CHAPTER VAX PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix . . VAX-3 VMS User's Network Working Group . . . . . . . VAX-19 INPUT/OUTPUT . . . . . . . . . . . . . . . . . VAX-32 VMS Notes from Nashville . . . . . . . . . . . VAX-107 VAX System SIG Committee List . . . . . . . . VAX-217 Forms at the End INPUT/OUTPUT Submission Form . . . . . . . . . VAX-222 System Improvement Request Submission Form . . VAX-224 VAX Systems SIG Fall 1987 SIR Ballot . . . . . VAX-226 PAGESWAPPER - August 1987 - Volume 9 Number 1 To register for on-line submission to the Pageswapper dial: (617) 262-6830 (in the United States) using a 1200 baud modem and log in with the username PAGESWAPPER. Articles for publication in the Pageswapper can be sent (US mail only -- no "express" services please) to: Larry Kilgallen, PAGESWAPPER Editor Box 81, MIT Station Cambridge, MA 02139-0901 USA Preference is given to material submitted as machine-readable text (best is Runoff source). Line length should not exceed 64 characters and the number of text lines per page should not exceed 48 (these limits are particularly important for sample commands, etc. where simple text justification will not produce a meaningful result). Please do not submit program source, as that is better distributed on the VAX SIG tape. Please do not submit "slides" from DECUS Symposia presentations (or other meetings) as they are generally a very incomplete treatment for those readers of the Pageswapper who are not so fortunate as to be able to travel to Symposia. Please DO write articles based on such slides to get the content across to a wider audience than is able to attend. Change of address, reports of non-receipt, and other circulation correspondence should be sent to: DECUS U.S. Chapter Attention: Publications Department 249 Northboro Road (BPO2) Marlborough, MA 01752 USA Only if discrepancies of the mailing system are reported can they be analyzed and corrected. VAX-2 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix System Security and Student Operators _D_D_O_O MMiixx John D. Hoinville and Richard D. Piccard Educational Computing Kalamazoo College Kalamazoo, Michigan _A_b_s_t_r_a_c_t We present techniques through which _a_l_l VMS user and system disk backups can be performed by operators whose _o_n_l_y extra user authorization entries are for the "OPER" and "GRPNAM" privileges and an identifier such as "STUDENT_OPERATOR". We place Access Control Lists on the tape and removable-media disk drives that deny access to _a_l_l users (including the operators) except those who have "BYPASS" privilege or a specific identifier. These methods have two primary advantages: all user disk backups are performed by command procedures, which reduces to minimal levels the chance of a catastrophic mistake, and the operators can legitimately be advertised to have _n_o special access to the files they are backing up. _I_. _I_n_t_r_o_d_u_c_t_i_o_n It is difficult to resolve the conflicting goals of system security; file system invulnerability to hardware, software, or human failures; system availability to users (measured in both connect time and CPU cycles), especially during normal working hours; and the effective use of professional staff time. At Kalamazoo College we have developed a method of meeting the file backup goal that has minimal impact on our ability to meet the other goals. Although there are some refinements that require recent VMS updates, variations of these techniques should be usable on any VMS Version 4 system. The approach described here should be of particular interest for small educational installations, but many other sites cannot afford to use skilled programmers as operators either. VAX-3 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix This paper is organized into four sections, as follows: I. Introduction II. Preparation A. Maintaining Operator Accounts B. Boot-time Measures C. File Access Control Lists III. Daily Operations A. Professional Staff B. Student Operators IV. Conclusion The heart of our approach is to perform all backups and system shutdowns from batch jobs that are submitted, /AFTER= a date in the far future, by user SYSTEM (or by someone else with "BYPASS" privilege). The student operator is granted access to the batch jobs' log files by an ACL entry that is automatically propagated to succeeding versions. The student operator can manipulate the queue to release or suspend a job, having OPER privilege. However, the student operator _c_a_n_n_o_t read other people's private files directly, having _n_e_i_t_h_e_r BYPASS nor READALL privileges, and also _c_a_n_n_o_t read them indirectly from the removable media (disk or tape) backup copies, because of Access Control List entries on the drives. The hard parts are first, getting the identifiers, ACLs, and privileges done correctly (examples are included below) and second, creating a batch job to shut the system down, since the queue manager and all batch jobs vanish like the Cheshire Cat fairly early in the shutdown process! The solution to the second problem is to have the batch job (which is _r_e_l_e_a_s_e_d by the student operator) include a RUN/DETACHED command to create a process that will shut the system down. This process is _o_u_t_s_i_d_e the purview of the job controller, hence survives the early shutdown of all queues. VAX-4 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix _I_I_. _P_r_e_p_a_r_a_t_i_o_n A. Maintaining Operator Accounts Our student operators use two accounts. The first is their personal student account, which has no special privileges or identifiers at all. The second is one of several long term accounts (usernames of the form OPERATORn, for n = 1, 2, ...) whose password is changed every academic term and whose OWNER field in SYSUAF.DAT is also changed to indicate the student presently authorized to use it. It is the OPERATORn accounts that have the STUDENT_OPERATOR identifier and the OPER and GRPNAM privileges. There are three steps involved in maintaining the student operator accounts: original creation, modification to operator functionality, and modification upon staff turnover. The original creation is by standard techniques, such as the ADDUSER.COM file presented in example 5-4 of the "Guide to VAX/VMS System Management and Daily Operations." Take pains to use a UIC code that gives all of the OPERATORn accounts the same group number, with no other users having that group number. The AUTHORIZE utility is used once interactively to create the STUDENT_OPERATOR identifier and later through a command procedure to grant it to the individual operator accounts. Using the OPERATORn accounts also helps for dealing with turnover in student workers. When a student leaves, only the OPERATORn account he or she used need be modified. This keeps management of the OPERATORn accounts (and hence, who has access to the OPER and GRPNAM privileges) to a minimal level. B. Boot-time Measures ACLs for Devices Since the tape and removable media disk drives are being used by the operator, the protection of those devices is especially important. There are often extended periods of time when previously used tapes or disk packs are physically mounted. This could constitute a terrible security hole, since the operators or other users might allocate the drive, mount the volume, and read the contents. The backup command procedure does indeed allocate the device, so that no interference will occur while it is in use, but that happens only when it is VAX-5 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix actually time to write the new backup copies of the files. We escape from this nightmare, as shown in Fig. 1, by the simple expedient of placing Access Control Lists (ACLs) on the drives. ================================================================ | Figure 1 | ============ $! $! Set protection so that users with the identifier $! EDCOMPSTAFF can access tape and disk drives. $! $ EDCOMP :== SET DEVICE/ACL=(- (IDENTIFIER=EDCOMPSTAFF,- ACCESS=READ+WRITE+EXECUTE+DELETE+CONTROL), - (IDENTIFIER=[*,*], ACCESS=NONE)) $! $ EDCOMP MSA0: $ IF ("''f$trnlnm("sys$sysdevice")'" .NES. "DJA0:") - THEN EDCOMP DJA0: $! ================================================================ Figure 1: Creating Device ACL's in SYSTARTUP.COM. The ACLs achieve several useful functions: the professional staff (who have need, or who know the SYSTEM passwords, and hence could get to it anyway) are granted full access to the drives; any job, typically with [SYSTEM] UIC, that has BYPASS privilege (such as the BACKUP jobs) gets access anyway; all other processes are _d_e_n_i_e_d access, including ordinary users _a_n_d the student operators. The ACL is applied to the RA-60 only if it is _n_o_t the system disk, since users MUST have access to many files on the system disk! Group Logical Names The student operators use their group logical name table to communicate to the backup batch job in the event that a fatal error or other catastrophe requires that the BACKUP command itself be re-issued. As discussed in part III, our backup command procedure includes code that must be executed only once, so that re-trying the BACKUP command must be possible. Besides, what system manager likes to be awakened by a phone call at at 6:20 AM from the student operator asking to have the backup job VAX-6 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix re-submitted! In order for the group logical name table to exist, at least one name must continue to be defined in it. Figures 2 and 3 show how this is done; the input command procedure specified in Fig. 2 is presented in Fig. 3. The technique of running LOGINOUT to process a DCL input file was documented in the Release Notes for VMS V4.1, page 4-2 (thanks go to the Colorado Customer Support Center staff for digging up this reference). ================================================================ | Figure 2 | ============ $! $! Run the job that creates the operator group [50,*] $! logical name table, for use by BATCH_SUBSETBACK. $! $ RUN/DETACH/UIC=[OPERATOR1]- /INPUT=SYS$MANAGER:OPERATOR_LNMTBL.COM- /OUTPUT=SYS$MANAGER:OPERATOR_LNMTBL.LOG- /PROCESS_NAME=OPERATOR_LNMTBL- SYS$SYSTEM:LOGINOUT $! ================================================================ Figure 2: Creating Group Logical Names in SYSTARTUP.COM. (Input file is presented in Fig. 3.) VAX-7 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix ================================================================ | Figure 3 | ============ $ DEFINE/GROUP STUDENT_OPERATOR_GROUP - "DO_NOT_DELETE_THIS_LOGICAL_NAME" $! $ EXIT $! $! +++++++++++++++++++++++++++++++++++++++ $! $! SYS$MANAGER:OPERATOR_LNMTBL.COM $! $! Executed by a RUN/DETACH/UIC=[OPERATOR] command in $! SYSTARTUP so that the [50,*] group logical name table $! will exist for reference by BATCH_SUBSETBACK.COM. $! $! +++++++++++++++++++++++++++++++++++++++ ================================================================ Figure 3: Input file for creating the OPERATORn group logical name table (see Fig. 2). PRIVBATCH The backup and system shutdown batch jobs are submitted to a privileged (or private) batch queue, PRIVBATCH, so that they cannot be shut out by user jobs. That queue is created as shown in Fig. 4. The memory management values proved essential for efficient analysis of accounting data, they are extravagant for backups. (Before we learned how large to set WSEXTENT, one quarterly accounting summary provoked 6 million page faults!) VAX-8 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix ================================================================ | Figure 4 | ============ $! $ INITIALIZE/QUEUE/BATCH/START/JOB_LIMIT=5- /WSDEFAULT=1024/WSQUOTA=3500/WSEXTENT=3500- /BASE_PRIORITY=4/PROTECTION=(S:RWED,O,G,W) - PRIVBATCH $! ================================================================ Figure 4: Creating PRIVBATCH in SYSTARTUP.COM. C. File Access Control Lists In order to allow a student operator to monitor the command procedures for errors, including BACKUP's soft tape write errors (which we track as an indicator of wear on re-cycled tapes), the student is granted the STUDENT_OPERATOR identifier. By giving the log files the correct ACLs, the student operators can access these files while leaving the rest of SYSTEM's log files secure from browsing. This reduces temptation the student operators might experience when it comes to looking into areas where they should not be allowed. Figure 5 shows the protection of a typical backup log file. Two methods can be used to create ACLs for the log files. One method is to set the ACL for the file with the VMS "SET ACL" command, the other method is to EDIT/ACL the file. An example of the former method is as follows: $ SET ACL/ACL=(IDENTIFIER=STUDENT_OPERATOR, - ACCESS=READ) EX.LOG Note that if this file had an existing ACL associated with it, this new ACL entry would be put at the top of the ACL list. The new entry could instead be added to the end of the ACL list with the "/AFTER" option. VAX-9 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix ================================================================ | Figure 5 | ============ $ DIRECTORY/SECURITY SYS$MANAGER:BATCHBACK.LOG Directory SYS$SYSROOT:[SYSMGR] BATCHBACK.LOG;24 [SYSTEM] (RWED,RWED,RE,) (IDENTIFIER=STUDENT_OPERATOR,ACCESS=READ) ================================================================ Figure 5: Log file ACL. Another way to create or modify an ACL is to invoke the ACL editor. By using the ACL editor, a file with the correct ACL can be created and then other files can be given the same ACL using the command $ SET ACL/LIKE=OBJECT_NAME=filename_with_ACL - filename_needing_ACL This command was not correctly described in early VMS Version 4 documentation. _I_I_I_. _D_a_i_l_y _O_p_e_r_a_t_i_o_n_s A. Professional Staff BACKUP Jobs The student operators can control the start of the backup because the submission command specifies execution at a later date that is hundreds of years in the future, so that the job will not run until an operator intervenes. Because of the complexity of the submission command, the system manager uses an interactive command procedure to perform the submission. We have such procedures for submitting the jobs that will perform overnight backups of new files, that will shut the system down for stand-alone backup of the system disk, that will perform VAX-10 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix full backups of specific user file directory trees, etc.. Figure 6 shows the critical portion of the command procedure that submits the procedure BATCH_RA60BACKUP.COM, which we use to write tape copies of RA-60 packs, typically the oldest existing disk pack copy of the system disk. The procedure prompts for the saveset name if the name was not supplied. Once a saveset name is supplied, the backup procedure is submitted to the batch queue named PRIVBATCH, for execution after the 18th of September in the year 3500. Of course that wait will never be fulfilled; the job is held until an operator intervenes, as discussed in part B of this section. By having the queue manager hold the job, there is virtually no CPU or memory used until it is actually time to execute the backup. In these respects it is superior to our original inspiration. That involved submission, by the backup or shutdown job, of another job that consisted of an infinite loop of 23 hour WAIT commands. Then, in the backup or shutdown job, that submission was followed by a SYNCHRONIZE command. When the time came to proceed, the student operator would delete the ================================================================ | Figure 6 | ============ $ BEGIN: $! $ IF P1 .NES. "" THEN GOTO START $ INQUIRE/NOPUNCTUATION P1 - "SUPPLY SAVE SET NAME (e.g., SYSTEMC.BCK): " $ GOTO BEGIN $! $ START: $! $ SUBMIT/QUEUE=PRIVBATCH/NOTIFY/PARAMETERS=("''P1'") - /NOPRINTER SYS$MANAGER:RA60BACKUP.COM - /NAME=BATCH_RA60BACKUP/AFTER=18-SEP-3500 $! $ EXIT ================================================================ Figure 6: A Simplified Submission Procedure. VAX-11 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix WAITing job, which would satisfy the SYNCHRONIZE and release the backup or shutdown job. This was clumsy, but it did work. It required a separate queue ("SYNCQUEUE"), consumed memory for both jobs, as well as a little CPU for the WAIT loop, and consumed active job slots in the PRIVBATCH queue. On our system we have located all user files in a directory tree whose root is DUA1:[USERFILES]. The files are split into three groups, in subdirectories MONDAY, WEDNESDAY, and FRIDAY, and the pseudodevices MON:, WED:, FRI:, and USER: are defined at boot time by the logical name assignments in Fig. 7. (USER: is a search list for the other three, for use when you don't know which group a given user's files are in.) We backup all files in each subtree once a week, on the obvious day. ================================================================ | Figure 7 | ============ $ ASSIGN/EXEC/SYSTEM/TRANSLATION=(CONCEALED,TERMINAL) - DUA1:[USERFILES.FRIDAY.] FRI: $ ASSIGN/EXEC/SYSTEM/TRANSLATION=(CONCEALED,TERMINAL) - DUA1:[USERFILES.WEDNESDAY.] WED: $ ASSIGN/EXEC/SYSTEM/TRANSLATION=(CONCEALED,TERMINAL) - DUA1:[USERFILES.MONDAY.] MON: $! $ ASSIGN/EXEC/SYSTEM FRI:,WED:,MON: USER: $! ================================================================ Figure 7: Creating pseudodevices in SYSTARTUP.COM. The procedure we actually use also performs several tasks at the beginning and end on specific days. Some of these we do _n_o_t want repeated, so provision must be made to loop back over the BACKUP command itself in case something should be wrong (fatal tape drive error, etc.). That is done through a group logical name for the student operators' UIC group (50 in our case). An alternative method to achieve the batch mode equivalent of the interactive INQUIRE command would be to write a program (in MACRO, FORTRAN, or PASCAL) that uses the SNDOPR (or SYS$SNDOPR) system service, which is documented on pages SYS-429 through SYS-442 of the _V_A_X_/_V_M_S _S_y_s_t_e_m _S_e_r_v_i_c_e_s _R_e_f_e_r_e_n_c_e _M_a_n_u_a_l_. Once the program has captured the text of the operator's reply, that VAX-12 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix text can be passed on to DCL through the LIB$SET_SYMBOL routine. We chose to implement the group logical name approach, rather than the SNDOPR approach, because the code is simpler and we could implement it with confidence more rapidly. Even if information more complex than a YES or NO type choice must be communicated, the group logical name approach should be considered. It will not work, of course, unless the batch process and the operator have either the same UIC group number, an identifier that matches an ACL on the group logical name table, or BYPASS privilege. (See the PAGESWAPPER for April, 1986, pages VAX-2 and VAX-3 for techniques of placing ACL's on logical name tables.) SHUTDOWN for S/A BACKUP Figures 8 and 9 provide the means for the student operator to shut the system down. Figure 8 is the batch job, submitted /AFTER=18-SEP-3500 by SYSTEM, and Fig. 9 is the input command file for the process that it creates after the operator releases it. There is no EXIT at the end of Fig. 9, because by then the system is shut down! This is essentially a variation on the technique used to create the group logical name table: run LOGINOUT so that DCL is available, and specify a command procedure as the input. Here, the input command procedure invokes SYS$SYSTEM:SHUTDOWN, with parameters specified so that no interactive session takes place. Presumably we are vulnerable here to changes in the parameter definitions of SHUTDOWN that DIGITAL might choose to make in future versions of VMS (are you listening, developers?). VAX-13 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix ================================================================ | Figure 8 | ============ $ SET NOON $ ON ERROR THEN EXIT $ SET LOGINS/INTERACTIVE=0 $ REPLY/USER/BELL/BELL - "THE SYSTEM IS SHUTTING DOWN FOR BACKUPS. PLEASE LOGOUT!" $ WAIT 0:2:0 $! $! Start detached process to shutdown the system $! $ RUN/DETACH/UIC=[SYSTEM]- /INPUT=SYS$MANAGER:CALL_SHUTDOWN.COM- /OUTPUT=SYS$MANAGER:SHUTDOWN.LOG- /PROCESS_NAME=CALL_SHUTDOWN - SYS$SYSTEM:LOGINOUT $! $ EXIT $! $! $! +++++++++++++++++++++++++++++++++++++++ $! $! SYS$MANAGER:BATCH_SHUTDOWN.COM $! $! Allows an operator to bring the system down. $! $! $! +++++++++++++++++++++++++++++++++++++++ ================================================================ Figure 8: SHUTDOWN Procedure Batch job. VAX-14 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix ================================================================ | Figure 9 | ============ $ ASSIGN OPA0: SYS$OUTPUT $ DEL*ETE :== DELETE/NOCONFIRM $ SET NOVERIFY $ @SYS$SYSTEM:SHUTDOWN 0 "Stand-alone backup" "N" "Y" - "BY 8 AM" "N" "REBOOT_CHECK" $! $! +++++++++++++++++++++++++++++++++++++++ $! $! SYS$MANAGER:CALL_SHUTDOWN.COM $! $! SHUTDOWN.COM parameters have the following meanings: $! $! P1 = Minutes until shutdown $! P2 = Reason for shutdown $! P3 = Spin down disk volumes $! P4 = Invoke site-specific shutdown procedure $! P5 = When will the system be rebooted? $! P6 = Perform automatic reboot? $! P7 = Shutdown option $! $! +++++++++++++++++++++++++++++++++++++++ ================================================================ Figure 9: SHUTDOWN Procedure Detached job input. B. Student Operators BACKUP Jobs The OPERATORn accounts have a LOGIN.COM that includes REPLY/ENABLE and then REPLY/STATUS to find out whether any outstanding requests need to be satisfied (such as mounting a second tape volume for the overnight backup). If such is the case, then the operator mounts the next tape and issues a command of the form VAX-15 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix $ REPLY/TO=request-number using the request number indicated by the REPLY/STATUS output. The operator checks the overnight incremental backup log file (with an EDIT/TPU/READ_ONLY command) to verify that all went well. Once the incremental backup is completed, the operator finds and mounts the first tape for that day's backup. Then the backup job needs to be released. There are two steps: first, find the entry number; second, SET that entry of the QUEUE to /NOAFTER as follows: $ SHOW QUEUE/BATCH/ALL PRIVBATCH . . . $ SET QUEUE/ENTRY=entry-number/NOAFTER PRIVBATCH After the above command is entered, the command procedure is immediately queued up for execution. The operator replies to mount requests and problems that may arise during the backup. The first request is usually that the operator mount a tape. The syntax of the reply is as discussed above. When each reel is finished, the operator receives another request to mount the next volume, loads the next tape, and replies to the request. If the procedure makes a request to "REPLY TO CONTINUE, USE DELETE/ENTRY= etc. TO ABORT JOB", then the procedure has encountered an error. _B_e_f_o_r_e _r_e_s_p_o_n_d_i_n_g, the operator logs in on a screen terminal and edits the log file with the command: $ EDIT/TPU/READ SYS$MANAGER:BATCH_SUBSETBACK.LOG If the only errors are "file access" errors that occurred while backing-up the files or "file not found" errors that occurred during the verify or the recording pass, or verification errors for blocks in MAIL or directory files, then all is normal for backups performed on an active system and the operator continues the procedure by replying to the request. If there is a FATAL type error message during the backup, we want to have it re-do just the tape writing part of the procedure. That will happen if, _b_e_f_o_r_e replying, the following line is executed by OPERATORn: VAX-16 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix $ ASSIGN/GROUP "SHUCKS" BACKUP_ABORTED Then the operator replies to the request so that the procedure continues. Immediately following the BACKUP (but not executed until after the REPLY, if BACKUP exits with warning or worse status) is a check of the logical name BACKUP_ABORTED in the group table for the [OPERATORn] UIC. If such a catastrophe has happened, and the above recovery started, then the request "Reply when FIRST tape is ready again." will appear. The operator waits for the tape on the drive to rewind and then mounts the _f_i_r_s_t volume of the tape set, again, and replies to the request. _A_s _s_o_o_n _a_s the backup is underway successfully, the following command must be issued by the operator, _b_e_f_o_r_e the backup completes: $ DEASSIGN/GROUP BACKUP_ABORTED In summary, the logic in BATCH_SUBSETBACK.COM is to test, immediately following completion of the tape-writing, whether the logical name BACKUP_ABORTED has any translation; if it does, then the procedure loops back to the BACKUP command itself. If one reel has a problem, the operator may get a message like "Request 8, Continue, Retry, or Abort". In that case the usual correct response is $ REPLY/TO=8 "RETRY" When the subset backup has completed, the operator _t_y_p_e_s the log file on the operator's console, with the command: $ TYPE SYS$MANAGER:BATCH_SUBSETBACK.LOG.0 This maintains a hardcopy record of these operations. SHUTDOWN and S/A BACKUP We instruct our student operators to use the REPLY/USER command to inform current users that a shutdown is impending. If time is not too pressing, the operator waits until all users have logged off before releasing the batch job to shut the system down (see Fig. 8). That job itself sets interactive logins to zero, announces the impending shutdown, and waits another two minutes. This extra time is invested as a matter of policy, to be sure that even if the student operator neglects to announce VAX-17 PAGESWAPPER - August 1987 - Volume 9 Number 1 System Security and Student Operators DO Mix the shutdown, our users will have had fair warning. Stand-alone BACKUP of the system disk is perhaps the most delicate of the procedures that we entrust to student operators. If they type the BACKUP command with the device names in the reversed order, then the system disk will be wiped out! We do reduce our vulnerability by _n_o_t pre-initializing the disk pack that they are supposed to be writing onto (but in this disaster scenario would copy from). Thus an older copy of the system disk would be "restored" and only software updates, batch jobs pending, and user authorization file changes, made during the intervening period (three weeks as we do these things), would be lost. All but the last week's worth would be recoverable from other packs. After the system has rebooted, the operator logs in and uses REPLY/ALL to notify waiting users that the system is up for the day. _I_V_. _C_o_n_c_l_u_s_i_o_n Performing backups from batch jobs liberates the operator, whether professional or student, from having to loiter in the machine room. Instead, we can perform useful work while the backups are in progress, knowing that OPCOM will alert us to the need to change volumes. In addition, the techniques described here permit less-skilled operators to relieve the more skilled staff both of the need to arrive in the wee hours to initiate backup jobs, and also of the time consuming tasks of supervising command procedures and changing tape volumes. We would like to thank several people who have contributed to the development of the techniques reported here. Amy Courter and Mary Haug developed the command procedures used for daily operation of our system. Forrest Piehl created the initial versions of the techniques reported here (using the SYNCHRONIZE command, as discussed above in section III, part A), including the code shown in Figs. 8 and 9 that permits student operators to shut the system down for stand-alone backup of the system disk. Finally, we thank the student operators who had to learn and deal with the various modifications and changes as we debugged each new inspiration: Kelly Baldrica, Daryl Dickhudt, Kenneth Dietz, In Bum Eom, Ernest Johnson, and Greg Lewis. VAX-18 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group VMS User's Network Working Group Jamie Hanrahan Working Group Chair This is a summary of the results of the meetings of the "Usenet for VMS Working Group" at the Spring '87 DECUS Symposium. _B_a_c_k_g_r_o_u_n_d (Those already familiar with the working group and its goals may skip this section.) Unix(TM) sites have for many years enjoyed the benefits of "Usenet",* a worldwide network of (mostly) Unix systems connected (mostly) via 1200 bps autodialed phone lines. The services provided by this network are electronic mail and "Netnews", an elaborate on-line conferencing system. The Usenet has been of incalculable value to the Unix community. The network encompasses over 10,000 systems, crossing both corporate and geopolitical boundaries; the body of knowledge that is available from the participants is not inconsiderable. Questions posted to Netnews on almost any topic are generally answered within a few days. Bug reports and fixes, patches to Unix system code, and public-domain software are all distributed through Usenet; many sites justify the telephone expenses (which, for sites with many connections, are not small) on this basis alone. Further, Usenet is effectively part of the Internet, which includes Arpanet, CSNET, and BITNET...#not to mention Easynet, DEC's internal DECnet network. ---------- * I am aware that, according to some authorities, the term "Usenet" only applies to Netnews, and the mail service should be referred to as "uucp mail" (a reference to the principal transport mechanism). But it is awkward to keep saying "Netnews and uucp mail". You know what I mean. VAX-19 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group VMS users who are aware of Usenet have for many years wished for a way to connect to it -- not only so we could talk to all those other folks, but also (or especially) so we could talk to each other. There has never been anything technically impossible about this, but the necessary software and information has not been cheaply available in the correct combinations. A few years ago some of us got together at a DECUS Symposium, started calling ourselves a Working Group of the VAX SIG, and set out to do something about the problem. To talk to the existing Unix systems in the network requires, first of all, a transport mechanism. The systems on Usenet mostly use uucp (Unix-to-Unix Copy), a facility that has been part of Unix since the early days. We considered writing uucp for VMS and rejected the notion, partly because AT&T regards the algorithms expressed in the Unix sources as trade secrets, partly because uucp isn't particularly efficient, and partly because we thought we hoped to find something else off the (low-cost) shelf. Our next candidate for the transport mechanism was Kermit. Kermit, like uucp, is a file transfer protocol, not a general protocol for task-to-task communications; but that is fine for our purposes, since all news and mail traffic on Usenet is shipped from one system to another as files. Good implementations of Kermit already exist for both VMS and Unix. The Unix version is written in C and can autodial, negotiate login sequences, and generally do everything that uucp can do. There is a VMS version of C-Kermit which needs a little bit of polishing but can also perform all of the required functions. (The standard VMS Kermit -- the one written in Bliss -- cannot. Polishing C-Kermit/VMS will be a far easier job than adding the autodialing, login script, etc., features to Bliss Kermit.) As for efficiency, standard Kermit does not do very well, but if "sliding window" support is added* it can approach 90% line utilization. Accordingly, I contacted Frank da Cruz about the possibility of my working on C-Kermit/VMS, and acquired the job. I got as far as getting automated, batch-mode file transfers to work between VMS systems, and my Unix-speaking acquaintances assured me that the scheme I was working on could be easily supported from the Unix side. This is where things stood before the Spring '87 ---------- *See _K_e_r_m_i_t _-_- _a _F_i_l_e _T_r_a_n_s_f_e_r _P_r_o_t_o_c_o_l by Frank da Cruz (Digital Press, 1987). VAX-20 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group Symposium in Nashville. _R_e_c_e_n_t _D_e_v_e_l_o_p_m_e_n_t_s Todd Aven appeared at the working group meeting and said (approximately), "Why don't you use PMDF?" Further discussion during the meeting, after the meeting, and at another meeting held later in the week established that: o PMDF (Pascal Memo Distribution Facility) is available at low cost. (Details are in a later section.) o PMDF can do everything we need in the way of a transport mechanism between VMS systems. o PMDF can speak to MMDF (Multi-channel Memo Distribution Facility), which is distributed with Berkeley 4.3 Unix systems, thus allowing connections with the Unix world (and thence with Internet). o PMDF includes an interface to VMS Mail, thus, part of the application layer is done. (The other part is a news system, about which more later.) o In addition to autodialed phone links, PMDF can use other channels, including DECnet, JNET*, and TCP/IP (Wollongong or Tektronix implementations). There are provisions for user-written support of additional channels. (A few people at the meeting expressed some dismay at the thought of throwing away what's been done so far on C-Kermit/VMS, but I didn't; very little of what I've done is specific to the use of Kermit as a network transport mechanism. I'm still working on the C-Kermit/VMS code and expect to ship some files to Columbia University within a few weeks [famous last words].) There are other nice things about PMDF. Since it can ---------- * JNET is a VMS-to-IBM_RSCS product, and is a product of Joiner Associates, Inc. VAX-21 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group conveniently use DECnet channels, and can delay use of such channels until a prescribed time of day or night, we may well be able to build a high-speed "backbone" for our network by using long-distance DECnet leased lines during off hours. (Many companies have such lines between facilities in widely-scattered locations, and nearly all such lines are idle at least twelve hours per day.) _W_h_a_t_'_s _B_e_i_n_g _D_o_n_e _N_o_w So, now that the software is available, we can start making connections, right? Well, almost. PMDF or UUCP? A public-domain uucp, "uuslave", is being developed. It does not at present do everything required (no autodial, no login script negotiation, etc.), but it does speak the uucp protocol and it has been certified by AT&T as _n_o_t having been derived from Unix sources. If this can be turned into a relatively complete uucp emulator for VMS, it would be preferable to PMDF. Not all Unix systems can talk to PMDF; if we could use uucp we'd have a vastly larger number of potential gateways to the existing Unix network. It would also simplify the node addressing issues (see the next few sections). Ad-Hoc Networks in the Internet Age If we let our network "just grow", we will fall prey to many of the problems currently faced by Usenet sites. We _c_o_u_l_d just start making connections, but the network will be far more beneficial to all of us if we do a bit of homework first. To make it as easy as possible to talk among ourselves and to the rest of the world, we should go to the trouble of becoming an officially recognized part of the Internet.* VAX-22 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group I am currently making inquiries as to how we might best connect with the existing Internet; until we have at least some information here, it would be wise not to proceed hastily. (But see "What You Should Do Now", below.) For some sites this will not be a problem: They will simply connect to a system that's already on the Internet and which is willing to perform domain server functions for them, and that will be that. But the rest of us will need, collectively, to become a zone, a domain, or a subdomain of our own, or else find a way to join an existing one. (Which is most suitable depends on what sort of resources we can scrounge up. It is not unreasonable to suppose that DECUS, or some part of DECUS, might provide the "responsible organization" to act as the administering authority of a new Internet domain [perhaps decus.org], but this remains to be seen.) Joining the UUCP Zone One option is to register new domains through Stargate Technologies, a not-for-profit, volunteer organization that is providing name registration services for the .uucp zone of Internet. To do this you need to find a nearby Unix system which has the MMDF software and whose administrators are willing to give you a connection. As long as your site is reachable by what appears to be uucp (i.e. via "bang style" addressing), it doesn't matter that the actual transport mechanism is something other than uucp. Registration for your organization through Stargate costs $150 per year; sign up and the rest of the Internet will recognize your organization as a domain and will properly handle mail to and from you. If you are the ABC company, your domain might be called abc.com . You can then add other systems to the routing tables on your "gateway" machine; the rest of the net doesn't need to know about those (since all traffic to and from the rest of the net will funnel through the gateway). Thus you can have ---------- * Please do not call or write to tell me that "only DoD-approved sites can connect to Internet". This was true when Arpanet and Internet were synonymous. Today, Arpanet is the ".arpa" zone of the larger Internet. Other zones include .bitnet, .csnet, and .uucp . VAX-23 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group eng.abc.com, sales.abc.com, etc., each subdomain representing a department. Stick a user name and an "@" on the beginning (i.e. yourname@eng.abc.com) and you have a complete Internet address for yourself, such as smith@eng.abc.com . News System Issues PMDF provides the transport mechanism and the mail application. Everyone involved in this project agrees that Netnews, or something like it, is also essential to the growth and health of this sort of network. A news system provides an effective replacement for network-wide "mailing lists", which can be expensive and difficult to maintain. Besides, with mail you can only exchange messages with already-known people; news lets you find people you might like to send mail to. The existing netnews software cannot, as far as I can determine, be compiled under the VAX-11 C compiler. (I've tried; if someone else has had better luck, please let me know.) It can be compiled and run under Eunice (Wollongong's Unix-on-top-of-VMS product), but it is not reasonable for us to require sites to buy Eunice just to participate in the network. Therefore, I am working on a VMS-specific news implementation. It is composed of four major components: Database server, import and export programs, and user interface. The components communicate with each other using a slightly modified version of Brian Kantor and Phil Lapsley's Network News Transfer Protocol (NNTP); the biggest change (and it's a minor one) is that the communication paths are VMS mailboxes or DECnet logical links, rather than character-oriented channels like TCP/IP (which is what NNTP expects). This design will permit users on different VAXes on a LAN to access a common news database on a single machine; it will not be necessary to maintain the news on every machine where it might be read. It will also be easy to build alternate user interfaces without having to deal with the complexities of the news database. The database itself stores the articles in individual text files, with RMS indexed files acting as indices to the articles. VAX-24 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group More information on the news software will appear in the next Working Group report, which I hope will be ready for next month's PAGESWAPPER. The job still looks relatively simple at this stage; I think I can comfortably promise that I'll have the news software ready before we get the addressing/domain questions resolved. Too Much of Anything, Even News, is Not Necessarily a Good Thing In the meantime, you might consider the following issue: Usenet is currently laboring under a proliferation of news text. With 10,000 sites and untold legions of users producing ten megabytes or so of news a week, the network backbone links are saturated or nearly so, and _n_o_b_o_d_y has time to read it all. Fortunately you don't have to read it all; it's easy to ignore the topics ("newsgroups") you're not interested in, just as you avoid subscribing to every magazine that's available. But the backbone sites must still ship all of it everywhere. (Perhaps nobody at your site reads the newsgroup on Celtic cooking, but someone at one of the sites you feed might do so.) As a result of the proliferation of newsgroups and contributions thereto, both the phone line time and the disk space are becoming prohibitively expensive for many Usenet participants. Many sites (other than those that make up the backbone) have ceased forwarding what they consider to be "nonessential" newsgroups. We will face this issue sooner or later. Should we decide at the outset that only technically-oriented newsgroups will be established? Or that there will be "mandatory" and "optional" newsgroups, and sites on our network will only be required to store and forward the "mandatory" ones? (Through the primary links, anyway. Sites would of course be free to establish additional links, over which they might receive and send news that they wouldn't get from the backbone. I believe that some of this is already happening on Usenet.) With any sort of restrictive policy, we face the questions of "who will make the decisions" and "how will they arrive at their decisions", and whatever policies are adopted, it seems certain that some people will be unhappy. (Should this working group, or some other agency of DECUS, set the policies? If so, will we VAX-25 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group be unable to use the network to exchange information (like product pricing) that conflicts with DECUS's canons of conduct?) But the Usenet experience indicates that without such a policy, the network will soon become untenable; presumably, this will make nearly everybody unhappy. _M_i_n_o_r _I_s_s_u_e_s_: _T_h_e _N_a_m_e _G_a_m_e Those of you who have been following this effort for some time will have noticed that the title of the article says "VMS User's Network Working Group" instead of "Usenet for VMS Working Group" -- the latter being our original name. This name change was made at Larry Kilgallen's suggestion that the idea of a "VMS User's Network" would have a broader appeal than the idea of "Usenet Access for VMS" -- since, after all, most VMS users would like a convenient way to talk to each other, while only a small fraction of the VMS community may be aware of Usenet and the desirability of connecting to it. I'm not entirely happy with the name "VMS User's Network" (which will probably be shortened to VUnet, which sounds too catchy not to have been used before, or even [God help us] Vusenet), but it seems to describe what we're trying to do. If you have a better idea for a name, I'd be happy to hear it. On the other hand, I'm not inclined to spend much energy worrying about this. _W_h_a_t _Y_o_u _S_h_o_u_l_d _D_o _N_o_w Acquire and Install PMDF At this writing it is likely that PMDF will play an important role in the network, so if you want to join quickly, you would be wise to acquire it. PMDF is available for the cost of distribution. VAX-26 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group Send a check for $50 (US) to: Ned Freed The PMDF Project Computing Services Harvey Mudd College Claremont, CA 91711 (714) 621-8006 It will come on nine-track tape (VMS BACKUP format) with about a hundred pages of documentation. (If your organization makes it difficult to buy software without prior approval, tell them that it's a book that happens to come with a tape instead of the other way around. Or just tell them it's a book.) Read the documentation, install PMDF on your system(s), and become as familiar with it as you can. (Even if you don't care about joining our network, you can use PMDF if you're using VMS Mail over DECnet. Have you ever been frustrated when trying to send mail to someone who lives on a node that's temporarily unreachable? PMDF will automatically spool the mail for transmission when the link is up. You'll have to give a "To:" address that specifies routing through PMDF, but that can be handled via logical names.) Learn About Internet Find someone in your organization, DECUS LUG, or other nearby tribe who knows something about Internet. This should not be difficult; most large universities and many small and large corporations are on the Internet, and there should be someone around who can either explain things to you or point you to someone else who can. They can in turn provide you with the various documents (called RFCs, for Request For Comment) which describe Internet addressing, standards for the exchange of Internet mail and network news, etc. (You can also obtain these by buying them from the Network Information Center at SRI International in Palo Alto, CA, but I do not have a complete list of the applicable documents yet, and you do not want to buy [let alone read] _a_l_l of the available RFCs. Next month's article will have a list of the RFCs to order, along with complete ordering information.) VAX-27 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group If you can't find a local Internet guru, don't despair: One of the things our Working Group is going to do is collect all of the necessary information and boil it down into easily-digested form. There's no need for more than the first few sites to dig it out ourselves! But if you want to be one of the first few sites, you'll need to know this material. (Then I can ask _y_o_u questions.) Acquire Hardware You'll need, at minimum, a 1200 or 2400 bps autodialing modem, attached to your VAX in such a way that software in the VAX can place outgoing calls, and incoming callers can be connected to your VAX. For traditional terminal multiplexers this means connecting the modem to a port that supports modem control (e.g. _n_o_t ports 2 through 7 of a DMF32). I don't know what it means for ports on Ethernet terminal servers, since we don't have any here and I don't know how they work for outgoing calls. (I do know that PMDF can negotiate the terminal server connection sequence for an incoming call; this would also work for modems attached to port selectors and the like.) Oh, yes -- order a phone line, preferably one that does not go through your company's PBX or Centrex or equivalent, and certainly not one that requires an operator to connect it to the VAX. Use it for a while with the modem and have the phone company correct any line noise problems before it's time to connect to the network. If you already have a suitable modem on your VAX, perhaps to support logins-from-home by employees or students at home, you can use that for PMDF links. Don't worry about interference with dialup terminal users: PMDF can be told to only place calls during hours when no one else will likely want the line, and if it finds the line in use (either when attempting to place an outgoing call or when calling into another system), it will just try again later. VAX-28 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group Talk To Your Management Obviously you're going to be spending some money for phone lines, line charges, disk space, etc. You will also be making known, to people at another site, a telephone number and username/password which can be used to access your VMS system. Granted the username will be a captive account, good only for sending and receiving mail via PMDF (the news traffic will be sent as "mail"); granted that there are thousands of Unix sites all around the country doing the same thing with uucp-captive accounts; you may still find it difficult to convince management that this is a good idea. (I don't know if dialback schemes will be usable with PMDF.) But convince them you must, if you want to join the network. Things will become easier after the network actually exists and has a few hundred sites on it; there will be a snowball effect once "everyone is doing it". But for the first hundred or so, things will be tough. Better start working on the problem now. (If all else fails, go to work for a more reasonable outfit.) Find Potential Network Partners Find other sites in your area to hook up to. In addition to VMS sites, you want to find a Unix site that is running Berkeley 4.3; they should have the MMDF software, and can therefore provide a gateway for you to the existing Unix-based network. (Whether they'll be willing to do so is another issue.) (Of course, if you have a Usenet-connected machine running 4.3 in your computer room, that problem is solved, but you should still look for other potential VMS sites in your area; if we're at all successful, they'll be wanting to connect to you.) Write To Me To build a reasonable network in as short a time as possible, we VAX-29 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group must hear from as many interested sites as possible. We need the following information: o Name, address, phone, etc. If you have any sort of electronic addresses (whether on networks or on dial-in machines like Compuserve), include those. o Do you think your site will be able to place and/or receive local and/or long distance telephone calls, in order to exchange mail and/or news with other sites? What do you think the cutoff point might be in terms of dollars per month? o Are any systems at your site connected to any part of Internet? (In the United States, Internet includes Arpanet, Milnet [but we can't use that], Bitnet, CSNET, and UUCP.) o Are you willing to be a pioneer, or would you prefer not to hook up to us until we have everything working reliably? o Does your company operate any long-distance VMS-to-VMS DECnet leased lines, whose excess (e.g. late-night) capacity might be used for mail and/or news? o ...anything else you think might be useful for us to know. NOTE: If you want to join the network, we want to hear from you, whether or not you can place outgoing calls, provide DECnet link capacity, etc. _C_o_n_c_l_u_s_i_o_n The light is (finally) visible at the end of the tunnel, and we are fairly certain that it's not a freight train. Right now it looks as though we might easily be up and running by the Fall Symposium in Anaheim. If you can help in any way, please let us know. Jamie Hanrahan VAX-30 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS User's Network Working Group Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 uucp: {akgua,hplabs!hp-sdd,sdcsvax,nosc}!crash!pnet01!jeh arpa: crash!pnet01!jeh@nosc.mil internet: jeh@pnet01.CTS.COM or: pnet01!jeh@crash.CTS.COM US194066 on the Pageswapper system (see page VAX-2) DECUServe: HANRAHAN VAX-31 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT INPUT/OUTPUT A SIG Information Interchange A form for INPUT/OUTPUT submissions is available at the back of the issue. To register for on-line submission to the Pageswapper dial: (617) 262-6830 (in the United States) using a 1200 baud modem and log in with the username PAGESWAPPER. ================================================================ Note 567.10 BACKUP Hints and Questions 10 of 13 "Kevin Angley" 19 lines 27-MAY-1987 14:45 -< To CRC or not to CRC >- ---------------------------------------------------------------- The /CRC or /NOCRC question has taken pseudo-religious airs. At Nashville, top people in DEC's storage group said that /CRC was a waste of time with new technology drives. However, the VMS Backup developer stressed the importance of reliability that a software CRC gives you. It comes down to the question of how far do you go in establishing a high confidence level. After all, you could do "paranoid" backups (small blocks, redundancy groups, /CRC, etc.) and THEN do fifty BACKUP/COMPARE's if you wanted to be EXTRA sure. But there's a point of diminishing returns, isn't there? Key to this discussion is the number of times "block recovered by redundancy group" is actually due to a software CRC error, or by a hardware reported error. The VMS developer said that 9 times out of 10 that you see that message, /CRC has saved you (seems unlikely to me). But the same message is reported for hardware detected errors, so you cannot build any kind of historical feeling for this. Perhaps if backup gave you a clue as to whether the hardware was the reason for tossing a block, or if the software CRC helped out, then we could judge this issue better. Kevin Angley VAX-32 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 ================================================================ Note 567.11 BACKUP Hints and Questions 11 of 13 "Linwood Ferguson" 24 lines 29-MAY-1987 21:08 -< /CRC gets you $25,000+ TK50 >- ---------------------------------------------------------------- /CRC and /NOCRC never seemed worth it, even for marginal gains on slow (i.e. 750 machines). Now it seems that it might be worth the risk: we tried our first MicroVax II with a TU81+. First experience, with our standard backups (/CRC by default) caused the language to singe the cabinet: it was as slow as a TK50; worse since you could see it sitting there and barely moving. Then I tried /NOCRC and no other changes (65534 blocks, /buff=5) and the tape streams at 75 ips with no stops (well, almost none depending on what it's copying). It is visibly the same speed as a 8500 we use (/CRC there). Actual timing put elapsed time "cost" of /CRC at 518% and CPU time cost at 1600%. CPU and elapsed were almost identical (does this make it a "reel time" backup?). So, for the moment, rather than tell the people paying the bill that they paid $25,000+ for a TK50, we're going with /NOCRC and crossing our fingers, and /VERIFY'ng more often. Question: can someone enlighten us as to whether there is such a thing as redundancy groups if you use /NOCRC, or what effect /GROUP has if you use /NOCRC? The manual seems to disconnect these somewhat. Is there any way to put redundancy into the backup without /CRC? Linwood Ferguson MJ Systems, Inc. P.O. Box 5223 2564 Ivy Road (22901) Charlottesville, Va. 22905-0223 804/977-2732 VAX-33 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 567.13 BACKUP Hints and Questions 13 of 13 "Frank J. Nagy" 12 lines 1-JUN-1987 08:48 -< /GROUP and /CRC are indeed different >- ---------------------------------------------------------------- The redundancy group (/GROUP) and cyclic redundancy (/CRC) are indeed separate. The redundancy group, according to the documentation, is a complete XOR block written every N blocks. That is, it seems to be the XOR of the previous N blocks. The redundancy group allows BACKUP to recover a block from the group if it disappears or is unreadable by using the other N-1 blocks of the group with the XOR block. Or, at least that seems to be the theory. Personally, I think that /NOCRC is probably OK on 6250 bpi tapes. We plan to proceed in this fashion once we get a TU81+ for our LAVC. Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 ================================================================ Note 577.18 Questions/Comments on TPU 18 of 18 "Offline Submission" 72 lines 29-JUN-1987 22:06 -< Customization in Germany >- ---------------------------------------------------------------- I am the leader of a software development group and we used a number of editors in the past. Getting our first VAX with EDT in 1983, it was our first full screen editor and all our people loved to work with it. But then came EVE. It should be the best you ever got, but nobody on our team could/would use it, because the man/machine interface was quite different from our well known EDT. But there was a small example in the TPU manual to make EVE like EDT. With this example and some additional changes I tailored the keypad interface of EVE to an EDT-like keypad. My people now loved EVE and - with some experience - had the advantages of EVE too. VAX-34 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT One of the best features of TPU is that you can extend the functionality of the programs stepwise without changing the previous characteristics. So later I extended EVE by adding several functions: - First I added typewriter-like tabulator functions, which insert a number of spaces instead of tabs (to get a readable output on a standard printer), working in overstrike and insert mode. - The next step was to implement functions for the replacement of rectangular text blocks, also working in overstrike/insert mode. - Then I included paragraph formating tools with semi-automatic hyphenation and right margin justification. - Then I added tools for pagination. - And last but not least I added tools for setting up tables of contents and indices. Finally we got a very powerful online text processing system. For our standard technical paper it fits our requirements on a WYSIWYG editor and nobody in our group will use the old-fashioned Runoff anymore. More and more of our marketing and management people will use this powerful and easy to use tool. But against all its advantages, TPU has sometimes mysterious and strange behaviour. In particular the pattern and search functions deliver sometimes unexpected results. We discovered a number of faults, and if you write a letter to the source you will get an answer like: "Thank you for bringing this to our attention. You are right... We plan to fix this problem in a future release of TPU." Some of the problems we encountered could have been prevented by better documentation, particularly in the case of side effects of the predeclared functions which are not explained (with some exceptions). VAX-35 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT If there is any interest in our text processing package we can hand it out. Unfortunately the program and the documentation is written in German (procedure names and messages) and probably someone would have to translate it. Rudolf Reusch AEG Aktiengesellschaft A46 V42 Hafenstr. 32 D 2000 Wedel/Germany Telephone: (0)40-700-565 Date: May 22, 1987 ================================================================ Note 585.8 Anyone use defrag programs? 8 of 10 "Kevin Angley" 15 lines 27-MAY-1987 14:56 -< Defrag alternatives >- ---------------------------------------------------------------- For sites that do not have much utilization during the wee hours, I suggest compressing as part of a nightly backup. We do an image backup of one of the user disks each night to a scratch disk. Then, if NO errors occur during that image backup (and be careful - backup's $SEVERITY is not to be trusted), we dismount the subject disk, and do a physical backup back to it. Then, we write the scratch disk off to tape. Now if RA's had programmable unit plugs, you wouldn't even have to do that. As far as third party defraggers go, I prefer to hold off for a year or two. I heard at Nashville that defrag tools are on DEC's possible do list for some possible unannounced release AFTER the next major unannounced not to be mentioned release. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 VAX-36 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 585.9 Anyone use defrag programs? 9 of 10 "Ken A L Coar" 12 lines 28-MAY-1987 11:47 -< Ah'm agin 'em >- ---------------------------------------------------------------- My personal opinion is that I don't want to TOUCH a third-party defragmentation program until the people who DESIGNED the file system have come up with one. This item has been high on the SIR list for a long time, as I recall, and if Digital hasn't done it, I surely don't want to trust my file systems to someone else's original code..! Particularly since the license would almost certainly read `no liability is assumed for loss or damage to ANYthing - data, files, business revenues, your hairline or hair colour - stemming from your use of this product.' If something goes dreadfully wrong, I want someone I can jump upon with hobnailed boots! :-)} #k Ken A L Coar General Dynamics Office Systems 12101 Woodcrest Executive Drive Creve Coeur, MO 63141 (314) 851.4003 (CST) ================================================================ Note 585.10 Anyone use defrag programs? 10 of 10 "Brian Tillman, Lear Siegler, Inc." 4 lines 24-JUN-1987 10:43 -< I'd love a defragger, but... >- ---------------------------------------------------------------- The problem I've encountered is that, try as I might, I can't justify the cost to management, because I can't get a handle on how much money I'm losing because of decreased throughput. Inefficiency due to fragmentation is extremely intangible. Brian Tillman Lear Siegler, Inc. 4141 Eastern Ave. MS121 Grand Rapids, MI 49518-8727 (616)241-8425 VAX-37 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 590.5 VAX 11/750 Word Processor 5 of 7 "Bob Ainsley" 12 lines 4-JUN-1987 14:32 -< MASS-11 Experience >- ---------------------------------------------------------------- We are using MASS-11 on 2 clusters with CPUs ranging from 750 to 8600. MASS-11 is a memory hog. It can be a CPU hog if something like a table of contents is requested. 15 to 20 users can bring a 780 to its knees. SPM and VPA seem to confirm that memory is more critical than CPU. We usually see 85-90% of 16 megs memory used on a dedicated 780 while CPU will be around 50%. Adding more memory (8 megs) seems to help. There is one bug that we can't seem to isolate. MASS-11 will sometimes go off into the weeds and get stuck in an infinite loop, consuming as much CPU that is available. The only recovery is stopping the offending process. Bob Ainsley DSC COMMUNICATIONS 1000 COIT RD PLANO TX 75075 2145193952 ================================================================ Note 590.6 VAX 11/750 Word Processor 6 of 7 "Barry L. Wallis" 13 lines 4-JUN-1987 19:04 -< More MASS-11 experience. >- ---------------------------------------------------------------- We also use MASS-11. We use it on a VAX-8600 with 20 megs of memory. I concur that memory seems to be the critical element (most processes seem to ring in between 1900 and 2000 pages [2000 pages is our WSQUOTA is 2000]). The only complaint I have is when one of my users does a large match/merge operation (to produce 5000 "personalized" form letters). At that point page faults so badly that I need to suspend until non-prime time so my other users can get useful work done! VAX-38 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT But all told, I would pick MASS-11 (over WPS and other word processors) again. Barry L. Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 ================================================================ Note 590.7 VAX 11/750 Word Processor 7 of 7 "John Osudar" 12 lines 4-JUN-1987 20:21 -< Another MASS-11 site >- ---------------------------------------------------------------- We use MASS-11 on a 785 with 32 megs, and a lot of other (interactive and batch) activity. I wonder if you've got the image installed shareable? The current version (6C), when installed, shows up as global sections totalling 715 pages. MASS-11 MAIL (if you use it) installs at another 616 pages. That 715 figure is for the "large" (monolithic) image. I agree that MASS-11 can be a memory hog, and especially a CPU hog, particularly when doing formatting for printing. The "infinite loop" problem that someone mentioned is a bug that we also saw, though I believe it was in a version before the current one. If you can feed MASS-11 the memory and cycles, though, it's a good product -- both our secretarial personnel and scientific staff use it. John Osudar Argonne National Laboratory 9700 S. Cass Ave. Bldg. 205 A-051 Argonne, IL 60439-4837 (312) 972-7505 VAX-39 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 599.7 Any PSI users out there? 7 of 9 "Ralph Sierra" 9 lines 7-JUN-1987 17:32 -< Should I be using PSI? >- ---------------------------------------------------------------- This may be a naive question but here goes (just charge me 5 negative points): What is PSI? Reply .6 said they were using PSI in a 'classic' application, for communicating via TYMNET. I have a user that calls various information services, in some cases, via TYMNET. This user is currently dialing out using VAXNET. How would this differ from using PSI? Ralph Sierra NKF ENGINEERING INC 12200 SUNRISE VALLEY DRIVE RESTON VA 22091 703/620-0900 ================================================================ Note 599.8 Any PSI users out there? 8 of 9 "Jamie Hanrahan" 118 lines 8-JUN-1987 15:33 -< PSI explained in a (large) nutshell >- ---------------------------------------------------------------- > What is PSI? PSI (technically, "VAX-11 PSI") is a layered product from DEC that allows VMS to access packet-switched data networks (PSDNs). If you're familiar with PSDNs, skip ahead to the section labelled "WHAT PSI DOES". (Difficult to do with NOTES, I know...) ABOUT PSDNs Without going into the details of packet switching, suffice it to say that, for the most part, a packet-switched data network is a privately-owned nationwide data communications network. The best- known examples in the US are Telenet and Tymnet. (In most other countries there is just one PSDN, and like the phone systems, they are run by each country's telephone and postal authority. Also, there are such things as private PSDNs, VAX-40 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT operated by and for a single company.) A PSDN allows you to connect multiple systems at various sites around the country or the world. Depending on the amount of traffic and the number of connections you need, there can be a large cost savings over leased lines. Suppose you have systems in Boston, Los Angeles, Seattle, and Houston, and you want each connected to all the others. You would need six long-distance leased lines. Or, you could sign up with a PSDN. A leased line runs from each system to one of the PSDN's nearby switches (four lines total). Each of your systems is assigned an "address" by the PSDN, quite analogous to a phone number. When one system wants to talk to one of the others, software in the first tells the PSDN to place a "call" over the PSDN to the target system's address. When the target system answers ("confirms the call"), the two systems can communicate just as if there was a leased line between them. Naturally you can set things up so that your system will only accept calls from certain other addresses (presumably the addresses of your other sites). Multiple calls can be multiplexed over a single connection to the PSDN, so all systems can be simultaneously connected to all others (within reason; there are some large limits on the number of simultaneous calls) with just one local leased line per system. Charges: You pay for the local leased lines, plus a monthly service fee, plus a charge for the amount of data (actually, the number of packets) you send. None of the charges are related to the distances between systems. The rate structures are such that if you have several sites to be connected, and you wouldn't use a leased line to more than, say, 10% capacity (remember this includes evening and weekends, during which time many leased lines are idle), you may well save money by using a PSDN. If you have just two systems to connect you may find that the cost of the single long-distance leased line is less than the cost of the two local leased lines plus the PSDN's monthly fees. (That was our experience.) Another use for a PSDN is to support people calling into your system with async terminals from around the country. You can set up a bunch of modems and let them make long-distance calls; or, you can hook up to a PSDN. Instead of calling your system directly they call a (usually) local number for the PSDN's nearby "X.29 PAD" (Packet Assembler/Disassembler). They then tell the PAD (via their terminal keyboard) the PSDN address of VAX-41 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT your machine. The PAD places a PSDN "call" to your machine, and software in your machine creates a virtual terminal, much like a DECnet remote terminal, in response to this call. You (not the caller) pay for line time on the PAD, plus (I believe) a charge for the amount of data. This is much cheaper than the equivalent long-distance phone line time would be, for almost any distance, and line errors are confined to the hop from the caller's modem to her local PAD (everything else is checksummed and corrected if errors are noted). The downside is that the round-trip time is considerably longer than for a regular dialup connection; use of things like screen editors, which require fast response to individual keystrokes, is severely hampered. WHAT PSI DOES VAX-11 PSI provides the software to place and accept calls and otherwise speak the proper protocol (X.25 over HDLC) to the PSDN. PSI provides four interfaces to users and programmers: 1. A $QIO interface for programmers, quite similar in function (though, of course, different in detail) to the $QIO interface for DECnet task-to-task communication. Processes on VAXes connected to the PSDN can communicate by using this feature of PSI. No other software (e.g. DECnet) is required. 2. X.29 (remote terminal) support. PSI can accept "calls" from PADs, allowing remote terminal users to connect to your system via the PSDN. A user logged into your system (usually locally, but not necessarily) can also place *outgoing* "remote terminal" calls to *other* systems on the PSDN (assuming that those systems have X.29 remote terminal support). 3. PSI MAIL. Users on machines connected to a PSDN can use VMS MAIL to send messages to each other, again without needing DECnet. 4. DECnet data link mapping (DLM). If you have DECnet and PSI on systems that are connected to a PSDN, DECnet can place PSDN "calls" between the systems. (The name "data link mapping" refers to the idea that the software "maps" a PSDN "call", or "virtual circuit" (a requested and confirmed call creates a virtual circuit), into a DECnet data link.) Then DECnet can use the data link for all the usual DECnet stuff: Task-to-task communications, file transfers, remote terminals, VMS MAIL, etc., etc. VAX-42 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Naturally, all of the above traffic can ride on the single leased line between your system and the PSDN's local switch. Does that help? (Experts: What have I missed or gotten wrong?) Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 599.9 Any PSI users out there? 9 of 9 "Jamie Hanrahan" 29 lines 8-JUN-1987 15:53 -< PSI vs. VAXNET? >- ---------------------------------------------------------------- Oops, almost forgot your other question: I have a user that calls various information services, in some cases, via TYMNET. This user is currently dialing out using VAXNET. How would this differ from using PSI? Right now your user is, presumably, using VAXNET* to dial out to TYMNET's nearby (I hope it's nearby) X.29 PAD. If you connect your VAX to TYMNET and install PSI (and there's no point in doing one without the other), your user could do this instead: $ SET HOST/X.29 other_system_name_or_psdn_address Benefits would be immunity from line noise (since every serial link is now a checksummed-and-corrected link) and, possibly, greater effective line speed, as the local hops to the PSDN switches are generally 9600 bps or more. But the cost would be exorbitant if this was all you were doing with the PSDN connection. ------ * (Footnotes in NOTES? Oh, well) Someone is sure to ask about VAXNET. VAXNET is a package that has appeared on several DECUS VAX SIG tapes. It is a very, very fancy SET HOST utility that includes autodialing capability for various modems, file transfer support, and many other features. It is quite a bit VAX-43 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT more efficient than the regular VMS SET HOST, according to reports. Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 637.1 New Version of MRGATE 1 of 2 "Ken A L Coar" 13 lines 28-MAY-1987 10:42 -< Not quite, for us at least >- ---------------------------------------------------------------- Another neat thing to do with MRGATE is to have your VMSmail automatically sent to ALL-IN-1 !! You can do this by setting the VMSmail forwarding address to A1_user::A1::MRGATE That doesn't work for us, because our ALL-IN-1 usernames are of the form `lname,xx' (mine, for example, is COAR,KA). In order to get this to work, we use MRGATE::"A1::a1username" e.g., MRGATE::"A1::COAR,KA" #k Ken A L Coar General Dynamics Office Systems 12101 Woodcrest Executive Drive Creve Coeur, MO 63141 (314) 851.4003 (CST) VAX-44 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 637.2 New Version of MRGATE 2 of 2 "Jack Patteeuw" 13 lines 1-JUN-1987 06:52 -< MAILUAF.COM Warning ! >- ---------------------------------------------------------------- Ken is very much correct ! In fact we have a A1 node on our network that assigning ALL-IN-1 user names in the format "Lname, Fname". Note the SPACE in between ! Just a warning for all you "System Manglers" out there. If you try and use MAILUAF.COM to set another users autoforward address be warned that the quotes used in .1 will screw up the database. This has to do with the way the .COM file re-builds the record to be written out to MAILUAF.DAT. Best protection is to double check the entry with a SHOW command. If it is screwed up, you will have to log into that account and fix via MAIL or fix the .COM file to properly handle the quotes (which I have not had time to do yet). Jack Patteeuw Ford Motor Co. Electrical and Electronics Division 31630 Wyoming Livonia, MI 48150 313-323-8643 ================================================================ Note 642.5 print&batch complaints 5 of 10 "Brian Tillman" 12 lines 24-JUN-1987 09:33 -< On the whole, they seem to work >- ---------------------------------------------------------------- RE: < Note 642.0 by NODE::US133673 "John Osudar" > As long as we're complaining about the queue manager/job controller/symbionts (e.g. note 630)... We find the Separation modules and the Setup modules generally work well. However, I do have to disagree with the form-level setup module. It comes out in front of every file in a job, not just when the form is changed, which seems to me to be all that's necessary. Notes 619 et. al. also address this issue for non-DEC lasers, which fits us as well. VAX-45 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Brian Tillman Lear Siegler, Inc. 4141 Eastern Ave. MS121 Grand Rapids, MI 49518-8727 (616)241-8425 ================================================================ Note 642.6 print&batch complaints 6 of 10 "Tony Carter" 4 lines 25-JUN-1987 07:39 -< Form-level setup modules useful >- ---------------------------------------------------------------- We use form-level setup modules with our LN03's. The setup module resets the laser printer to a known state. We had users changing all sorts of printer parameters and not resetting after they were done. This way it is transparent to the users. Tony Carter P.O. Box 846 Middleton, MA 01949 (617) 245-6600 ================================================================ Note 642.7 print&batch complaints 7 of 10 "John Osudar" 6 lines 25-JUN-1987 15:57 -< re: 642.6 >- ---------------------------------------------------------------- That's what "job reset modules" are for -- to reset your printer to a known state after a job finishes. The form-level stuff in intended to set up new characteristics specific to the form being used. While you can obviously put things anywhere you want, it's a lot easier to put reset stuff in the job reset modules, setup stuff in the setup modules, etc. John Osudar Argonne National Laboratory 9700 S. Cass Ave. Bldg. 205 A-051 Argonne, IL 60439-4837 (312) 972-7505 VAX-46 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 642.8 print&batch complaints 8 of 10 "Tony Carter" 6 lines 26-JUN-1987 08:02 -< Re: 642.7 >- ---------------------------------------------------------------- Using "job reset modules" is fine if you know that all files in one job are to be printed with the same characteristics. But if one file is to be printed in 132 columns and another in 80 columns then it is not good enough. Though your point is well taken. It may be worth putting up with that problem to put things where they really belong. I'll check it out. Tony Carter P.O. Box 846 Middleton, MA 01949 (617) 245-6600 ================================================================ Note 642.9 print&batch complaints 9 of 10 "Tony Carter" 4 lines 28-JUN-1987 12:38 -< Form-level modules vs Job reset modules >- ---------------------------------------------------------------- Using the LN03 reset escape sequence in the job reset module causes an extra page to be ejected. Using it in the form-level module works fine. There may be a work around for this, but I haven't figured it out yet. Tony Carter P.O. Box 846 Middleton, MA 01949 (617) 245-6600 VAX-47 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 642.10 print&batch complaints 10 of 10 "Chris Erskine" 7 lines 29-JUN-1987 12:17 -< Repeat of how to eliminate the extra pages >- ---------------------------------------------------------------- A repeat of replies from myself and others. By putting a on the first line of the RESET module with escape sequences also on the first line appears to fix the problem. I have run this on an LN03 without problem but have not been back to my other site to test on an LN01 but I think the extra page is from the spooler so the type of printer should not matter. Chris Erskine 6001 Adams Rd. Bloomfield Hills, MI 48013 (313) 258-4049 ================================================================ Note 644.15 Wanted: MAIL improvements 15 of 18 "John Osudar" 20 lines 4-JUN-1987 20:37 -< another wish list item >- ---------------------------------------------------------------- Just to add to my previous list of desired improvements: VMS MAIL lets you SET FORWARD your mail -- and this is great, since I have accounts on four or five different VAXes located in different places. However, automatic forwarding could be better: first of all, it ought to be trivial for MAIL to continue to broadcast a message telling me (if I'm logged in, of course) that someone just sent me mail on node ABC, even though that mail was forwarded to node XYZ as I directed. (In fact, it should be easy to remind me at the same time that the forwarding actually happened!) This would be a low-cost feature that would help users get more timely notification of their messages, and can be disabled by any user that doesn't want it. The other thing I don't like about forwarding probably can't be fixed without store-and-forward, which I don't think belongs in the basic VMS MAIL (although I wouldn't complain if someone gave it to me!) -- and that is the problem that, if the node to which your mail is forwarded can't be reached, the message is lost. This can be really confusing for someone who thinks they're sending a local mail message and is told that the node is unreachable. However, as I said before, I don't expect an easy fix for this one... VAX-48 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT John Osudar Argonne National Laboratory 9700 S. Cass Ave. Bldg. 205 A-051 Argonne, IL 60439-4837 (312) 972-7505 ================================================================ Note 644.16 Wanted: MAIL improvements 16 of 18 "Jamie Hanrahan" 15 lines 8-JUN-1987 14:37 -< PMDF can do store-and-forward >- ---------------------------------------------------------------- You can get store-and-forward for VMS mail at low cost by using PMDF (Pascal Memo Distribution Facility). PMDF is also the transport mechanism we've selected for the "VMS User's Network" (formerly Usenet-for-VMS; see note 511), so if you're interested in that you need PMDF anyway. To get PMDF, send $50 to: Ned Freed The PMDF Project Computing Services Harvey Mudd College Claremont, CA 91711 It comes on a magtape with all sources, doc. as .RNO and .MEM files, and hardcopy documentation as well. Highly recommended. Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 VAX-49 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.0 Applications software standards 26 replies "Kevin Angley" 70 lines 27-MAY-1987 15:05 ---------------------------------------------------------------- We are in the process of putting together an "applications software" standard which we will at least try to get third party software suppliers to adhere to. I am putting it in this forum to generate some discussion: Are these reasonable/practical? Can you think of other standards? Here is the condensed first draft: Most writers of application software develop their application and installation procedures with the naive notion that their package is the only package to run on the system. They make assumptions about the practicality of "system" level modifications, logical names, and privilege. Because we must maintain multiple applications, often for similar purposes, and because many applications are actually used by a small percentage of the user population, it is mandatory for us to provide support/changes to a user's process for a particular application only if that user is going to use that application. This means that all logicals, verb definitions, symbols, etc. will be done on a per process basis, upon the user's request. We have implemented a "SUPPORT" command procedure which does the necessary setup for each application. The application software standards follow. In some cases, minor exceptions can be made as long as the vendor is willing to cooperate in providing the software support necessary to comply with the intent, if not the letter of the standards. VMS and layered products from Digital are exempt. o Command verbs will be added on a per process basis via SET COMMAND's rather than modification of the network-wide DCLTABLES. o Symbol definitions will not be made in the cluster-wide SYLOGIN.COM. Rather, symbol definitions can be made in the application setup command procedure or in command procedure(s) nested in the application setup command procedure. VAX-50 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT o Logical names may be entered only in the process tables. Group and system logical name tables, or user-created logical name tables which require privileges to create cannot be used. o The application software must reside in directories of its own, and may not place any files in "system directories", including SYS$SYSTEM, SYS$HELP, SYS$LIBRARY, SYS$MANAGER, etc. o Applications may not depend on users having privileges beyond TMPMBX and NETMBX. o Applications may not require that images be installed with any privileges. Exceptions may be made for images that require privilege to interact with dedicated hardware (e.g. a device driver for a graphics station). In these exception cases, source code must be provided, and the image built from this source at our site. o Logical names must be of the form "xxx_name" where xxx is a unique vendor designation, the underscore must be present, and name is any acceptable name not containing "$". o Magnetic tape written in VMS "backup" format (either 6250 or 1600 BPI) is the preferred distribution media. Files-11 formatted tapes, TK50 tape cartridges, or RX50 diskettes may also be acceptable. Distribution media are kept onsite as long as they are current level, and will thereafter be destroyed if requested by the vendor. We do not return media to the vendor under any circumstances. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 VAX-51 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.1 Applications software standards 1 of 26 "Jamie Hanrahan" 11 lines 27-MAY-1987 17:40 -< jolly good show >- ---------------------------------------------------------------- I like it! I can see the need for exceptions here and there, of course...but you've allowed for that. I'd even go further re. directories: The customer should be able to specify where the files that make up the files will live (device and directory) at installation time; moreover, moving the files to some other directory(ies) should not necessitate the reinstallation of the product. Also, all vendors of VMS software should register their products through Digital's facility name registration service. Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 660.2 Applications software standards 2 of 26 "Larry Kilgallen" 160 lines 27-MAY-1987 19:32 -< Some difference of opinion >- ---------------------------------------------------------------- I have spent, and continue to spend, time on both sides of the fence -- working for the recipients/victims and for the sources/perpetrators of third party software. I admire the purpose of setting standards, but feel it is important to base such rules on what is viable, rather than have something which might (appear to) have the effect of keeping out all third party software. There are certain applications which you will NEVER get onto a VAX if you wait for a DEC product. On to specific comments: VAX-52 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT This means that all logicals, verb definitions, symbols, etc. will be done on a per process basis, upon the user's request. If a vendor has registered their product with DEC and received approval for a facility name, they have a "lock" of sorts on filenames which start with that facility name followed by a single underscore. (In fact it is a lock in the exact sense of the VMS Lock Manager, providing protection only against cooperating vendors.) I think that logical names need not be restricted to a user's process if a vendor follows the DEC registration rules. (Incidentally, for those who are not involved in the development process, this DEC registration is entirely a technical process. It has NOTHING to do with a vendor being a DEC cooperative anything marketing anything [thank goodness].) As far as DCL symbols are concerned, I consider the mark of a solid product the fact that it doesn't require ANY symbols be used. I don't like ANYONE slowing down logins, even if it is for a product I want to use. VMS and layered products from Digital are exempt. I am not certain exempting DEC from standards is quite reasonable. Is it because they seem to be an immovable object, or because you feel their code is so much better? If the latter, then a judgement of implementation quality from a third party should be the path to letting them play by (almost) the same rules as DEC. o Command verbs will be added on a per process basis via SET COMMAND's rather than modification of the network-wide DCLTABLES. Verb definitions are indeed a problem, because DEC provides no registration service for them. I suppose site-specific decisions on how to handle those (main table, alternate table, or per-process at runtime) is a good idea, but there is a decided lack of standards and assistance in this area from DEC. There is no [SYSCLD] directory, since DEC does not need one as they DO run a registry for their OWN products use of verbs. Perhaps DECUS should push DEC to set up a standard area for CLD files. VAX-53 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT o Symbol definitions will not be made in the cluster-wide SYLOGIN.COM. Rather, symbol definitions can be made in the application setup command procedure or in command procedure(s) nested in the application setup command procedure. From a vendor perspective, I think the best approach is not to make the assumption that there IS any logical name SYS$SYLOGIN. There is no rule that a system-wide login command procedure has to exist, the feature is just there if a site should WANT to use it. Given that setting, vendors who want to sell their product into a maximum number of sites will make it work without such symbols. o Logical names may be entered only in the process tables. Group and system logical name tables, or user-created logical name tables which require privileges to create cannot be used. On this I disagree. Provided that a vendor has followed the DEC facility registration and naming convention for logical names, there should be no problem putting them into the system logical name table. To implement certain site goals (such as Jamie's comment in .1 about putting the data on a different disk) are most cleanly implemented with system logical names (conforming to the facility registration naming standard, of course). o The application software must reside in directories of its own, and may not place any files in "system directories", including SYS$SYSTEM, SYS$HELP, SYS$LIBRARY, SYS$MANAGER, etc. Again, I disagree. I want the FOOBAR software (a product of FOOBAR, International Megacorp) release notes for version 5.8 to be just where I would expect to find them: SYS$COMMON:[SYSHLP]FOOBAR*058.RELEASE_NOTES. That is where VMSINSTAL puts them, and I want vendors to conform to EVERY last nit regarding VMSINSTAL usage, so their behaviour is PREDICTABLE (there being so many of them). VAX-54 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Furthermore, only by following the VMSINSTAL standards can I be assured that a product will install properly on an ALTERNATE ROOT (option R to VMSINSTAL). This is the case where one carries the system pack from a target machine over to the machine which has a tape drive to do the installation. This goes against Jamie's goal of requiring the vendor to put things in any location the site specifies. I think that is good in certain situations (large data files), but for the typical product I would much prefer they follow the DEC standards so I know that defining a process logical for their image name will direct it elsewhere (knowing it would be in SYS$SYSTEM, they just refer to it by name, rather than full filespec). o Applications may not require that images be installed with any privileges. Exceptions may be made for images that require privilege to interact with dedicated hardware (e.g. a device driver for a graphics station). In these exception cases, source code must be provided, and the image built from this source at our site. Here we get to the dual standard. How many DEC products provide source? Have you never had a DEC product crash your VAX? I think a more reasonable standard would have to do with ensuring that any required privileges for installing images be "appropriate" for the product. Consider a database, for instance. There are at least some special-purpose databases which do things in a significantly different fashion than DBMS, Rdb or RMS. One example which comes to mind is Basis, which is centered around text-retrieval. If I had an application need particularly suited to Basis, I would want to be assured that they use appropriate technology for VMS (so many "portable" products do not; I have no idea whether Basis does or not). In databases, on VMS, "appropriate technology" in my view means making use of the VMS lock manager. But this simply cannot be done across groups (another feature I want them to support) without using the SYSLCK privilege. Now the hazards of SYSLCK misuse are almost entirely restricted to denial of service, but that can only happen if someone misdeclares their root lock resource name to match that of some other facility. You have to trust a vendor to some degree. After all, you are trusting your organization's data to their care! VAX-55 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT On the other hand, one database vendor I DO know about, Software AG, claimed up and down that they needed to run with LOG_IO privilege in order to run with their database file aligned on cylinder boundaries EVEN THOUGH THE DATABASE FILE HAD BEEN PREALLOCATED. The same database could not survive a successful image backup. It's not easy to do a thorough analysis of third party products as fast as other forces in your organization throw them at you, but sharing information with other DECUS members can help. o Logical names must be of the form "xxx_name" where xxx is a unique vendor designation, the underscore must be present, and name is any acceptable name not containing "$". And xxx must be registered with DEC! o Magnetic tape written in VMS "backup" format (either 6250 or 1600 BPI) is the preferred distribution media. Files-11 formatted tapes, TK50 tape cartridges, or RX50 diskettes may also be acceptable. Distribution media are kept onsite as long as they are current level, and will thereafter be destroyed if requested by the vendor. We do not return media to the vendor under any circumstances. I do not consider anything but Backup to be acceptable, both because of reliability concerns and because one must be using Backup to follow what I consider to be the number one rule and litmus test: THOU SHALT USE VMSINSTAL !!! I do not consider 6250 tape to be reasonable, since 1600 is the density that all 9-track drives DEC has provided will handle (and the deviation of third party drive vendors tends to be to support MORE densities than DEC, rather than fewer). If people deal in 6250 tapes for software distribution, it increases the chance that the vendor will ship one to a site which only has 1600 bpi. Naturally an option may be appropriate where a product will not fit on a single large reel at 1600 bpi, but the only case that large of which I am aware is the VMS source kit! Larry Kilgallen VAX-56 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 660.3 Applications software standards 3 of 26 "Jamie Hanrahan" 34 lines 28-MAY-1987 19:57 -< compromise re. file placement >- ---------------------------------------------------------------- The desire to have third-party application software live outside of the VMS-defined system directory tree stems, in my case, from several sources. First, DEC has told us in several places to not put anything other than DEC-defined files in system directories. Second, I'm paranoid: I want to know, or at least be able to find out, what each and every file on the system disk is for, who put it there and when, etc. If I can put third-party products in their own trees, this task is made easier. Third, there often comes a time when we want to REMOVE a product from the system disk. Again, if the product lives in its own tree, there's less work to do. (I've been making do with a command procedure which does a directory/date/security sys$sysdevice:[*...] into a file, then uses differences to compare that file with the previous version. This is run weekly, and also immediately before and after each s/w installation.) I think we can all agree on the following, which would go a long way toward satisfying my desires in this area: 1. The installation (of other than a DEC product) should not modify any DEC-supplied files, specifically including DCLTABLES.EXE and *.HLB, but excluding directory files. 2. The product documentation should include a complete list of EVERYTHING that the install procedure does to the target disk(s). 3. A command procedure or other means should be supplied whereby the product can be removed from the disk. If someone installs the product and then removes it on an otherwise-idle system, the free block count of the disk should show a net zero change. Jamie Hanrahan Simpact Associates VAX-57 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 660.6 Applications software standards 6 of 26 "Frank J. Nagy" 26 lines 1-JUN-1987 09:04 -< SET COMMAND: slow and non-shared >- ---------------------------------------------------------------- Well, one problem is that SET COMMAND business. As most of us are aware, SET COMMAND is just not very fast. In fact, you can almost count how many are in your LOGIN file by how slow it is. There is also the hidden problem of the DCLTABLES section. At login, each user is mapped to this shared global section. When you do a SET COMMAND, a private copy of DCLTABLES is made for that user. If EVERYONE on the system does a SET COMMAND, then lots of memory gets used and lots of page file and lots of paging since everyone has a separate copy of DCLTABLES! This has lead me to writing utilities which: a) use the LIB$GET_FOREIGN command line facility to get the command line b) tack a verb onto the front of the return from GET_FOREIGN. c) call CLI$DCL_PARSE with the resultant string and the CLD tables which were compiled into an object module and linked into the utility. This allows quick definition of the utility "verb" as a foreign command while still allowing my code to make use of the CLI command-decoding services and providing DCL-like syntax for my users. Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 VAX-58 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.7 Applications software standards 7 of 26 "Frank J. Nagy" 10 lines 1-JUN-1987 09:09 -< Foreign entries: DCLTABLES, HELPLIB... >- ---------------------------------------------------------------- When I installed VMS V4.4, I had to do a bit of playing about. I seem to remember noticing that the VMS upgrade procedures went about figuring out what "foreign" items were in DCLTABLES.EXE, HELPLIB.LIB and IMAGELIB.OLB (maybe others also). These "foreign" items were extracted and moved into the new copies of the files being installed with V4.4. It seems that DEC has developed itself into this hole since the "foreign" items being moved were for various DEC layered products! Maybe "user" items in these common files are "safe" after all (provided you don't conflict with a Digital name). Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 ================================================================ Note 660.8 Applications software standards 8 of 26 "Larry Kilgallen" 16 lines 1-JUN-1987 12:44 -< DCLTABLES entry preservation is not guaranteed >- ---------------------------------------------------------------- RE: 660.7 -- Upgrades preserving DCLTABLES entries Indeed, upgrades since VMS V4.0 have preserved such entries (except on MicroVMS, where the "system manager" is most likely to need such protection). I once considered looking through the upgrade kit to find the program they use (some DEC customer must have done so, but I never had the time). I presume the technique is similar to that used by programs on various DECUS tapes to dump the contents of DCLTABLES. BUT, we cannot COUNT on this compatibility unless DEC commits to providing it. I presume one good reason they might not commit to it is the possibility that the rules for writing CLD files might change in the future. A change as large as the one from V3 to V4 might be grounds for non-support. VAX-59 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 660.9 Applications software standards 9 of 26 "Michael R. Pizolato" 30 lines 2-JUN-1987 16:47 -< Thoughts... >- ---------------------------------------------------------------- Some thoughts: 1. ALL SOFTWARE MUST USE VMSINSTAL FOR INSTALLATION. GET THE MANUAL "DEVELOPER'S GUIDE TO VMSINSTAL!" 2. To me, one of the better definitions of user-friendly is "consistent." That is, I want everything on my system to look and act like everything else. So, when I install a third-party software product, I expect it to be like a DEC product. I don't believe in restricting command definitions, logical name or file placement, etc., except to stay within the product registration rules. 3. I think that the suggestion of a "de-install" procedure is a good one. Even DEC products should have such an animal. Perhaps a "VMSDESTAL.COM" procedure is in order here. 4. Any good product will not require global symbols in SYLOGIN.COM to run, unless it's a command procedure (yes, I do believe that major tools can be implemented in DCL). 5. Including a description of EVERYTHING that an installation does to the system should be mandatory. 6. Is product registration restricted to those vendors who are going to sell their stuff? I have a thing or two that I'd like to distribute through DECUS for free (and I'm always working on more); they use VMSINSTAL, and I'd like to register them, or at least find out what all the standards are so I can be sure I'm following them. Can I do that? VAX-60 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Michael R. Pizolato AT&T Technology Systems Dept. 323610 555 Union Blvd. Allentown, PA 18103 215/439-5500 ================================================================ Note 660.10 Applications software standards 10 of 26 "Larry Kilgallen" 15 lines 2-JUN-1987 19:49 -< No Financial Test for Facility Registration >- ---------------------------------------------------------------- Absolutely you can get a facility name from DEC for things you plan to give away as well as things you plan to sell. You don't even have to know which it is at the time. The major test should be whether the software is going to be widely used outside your organization. There is no need for registration of site-specific software, so do DEC a favor and police that on your own. Another consideration is that there is a limited address space for "facility codes" used in message files. This is really only required if you have something which is going to be called by software that others write, so don't say you need a facility code unless you REALLY need one. (This is my own advice, it does not represent anything DEC has suggested.) Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 660.11 Applications software standards 11 of 26 "Kevin Angley" 35 lines 15-JUN-1987 16:55 -< The Devil's Advocate >- ---------------------------------------------------------------- Thanks for the comments so far on applications installation standards. It seems that most people are "gotta do VMSINSTAL, just like a DEC layered product" kind of guys. Hmmm... let me voice the opposite school of thought. VAX-61 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT I think VMSINSTAL is real neat for DEC layered products. DEC keeps internal track of what product plays with what, and the effect of having combinations on your system. Also, new versions of VMS or layered products take all this into account when doing updates. This same compatibility is not there with the various vendors. It has also been my experience that vendors don't consider all possibilities of configurations out there (like the differences in system directory structures on clustered and non-clustered systems). I have come to abhor a vendor using VMSINSTAL because it turns over my system to a possibly ignorant vendor to automatically go to town with all privileges. I much prefer a manual installation. What? Yes, I prefer to see what each and every step is doing. In my mind, the best installation instructions would read as follows: 1) Copy the directory tree off the tape to anywhere you want it. 2) Point a logical name to the base of this directory. 3) Tell the user to do $ @logical_name:SETUP in their LOGIN.COM if they want to use this application. and that would be it. I can see some advantages to VMSINSTAL, but I like this way much better. Further comments (perhaps reread 660.0 first). Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 VAX-62 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.12 Applications software standards 12 of 26 "Chris Erskine" 108 lines 16-JUN-1987 08:46 -< Overall Comments on the Standards >- ---------------------------------------------------------------- One thought for these standards. Are you going to get all companies to follow them. I know of one company which you will have a very hard time getting to change their "standards". This company is the one that we all love and curse (DEC). o Command verbs will be added on a per process basis via SET COMMAND's rather than modification of the network-wide DCLTABLES. As stated else where, SET COMMAND is a DOG. Why not put the symbols in the standard DCL tables where they belong. So what if a DEC update replaces the tables and does not preserve the old entries. If it is due to a change in the rules for writing the CLD files then even if you were to do SET COMMAND commands for the users, they would still break. There are times now when you upgrade your system that you have to reinstall the DEC standard products. o Symbol definitions will not be made in the cluster-wide SYLOGIN.COM. Rather, symbol definitions can be made in the application setup command procedure or in command procedure(s) nested in the application setup command procedure. How many people have had to define symbols for DEC products in their SYLOGIN file? The big one that I can think of is DATATRIEVE. o Logical names may be entered only in the process tables. Group and system logical name tables, or user-created logical name tables which require privileges to create cannot be used. VAX-63 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT I can see not putting entries in a group logical table since DEC states that these may not be supported at a later time but why not user-created tables which are created at startup. As far as I can tell this is where most of these logicals may belong. I have seen locations where the put hundreds of logicals in the system tables. This slows down the system for translating them and also makes it hard to find a given logical when you need to. By creating a logical table which uses the registered logical name suffix, you can find the logicals which relate to the given product. The logicals which would be placed in system space would be some of the generic ones such as pointers to the disk that the applications lives on. This requirement is one which should be taken back to DEC for some of its products. (If you have an active All-In-1 mail system, you can have hundreds of OA$SHAREn directories. This system belongs in a named table.) o The application software must reside in directories of its own, and may not place any files in "system directories", including SYS$SYSTEM, SYS$HELP, SYS$LIBRARY, SYS$MANAGER, etc. Nice idea but I like some of the features that come with these directories. Shared libraries must either be installed from SYS$LIBRARY or be defined as a logical. Logicals are nice but they do slow down the system for translations. Standard locations for startup files also help maintain the system. o Applications may not depend on users having privileges beyond TMPMBX and NETMBX. This is dependent on the application. I know of some "applications" which require the user to have privileges to run it and that is the way it should be. This includes "applications" which are system support type of applications. o Applications may not require that images be installed with any privileges. Exceptions may be made for images that require privilege to interact with dedicated hardware (e.g. a device driver for a graphics station). In these exception cases, source code must be provided, and the image built from this source at our site. VAX-64 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT The SYSLCK privilege was already stated but there are additional privileges which applications need to protect the system and allow the application to work. One that comes to mind is a file transfer application which puts files in user directories. This program requires SYSPRV at least if not BYPASS depending on how the system was put together. I can understand your worrying about what these applications may be doing to your system by way of security but if you can not trust the third party to write code which does not change your system or play games with it then you should be writing your own code. Towards creating these standards, I think we need to come work with DEC on these standards so that they do not then come out with something which will break the applications which meet our standards. These things would include not only the registration of the logical names prefixes but also thing like command verbs, locations these files should be placed when installing them, ... By working with DEC to develop the standards, the required utilities to preserve the locations or code could also be developed. This includes things like extracting the foreign DCL verbs and help files. The idea of the VMSDESTAL procedure is very good and needs to be expanded. This would also help DEC for their products for such things as loaner products and monthly rentals of software. Chris Erskine 6001 Adams Rd. Bloomfield Hills, MI 48013 (313) 258-4049 VAX-65 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.13 Applications software standards 13 of 26 "Chris Erskine" 19 lines 16-JUN-1987 09:00 -< The Devil Loses Again >- ---------------------------------------------------------------- Towards .11 The VMSINSTAL is the way to install software. What is needed is good information on installing products and using VMSINSTAL. The DEC manual DEVELOPER'S GUIDE TO VMSINSTAL is a starting point but Kevin's comment to understanding the different types of systems goes for DEC as well as third party vendors. The DEC product development groups do not understand the different types of system which are used. This includes products like All-in-1 and PCFS (PC Files Services) which require the SYSUAF file to be in SYS$SYSTEM: rather than using the SYSUAF logical if it is defined. PCFS even defines its own SYSUAF logical!!!! Other products which do not understand the different systems are a number of the communication type of products. The Terminal Server groups are one of these. The new DSV_CONFIG procedure installs into SYS$COMMON but then tries to use its data file from SYS$SPECIFIC. Chris Erskine 6001 Adams Rd. Bloomfield Hills, MI 48013 (313) 258-4049 ================================================================ Note 660.14 Applications software standards 14 of 26 "Michael R. Pizolato" 16 lines 16-JUN-1987 12:08 -< Group logicals to go? No!! >- ---------------------------------------------------------------- In reply to .12: I can see not putting entries in a group logical table since DEC states that these may not be supported at a later time VAX-66 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT WHAT!?#@$&^%*!!?*#&@^%%@ This cannot be! We use group logical names all the time! They are an integral part of the way I run my system! No, no, no! For the one who would do such a thing, a plague on him and his seed for a thousand generations! |-( Really, though, I guess I could get around it with user-created tables and ACLs. But, please, don't give me a heart attack; I'm only 26. When and where was this announced? What other such goodies lie in wait? Michael R. Pizolato AT&T Technology Systems Dept. 323610 555 Union Blvd. Allentown, PA 18103 215/439-5500 ================================================================ Note 660.15 Applications software standards 15 of 26 "Reece Pollack" 14 lines 16-JUN-1987 14:04 -< Only process logical names? >- ---------------------------------------------------------------- Logical names only in the process logical name table? You must really love your coffee breaks, at the rate of one per application system entry. We are working on a scheme to place an application's logical names in a logical name table specific to the application, and adding the table to the user's table search list upon entry to the application. This requires only a few line of DCL rather than all of the defines each time. When the user leaves the application, the table is removed from the search list and they're all gone. I once worked at a site where the system login command file took almost 5 minutes to run on an unloaded 11/780, because the MIS director didn't want anyone to have to do anything in his own login file to get to anything. This has made me *very* careful about what I force a user to go through to get to an application. Reece Pollack American Satellite Company MS 34/MIS 1801 Research Blvd Rockville, MD 20850 VAX-67 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.18 Applications software standards 18 of 26 "Jamie Hanrahan" 34 lines 16-JUN-1987 16:35 -< VMS should have more support for extensions >- ---------------------------------------------------------------- Re. Chris Erskine's comments: o Command verbs will be added on a per process basis via SET COMMAND's rather than modification of the network-wide DCLTABLES. As stated else where, SET COMMAND is a DOG. Why not put the symbols in the standard DCL tables where they belong. So what if a DEC update replaces the tables and does not preserve the old entries. The application software must reside in directories of its own, and may not place any files in "system directories", including SYS$SYSTEM, SYS$HELP, SYS$LIBRARY, SYS$MANAGER, etc. Nice idea but I like some of the features that come with these directories. Shared libraries must either be installed from SYS$LIBRARY or be defined as a logical. Logicals are nice but they do slow down the system for translations. Standard locations for startup files also help maintain the system. Both of these ought to be addressed by VMS. For each "special" directory like SYS$LIBRARY there should be another directory for site-specific files analogous to those in the "system defined" directory. Similarly for DCL tables: There should be a way to provide a site-specific extension to the DCL tables without modifying DCLTABLES.EXE *and* without doing SET COMMANDs to modify the process-private tables (which do indeed take a while). VMS provides many mechanisms for extending certain parts of the system "non invasively". It should provide some more. Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 VAX-68 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.19 Applications software standards 19 of 26 "Larry Kilgallen" 23 lines 16-JUN-1987 16:49 -< Extensions come in many flavors >- ---------------------------------------------------------------- Jamie's comments on extensibility are well founded, but let's not be limited in vision when considering needed extensions. I can think of several name spaces: DEC other vendors (registered with SQM) site-specific cluster-specific node-specific This has come up in a discussion of MODPARAMS.DAT. I may have a vendor-inspired requirement (or desire, for performance sake) that MAXBUF be at least 3000. I may have a management decree for the site that LGI_xxx be set to 7. I may have a usage requirement on this cluster that VIRTUALPAGCNT be 300000. I may have a requirement on just this node that WSMAX be only 500 (due to low memory). When the management decree changes to make LGI_xxx be 9, I want to be able to COPY the standard file around via DECnet, but I certainly don't want to wipe out node-specific values for other parameters!!! Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 660.20 Applications software standards 20 of 26 "John Saunders" 13 lines 22-JUN-1987 21:32 -< WHAT Registration Service? >- ---------------------------------------------------------------- Also, all vendors of VMS software should register their products through Digital's facility name registration service. VAX-69 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT As a vendor or VMS software, I suppose this means me! However, I never heard of the facility name registration service until I read this note. I've spent the last two weeks trying to find someone at DEC who knows about this, but haven't had any luck. Apparently, the VMS product manager doesn't even know about it! Would someone please tell me who I should call about registering our facility name? Also, does this service handle device name registration? John Saunders FEL Computing PO Box 72 Williamsville, VT, 05362 (802) 348-7171 ================================================================ Note 660.21 Applications software standards 21 of 26 "Larry Kilgallen" 11 lines 23-JUN-1987 06:41 -< THE Registration Service >- ---------------------------------------------------------------- The registration service is described in some detail starting on page VAX-24 of the November 1986 Pageswapper (PAGSWP804.MEM on the Fall 1986 SIG tape if you do not have the hard copy). The official DEC documentation is in the Developer's Guide to VMSINSTAL, an optional book (i.e., not in the "complete" VMS documentation set) released at the time of VMS V4.4, about a year ago. They do not register device names. Yes, it would be nice if they did. They know that. Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 VAX-70 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.22 Applications software standards 22 of 26 "Brian Tillman, Lear Siegler, Inc." 9 lines 24-JUN-1987 11:14 -< Need multiple CLI tables. >- ---------------------------------------------------------------- RE: < Note 660.12 by NODE::US204054 "Chris Erskine" > -< Overall Comments on the Standards >- As stated else where, SET COMMAND is a DOG. Why not put the symbols in the standard DCL tables where they belong. I think DEC should support multiple CLI tables, like search lists. Then I could put all additional commands in an additional table and leave DCLTABLES alone. Brian Tillman Lear Siegler, Inc. 4141 Eastern Ave. MS121 Grand Rapids, MI 49518-8727 (616)241-8425 ================================================================ Note 660.23 Applications software standards 23 of 26 "Chris Erskine" 12 lines 25-JUN-1987 08:11 -< Group Logical Tables >- ---------------------------------------------------------------- The information on group logical tables going away was told to me by a DEC instructor when I took the V4.0 VMS Update course. Some if this can be seen with the literal UIC symbols. The fact the the group name is added in AUTHORIZE by using the addition of account says that DEC is thinking of changing something. Also the Update manual talks about the group table being used for compatibility with previous releases of VMS. Who knows what lurks in the halls of DEC. Chris Erskine 6001 Adams Rd. Bloomfield Hills, MI 48013 (313) 258-4049 VAX-71 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.24 Applications software standards 24 of 26 "Brian Tillman, Lear Siegler, Inc." 6 lines 25-JUN-1987 14:53 -< Bye, bye, groupie >- ---------------------------------------------------------------- Page 196 of "VAX/VMS Internals and Data Structures", first edition, also gives some indication that group logical name tables are going away. There is no symbolic name for the group logical name table mutex, implying it will someday go away. It doesn't matter, though, since anything you can do with group tables can be done with a table you create yourself. Brian Tillman Lear Siegler, Inc. 4141 Eastern Ave. MS121 Grand Rapids, MI 49518-8727 (616)241-8425 ================================================================ Note 660.25 Applications software standards 25 of 26 "Frank J. Nagy" 5 lines 27-JUN-1987 10:32 -< Good idea: Multiple DCL tables >- ---------------------------------------------------------------- I think one of the best ideas to come out this discussion and one which should not be too difficult to implement is the idea about having multiple DCL tables or some sort of search list for DCL command tables. Lets put it on the SIR ballot and stuff the ballot box! Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 VAX-72 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 660.26 Applications software standards 26 of 26 "Kevin Angley" 13 lines 27-JUN-1987 11:34 -< A couple random replys >- ---------------------------------------------------------------- The "applications standard" list was meant to inspire discussion and it looks like it has. We have seen many suggestions on how VMS could be improved to make the applications world play together better. The concept of a VMSDEINSTAL would alleviate most of my concerns about putting things in "system" directories. Just as an aside, I wonder if version 5's VMSDEINSTAL_THOSE_PARTS_OF_VMS_I_DONT_NEED will be along the lines of a VMSDEINSTAL. A previous reply about the system manager forcing all users to go through a 5-minute system login to setup support for all these different applications whether the user needed them or not is exactly the type mentality that the "standards" were meant to avoid. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 ================================================================ Note 661.0 Info on DMR option switches needed No replies "Jamie Hanrahan" 7 lines 27-MAY-1987 17:30 ---------------------------------------------------------------- Does anyone know EXACTLY what the "DMC compatibility mode" and "high-speed mode" switches on the DMR11 line unit are doing, in terms of what's going on on the communications line? A guess is that DMC mode causes a change in the line protocol startup sequence, and that high-speed mode causes the beast to not send abutting packets, but this is only a guess. Authoritative information would be appreciated. Jamie Hanrahan Simpact Associates VAX-73 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 662.0 VAX C Bug 3 replies "Jack Patteeuw" 16 lines 1-JUN-1987 07:29 ---------------------------------------------------------------- The other day I had the unfortunate job of telling a UN*X "freak" that he was right and that VAX C (V2.3) was wrong ! Seems that if you try and print a long string (>512 characters) using the printf function with a %s format conversion specifier, the program will Access Violate ! The wonderful people at Colorado Spring said "sure, we know about this !" As a matter of fact, they were the ones to tell my the 512 number. Of course NONE of this is documented ! Come on DEC, **NO** compiler/run-time library should generate/contain code that can cause a program to Access Violate. With problems and answers like that, how long can we VMSers keep those little UN*X "devils" out ? Jack Patteeuw Ford Motor Co. Electrical and Electronics Division 31630 Wyoming Livonia, MI 48150 313-323-8643 VAX-74 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 662.1 VAX C Bug 1 of 3 "Bill Mayhew" 26 lines 11-JUN-1987 23:18 -< ...and lseek() won't! >- ---------------------------------------------------------------- I have another VAX C Bug, or at least dreadfully poor implementation decision, to add to the list. It seems that lseek() doesn't. If you have a "normal" variable-record-length RMS file, and open that file using VAX C's "UNIX-compatible" open(), and seek to the end of it (or within it, for that matter) using the "UNIX-compatible" lseek(), lseek() *reads the whole bloody file* to get there. I found this out when I moved a "high-performance" database manager which had been developed on a UNIX clone using Whitesmiths' C, and subsequently ported to RSX using DECUS C and our own RMS interface library, to VAX C, and performance fell through the floor. To open and seek to the end of a 2500-block file is taking from 12 to 30 seconds (depending on load) on a 785 with RA81s; on a VS2000 with an RD32, it's taking 6 seconds to deal with a 1000-block file. Now yes, I understand that you can't expect to do arbitrary seeks within a variable-record-length RMS file without some troubles, but if your seeks are at record boundaries there's a straightforward way for RMS to do the job; and there's certainly a straightforward way for RMS to move you to end-of-file. Unfortunately, lseek() isn't doing it. The Telephone Support Center has been researching this for a week with no real progress. I would be deliriously happy if someone knows where open() leaves its FAB and (particularly) RAB so I can build a suitably-intelligent lseek() for my application. Still waiting for the heads to move, Bill Mayhew Village Systems Workshop Inc PO Box 642 Natick MA 01760 617-237-0238 VAX-75 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 662.2 VAX C Bug 2 of 3 "Bill Hoelzer" 15 lines 17-JUN-1987 08:49 -< Changes to C 2.3 >- ---------------------------------------------------------------- We have found a change between version 2.2 and version 2.3 of C that has caused a great deal of trouble at our site. Function definitions have been moved to different include files. For example, the function atof had been declared in the file math.h. It is now declared in the file stdlib.h. Therefore, any program that used this function and included the file math.h will no longer work correctly unless the file stdlib.h is also included. Please note that the help file containing the information about the run time functions has not been updated to reflect this change. Bill Hoelzer Quorum Systems 5165 Campus Drive Plymouth Meeting, PA 19462 ================================================================ Note 662.3 VAX C Bug 3 of 3 "Frank J. Nagy" 5 lines 20-JUN-1987 10:46 -< V2.3 changes not a bug! >- ---------------------------------------------------------------- I believe that the V2.3 change to move the definition of atof() from math.h to stdlib.h was called for in the ANSI standard. There are likely to be other such changes lurking about since V2.3 of VAX C is compatible with the draft ANSI C standard now up for review/acceptance. Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 VAX-76 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.0 Comments on the SPR process 24 replies "Jack Patteeuw" 36 lines 1-JUN-1987 07:55 ---------------------------------------------------------------- When I first became a VAX System Manager about 4 years ago, I thought the SPR process was a very good idea; a path directly into VMS engineering! Furthermore, I felt it was my "duty" to my fellow System Managers to report ALL problems as quickly as possible via SPR so they fixed or at least documented for others. Lately I have written very few SPR's for the following reasons: First, it take **MONTHS** to get a response from a written SPR. I like to write (actually have them typed) my SPR's because that way I have a written document of the exact correspondence and know the date and other particulars. I've been told, however, that all SPR's are forwarded to Colorado Springs, so this probably adds a week or two to the processing time. Second, most of the time I get a "form letter" response that either a) does not really address my problem (like the one that discussed queues on clusters when I clearly stated that I had a standalone 11/780) or b) states that they don't have enough data (and this was after I sent them 5 RX01's worth of ALL-IN-1 traces, forms, named data, etc. !!) I even had one case where the response said they couldn't verify my problem because they did not have the same type of terminal I had (yes it was a DEC terminal). Hey DEC, if you don't understand my problem description (I know I'm not the world greatest writer) why not **CALL ME**, you got my phone number right there on the form ! The only way I am really assured of getting an answer to a serious problems is by first submitting an SPR and waiting for the silly ass response and then kicking my local office (Thank God we can afford a Specialist) to go through the back door and get the problem resolved. What about the rest of you in DECUS land ? Good or Bad ? Can we (DECUS) do something about making it better ? Jack Patteeuw Ford Motor Co. VAX-77 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Electrical and Electronics Division 31630 Wyoming Livonia, MI 48150 313-323-8643 ================================================================ Note 663.1 Comments on the SPR process 1 of 24 "Larry Kilgallen" 34 lines 1-JUN-1987 12:57 -< SPR responses are horrible -- A Proposal >- ---------------------------------------------------------------- I find the SPR response situation uniformly horrible. I understand the potential for overload, but the Colorado crew is so used to dealing with people who have not read the manual, that they can think of nothing else. I had a developer tell me at Nashville that a bug I had noted in the Pageswapper was now fixed. The developer never mentioned having received the corresponding SPR, which was sent in a minimum of one month before the Pageswapper hit the streets (newsletter production times being what they are). Realizing the potential for inundating DEC with misformed SPRs, I would propose the establishment of a "point" system. The INDIVIDUAL SUBMITTER would be given 5 negative points for every blatantly incorrect SPR (e.g., where the answer is documented, etc.). The INDIVIDUAL SUBMITTER would be given 5 positive points for every incorrect response received (e.g., responder did not understand a clear statement, etc.). The INDIVIDUAL RESPONDER would be given 5 negative points for every incorrect response generated. The INDIVIDUAL RESPONDER would be given 5 positive points for every non-trivial CORRECT response given. All scores would have their absolute values decrease as some function of time. And now for their prize: high scoring submitters would be guaranteed THEIR SPRs would be handled by high scoring responders. VAX-78 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Question: Does the customer get a software maintenance refund if there ARE NO high scoring responders? Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 663.2 Comments on the SPR process 2 of 24 "Larry Kilgallen" 6 lines 1-JUN-1987 13:00 -< DUTY to report bugs, vs SPR morass >- ---------------------------------------------------------------- I think we have all felt Jack's "sense of duty" for reporting bugs. If DEC cannot make the system work, I would suggest using DECUS channels to share information about known defects. Perhaps that will shame the folks responsible for the current SPR debacle into cleaning up their act. Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 663.3 Comments on the SPR process 3 of 24 "Bob Hassinger" 77 lines 1-JUN-1987 13:46 -< We have problems too, how about you? >- ---------------------------------------------------------------- > What about the rest of you in DECUS land ? Good or Bad ? I have a similar complaint. In recent years I have tended to not submit SPRs because they do not seem to be getting much attention and they have not been yielding useful results, either directly or through corrections to future releases. It has been several years since I received a useful answer via the SPR process and response time has been very bad by previous standards. At one time DEC stated a commitment to a 30 day response. There seems to be no trace of that commitment left now, based on my experience. VAX-79 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT My most recent experience has been with an SPR regarding a potentially serious difficulty with BACKUP. The problem seems particularly bad because you do not find out about it until you go to restore your large crashed disk from the image plus incremental backups you have been making per DEC's instructions. I think it would be unusual for most users to test their backup procedures for the particular conditions that produce the problem but in practice they occur easily and are quite common given the current recommended procedures. First I tried everything I could think of to resolve the problem or find what I was doing wrong when I could not I reduced the problem to a very simple case and contacted the Telephone Support Center. We went around about it for a week or so and they finally confirmed that it did seem to be a real problem, they said they could not resolve it and they suggested I might want to submit an SPR. On 12 Feb 87 I submitted an SPR. I outlined the problem in this conference (IO 588) the same day. I received an acknowledgment dated 19 Feb 87 with the reference number "ICA-4265" (a rather odd looking number compared to past experience by the way). Since then I have heard NOTHING from DEC. I have written DEC and I have gone through the designated contacts at my local office in an effort to get a response. I asked about the problem at our LUG meetings and eventually, on 1 May 87, I even asked about it in the VAX Advanced Q and A session at the Nashville Symposium. The responsible DEC person on the panel claimed to have never heard of the problem and promised I would get an answer but it is now another month and so far nothing! Reviewing IO 588 you can see that this is either a simple problem that can be resolved by reviewing the material I submitted and telling me what I am doing wrong or else it is a serious problem that must be recognized, the users warned, work-arounds published and a fix made. I think that to just totally ignore all my efforts to report the problem and trace what has happened to the report is unacceptable. We pay a lot of money for SPR service and it seems to me we deserve better service and treatment! Is anyone else having problems like this or are Jack and I isolated cases? VAX-80 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT > Can we (DECUS) do something about making it better ? Remember a few years back the RSX community made things very uncomfortable for DEC when many felt they were not getting what they needed and were paying for in the SPR process. I think the first step might be for DECUS to insist that DEC document what their current goals/commitments for SPR response time and quality are. Then we could each evaluate what we receive against what they promise. On the other hand, if they fail to responded constructively then that would serve as a basis for the next step. Bob Hassinger Liberty Mutual Research Center 71 Frankland Road Hopkinton, MA 01748 617-435-9061 ================================================================ Note 663.4 Comments on the SPR process 4 of 24 "Jamie Hanrahan" 25 lines 1-JUN-1987 15:20 -< We have the technology... >- ---------------------------------------------------------------- I'm afraid I too have gotten out of the habit of submitting SPRs. Mostly this is because the form says "MUST BE TYPEWRITTEN"; there are exactly two typewriters in this 120+ person company, and neither is convenient to my desk. Nor is either machine particularly good at making six carbons plus original, nor am I good at typing things exactly right the first time (the typewriters have lift-off correction systems, but that's of little use on the carbons). On the other hand, since we started running VMS field test software, I've been dialing into the field test Quality Assurance Report system at least once a week. That's where I report bugs, etc. Since we're on self-maintenance we don't have access to DEC's "Software Information Network". Why can't DEC set up an on-line SPR system, one that would be accessible even to self-maintenance customers? I think that most programmers and system managers would be far more likely to submit SPRs through such a system than they are to send them in by mail. I know I would be. VAX-81 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT The current paper mail SPR system is "stone knives and bear skins" technology. You would think that DEC, the company that prides itself on its networking capabilities, could move its SPR system at least into the clay tablet era... Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 663.5 Comments on the SPR process 5 of 24 "Kevin Angley" 23 lines 1-JUN-1987 16:10 -< DSIN is in-DSIN't >- ---------------------------------------------------------------- Well, Jamie, you could use the SPR generator from Jim Downward's KMSKIT rather than a typewriter (DEC will take these forms as long as you staple it to a blank SPR form) ... but that probably misses your point. Response to SPR's, bugs reported to Telephone Support, etc. has been disgusting. Meanwhile, DEC is ADVERTISING their whizbang up to the minute news flash Software Information Network ... but don't believe a word of it. Has anyone ever seen a bug they reported show up in DSIN?? I haven't. Even the potentially disastrous F11BXQP 4.5 bug (index file extents in a cluster) has not appeared on DSIN. Neither was the fact that the patch being sent out for that bug had an even worse bug. Nope ... the only "news flashes" you'll find on DSIN are things like "VMS 4.0 ANNOUNCED!!" (no kidding). It is a real shame ... DSIN could have been a great tool ... but what it needs to be is access (read only is O.K.) to the same software problem reporting database used by Digital. It is not. The ONLY items that get put on DSIN are those which one of the support people physically enter as an "article" and this is not a priority thing with them. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 VAX-82 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.8 Comments on the SPR process 8 of 24 "Michael R. Pizolato" 5 lines 2-JUN-1987 16:06 -< It's bad, bad, bad. >- ---------------------------------------------------------------- My experiences with SPR's (and Problem Submitals to Colorado) has been so bad that I just call Colorado for ALL my troubles, and to hell with waiting for anything. That's probably not being a good neighbor, but it's the only way I can get answers. Michael R. Pizolato AT&T Technology Systems Dept. 323610 555 Union Blvd. Allentown, PA 18103 215/439-5500 ================================================================ Note 663.9 Comments on the SPR process 9 of 24 "Michael R. Pizolato" 8 lines 2-JUN-1987 16:13 -< Self-maintenance and no DSIN? >- ---------------------------------------------------------------- In reply to .4: Since we're on self-maintenance we don't have access to DEC's "Software Information Network". What kind of "self-maintenance" are you on? We are on hardware self-maintenance but I still have DSIN access because I'm on a software contract. Michael R. Pizolato AT&T Technology Systems Dept. 323610 555 Union Blvd. Allentown, PA 18103 215/439-5500 VAX-83 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.10 Comments on the SPR process 10 of 24 "Larry Kilgallen" 18 lines 2-JUN-1987 19:42 -< SOFTWARE self-maintenance >- ---------------------------------------------------------------- Regarding 663.9: Self-maintenance In addition to levels of hardware maintenance, there are levels of software maintenance from DEC. In order of decreasing cost: DECservice - someone from the local office comes out and does your installations Basic - you do installations, get to submit SPRs and call Colorado Self-maintenance - you do installations, get to submit SPRs and do NOT get to call Colorado Based on my experience with the local office and with Colorado at various sites, I think self-maintenance is the hands-down winner. Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 663.11 Comments on the SPR process 11 of 24 "Michael R. Pizolato" 6 lines 3-JUN-1987 14:31 -< Colorado is good for something! >- ---------------------------------------------------------------- RE: .10 I vote for Basic. That's apparently what we have. Even though Colorado is abominably slow to deal with problem submittals and SPR's, I have received a great deal of help over the phone. Michael R. Pizolato AT&T Technology Systems Dept. 323610 555 Union Blvd. Allentown, PA 18103 VAX-84 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 215/439-5500 ================================================================ Note 663.12 Comments on the SPR process 12 of 24 "Barry L. Wallis" 17 lines 3-JUN-1987 16:37 -< SPRs do have a place (sort of) >- ---------------------------------------------------------------- SPRs are next to useless! I am convinced they are mostly for DEC's benefit. At least when the DEC-20s were still alive their software dispatch had REAL answers (even patches most of the time) rather than "we will consider fixing this is a future release {if we feel like it}". Since I have moved into the VAX world I have all but stopped submitting SPRs (out of frustration). I am once again beginning to submit them. Why? Because they are a sure fire way of documenting when you told DEC about a problem. If you bring them to DECUS and can search out the appropriate developer you can usually get some kind of resolution soon after DECUS (if not right there on the spot). I guess getting a response every six months is better than getting a response every two years (and if you can document how much money was saved by the answer it can help justify your going to every symposium). Barry L. Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 VAX-85 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.13 Comments on the SPR process 13 of 24 "M. Erik Husby" 36 lines 4-JUN-1987 17:29 -< More comments on SPR's >- ---------------------------------------------------------------- My experiences with SPR's are the same as the others. I've had an SPR outstanding from September of last year despite the HIGH priority item being checked. Fortunately, I was able to develop a way around the problem. I've had an SPR take over a year to get a non-response. I've had a series of SPR's (3 submissions of the same problem) where it was obvious that the responder was not reading the SPR when writing the response. In that case I stumbled upon an undocumented "feature" that was the cause of the problem and was thus able to solve it. I had a response to an SPR that said what I wanted was undoable and then discovered in the next release of VMS that they had done what I requested. I would submit SPR's if it was easy and if I could be assured of getting a reasonable response in a reasonable time, generally 1 to 2 months. I think the problem stems from the bad reputation that program maintenance has. It is a task that is given to the beginners. Having done both maintenance and development, I can say that to do maintenance well requires as much if not more skill than development. I have mixed feelings about the TSC but feel that they have helped more often than not. DSIN seems to be a lot of fluff. FLAME OFF M. Erik Husby Project Software & Development 14 Story St. Cambridge, MA. 02138 (617)-661-1666 VAX-86 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.14 Comments on the SPR process 14 of 24 "Linwood Ferguson" 84 lines 7-JUN-1987 11:31 -< CSC yes, SPR's and DISN no >- ---------------------------------------------------------------- We support a lot of sites, mostly run by computer novices. Their talent for finding weird things is exceeded only by their insistence that it be fixed today! We've tried the route of local specialists, DISN, CSC (i.e. the telephone folks), SPR's, reading the manuals, and prayer (not necessarily in that order). Opinions, for what they're worth, follow: DISN is mostly worthless. I have spent hours logged on reading things, and probably twice found marginally useful info. SPR's submitted through them disappear. Problems to be researched seem to work, but appear to go to the same people you get on the phone, require many days to get a return call, and (considering the big advantage would be they could RESEARCH it before calling) you still have to re-explain it over the phone and wait while they do their thing. So... Also, I agree with .5 in that NO major bugs I've been aware of appear in DISN until far, far later than reasonable; and usually they do not appear at all. I have see lots of examples. Two recent ones: we had a problem with DYNSWITCH that went on about 6 months, they sent a patch, and for several months afterwards I never saw it in DISN (I haven't looked recently); next: LAVC has a major bug in the XQDRIVER that was crashing our LAVC almost daily. CSC provided a patch, it works, but it isn't in DISN! CSC, however, is not worthless to us. Quite the reverse. While we have had all the problems others have with lack of understanding, being treated as though you also don't know what a manual is, slow response, lost calls, and "beginner" support people, it is still the most valuable service Dec has offered us. We have to do telephone support to users, and it is THE hardest thing to do when dealing with computers- development, hardware and other glamour jobs notwithstanding. Considering that they place no requirements on the training of the people that call them, no limit on the number of calls, and help with ANYTHING VAX-87 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT (at least for awhile), I'm surprised that the people on the other end sound even slightly sane. Yet in spite of this, they are the ONLY source from which we have had help on serious problems. We have received several (about 6) patches for very serious problems (some never reported anywhere as near as I can tell!), quite a few immediate answers to serious problems from their database, and lots of "no reports of that" (which is in itself good to know). So consider this my vote for Basic Support; CSC isn't perfect, they maybe aren't even good, but they're the best thing going at this point. (Notice I omit local analysts: all they have ever done for us is call CSC; we have not had any that did anything productive except send BIG bills. I am sure that from discussions with others, this is not always the case and some are quite useful, but here you really "pays you money and takes your chances) Finally, SPR's: I have never, ever received a full answer to a problem submitted (I do not consider "will be fixed" an answer unless it says when! especially when they never are). Most are non-answers on form letters. Many recent ones (i.e. the last year) just plain disappear. So I too have virtually stopped submitting them. Sad but true that Dec has effectively cut its engineering off from this nearly-last communication with users. We use to deal with a competitor of Dec's (who's oriental name shall not be mentioned); they had NO means of communicating with engineering. Things widely known in the field NEVER got fixed (want an example: every system in the country failed on December 31, 1980 due to an error in handling leap years and counting days and file expirations. Result- massive complaints. Guess what: numerous releases later, on December 31, 1984, the same thing happened. We junked ours, not just because of 1988 being next year, but the presence of the same bugs (and their were LOTS more, like Cobol consistently not being able to ADD right always) for literally a decade shows they STOPPED LISTENING TO USERS). Suggestion: Decus encourage not only that Dec publish the standards it plans to meet for SPR response, but ALSO publishes on-going statistics that tell how its actually doing. In that kind of embarrassing spotlight, maybe they would take it seriously. (Actually, I liked the suggestion for a point system, but they would just make DISN keep score, and our records would follow our SPR's into the black hole). Linwood Ferguson VAX-88 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT MJ Systems, Inc. P.O. Box 5223 2564 Ivy Road (22901) Charlottesville, Va. 22905-0223 804/977-2732 ================================================================ Note 663.15 Comments on the SPR process 15 of 24 "Greg Isett" 9 lines 8-JUN-1987 00:03 -< DEC's Internal Database >- ---------------------------------------------------------------- One of the folks at CSC told me that DEC has an INTERNAL (to DEC) database of SPRs that are not accessible to DSIN users. Why not? Also, most of my responses to SPRs have been "It will be fixed in the next release..." My immediate questions (which I call DEC S/W Support) are "When will the next release be?" and "Why couldn't DEC send me a patch so I could use their expensive software right now?" Greg Isett hrb-singer incorporated department 125 p o box 60 state college pa 16804 ================================================================ Note 663.16 Comments on the SPR process 16 of 24 "M. Erik Husby" 14 lines 8-JUN-1987 09:56 -< An alternate SPR proposal >- ---------------------------------------------------------------- The RSX MultiTasker used to publish SPR's, perhaps if all VMS SPR submitters also forwarded a copy to the PAGESWAPPER, with a one or two line abstract, the PAGESWAPPER could publish the abstract along with the submitter's name and address. Then when one received Digital's response, one could notify the PAGESWAPPER and we could gather statistics. VAX-89 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT If Larry feels that this means more work for him, I will volunteer for the job, after all I suggested it. Any comments? M. Erik Husby Project Software & Development, Inc. 20 University Rd. Cambridge, MA. 02138 M. Erik Husby Project Software & Development 14 Story St. Cambridge, MA. 02138 (617)-661-1666 ================================================================ Note 663.17 Comments on the SPR process 17 of 24 "Barry L. Wallis" 39 lines 8-JUN-1987 19:36 -< DECs SPR System for the DEC-20s Used to Work >- ---------------------------------------------------------------- re: 663.15 Also, most of my responses to SPRs have been "It will be fixed in the next release..." My immediate questions (which I call DEC S/W Support) are "When will the next release be?" and "Why couldn't DEC send me a patch so I could use their expensive software right now?" We ex DEC-20 users used to have a wonderful service called "The Software Dispatch". Although it had the same name as the VAX/VMS Software Dispatch it was different (in that it was useful). It was split into different sections as follows: 1) A list of both system software and layered products including: o Name o Latest version o The date it was initially available 2) Solutions to problems found by DEC developers (i.e., a customer did not send in an SPR to generate the response). VAX-90 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 3) Solutions to published SPRs. This included a legible photocopy of the actual SPR the user sent in (and any solution he may have envisioned). 4) Unanswered SPRs. This is a photocopy of all SPRs received since the last issue (whether or not there is an answer to it yet). This is helpful because it includes the customers name and phone number so you can get in contact with them directly if you have a potential solution/workaround. They actual published answers (patches you could apply by hand if you chose) other than fixed in next release. Also the magazine came out on the 1st and the 15th of each month rather than every other month. It unfortunately seems as though DECs trendline is downward so far as SPR responses are concerned. I would certainly be willing to form an SPR working group within DECUS if that is an appropriate way to address the problem. Barry L. Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 ================================================================ Note 663.18 Comments on the SPR process 18 of 24 "Mark Hartman" 41 lines 9-JUN-1987 01:39 -< More steam... >- ---------------------------------------------------------------- I, too, must speak in defense of the CSC/CS... Our shop is a relative newcomer to VMS (migration/coexistence with RSTS/E). Much of our conversion could not have been accomplished without CSC. However... VAX-91 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT The RSTS Software Dispatch is published monthly. It clearly documents problems, and where and how to fix them (instead of just promising "a future release" [mostly]). It gets published every month. It's USEFUL! The VMS Dispatch is nearly useless. Nearly a third of the putatively useful pages in the last issue were taken up by a comparison of DEC products which are not bundled with VMS and in which I, for one, have no current interest. The article should have been a sales publication, not something we have to pay money for. Why, when VMS is supposed to be the flagship operating system, do we actually get better information on the PDP operating systems? Not that I'm objecting to good service SOMEWHERE, you understand, but this is supposed to be about VMS. It seems to me that a relatively short and simple program (perhaps even running on this same system [sorry, Larry, for volunteering your computer]) could handle collecting: o Submitter o One-line summary o Date submitted to DEC o Priority with which submitted o Date SPR ack'd o Date reply received o Quality of reply Something even this simple (and yes, I'll write it if necessary) could supply the entire VMS community with some raw information to beat DEC over the head with. Mark Hartman Jadtec Computer Group 546 W Katella Ave Orange CA 92667 VAX-92 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.19 Comments on the SPR process 19 of 24 "Bob Hassinger" 45 lines 10-JUN-1987 12:09 -< OK, what can we do about it? >- ---------------------------------------------------------------- OK, based on what I see here, at my LUG, and on DECUServe (which has had a similar discussion recently), it looks like there *is* a problem with SPRs and it seems to be specific to VMS. The question becomes what can we (DECUS) do about it? Part of these SPR problems usually turns out to be in the Maynard group that processes and tracks SPRs (i.e. "Box F, Maynard MA") and it looks like a group (CSC?) in Colorado has joined the act, filtering, reviewing and delaying things, and of course then finally we get to the VMS and layered product development groups. The first part of this path is common to all product areas but it seems that most of the current problems are in the parts of the process that are unique to the VAX and VMS related products. I would suggest that we concentrate our efforts here in the VAX SIG from the VAX and VMS point of view rather than trying to solve all the SPR problems for everyone. A little more focus will help I think. We talked at our LUG about this yesterday. Here are some ideas. 1) Send copies of your SPRs to the Pageswapper. This is a continuation of the tradition established many years ago by SIGs like RSX, 12 Bit and so on in efforts to focus attention on similar problems and to provide an alternate path to communicate problems and some of the solutions. 663.17 is a good suggestion that carries this idea further. 2) Put the issue on the SIR list. This will take quite a while but it is worth doing. 3) Get the VAX SIG to focus attention on the issue in an effort to get improvements. A session at the Fall Symposium to discuss the nature and source of the problem and possible solutions with DEC would get this process started. If I were funded to go to the Fall meeting I would submit the session. I can not do this, however, but if someone else is interested I will help them submit the session and so on. The deadline is very soon so you have to get on it right away. VAX-93 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT 4) We can all start writing to Ken Olsen. This can be a good way to get some attention to this sort of problem. Of course Ken O does not see most of this kind of mail himself but he does have people who are in charge of seeing to it that it gets the attention it requires. Bob Hassinger Liberty Mutual Research Center 71 Frankland Road Hopkinton, MA 01748 617-435-9061 ================================================================ Note 663.20 Comments on the SPR process 20 of 24 "Ken A L Coar" 7 lines 10-JUN-1987 16:54 -< Worse yet.. >- ---------------------------------------------------------------- We've started receiving an even more distressing response than 'will be fixed..' nowadays they tell us it 'will be CONSIDERED for fixing...' Needless to say, when the issue involved is an actual problem, not a suggestion, we're not terribly appreciative of the phrasing, particularly if they mean what they say! #k :-(} Ken A L Coar General Dynamics Office Systems 12101 Woodcrest Executive Drive Creve Coeur, MO 63141 (314) 851.4003 (CST) VAX-94 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.21 Comments on the SPR process 21 of 24 "Bill Mayhew" 6 lines 11-JUN-1987 23:22 -< *what* local office? >- ---------------------------------------------------------------- I'm glad to hear you *have* a local office to resort to. Three months of trying to find one where I am (a mere stone's throw from Maynard) has led to nothing but frustration. I am on the verge of tossing out all my software services because it seems like a total waste of money. Bill Mayhew Village Systems Workshop Inc PO Box 642 Natick MA 01760 617-237-0238 ================================================================ Note 663.22 Comments on the SPR process 22 of 24 "Bob Hassinger" 22 lines 12-JUN-1987 17:20 -< Ah, about "local offices" for software support... >- ---------------------------------------------------------------- My "local office" is in another state. It happens that our company has it's mega-IBM operation there and the last year or so DEC has been making a big push to establish a beach head there. Since our little research center has been the one and only DEC operation in the company for twenty years or so, someone at DEC thought it would be good to service our location from the same office (we had been making a good bit of noise about poor sales efforts out of our real local sales office). Anyway, only because of this special situation do I have someone I am supposed to be able to call at the moment (as noted it has not done me any good on this problem however - they never got back to me). When talking to DEC types about software problems they always seem to think everyone has a "local office" for software support issues but I have not had someone I could call since the early 70s. Maybe they are thinking of sites that have the high class software support - DECservice or whatever - but that is only a small fraction of the total. I have never been able to figure out what the rest of the world is suppose to do. I assume this is your problem too Bill, you do have a "local *sales* office". VAX-95 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Bob Hassinger Liberty Mutual Research Center 71 Frankland Road Hopkinton, MA 01748 617-435-9061 ================================================================ Note 663.23 Comments on the SPR process 23 of 24 "Bill Mayhew" 15 lines 13-JUN-1987 17:33 -< "Software Support": an oxymoron? >- ---------------------------------------------------------------- Yes, that's exactly my problem. I do have a "local sales office" .. which is its own problem (so what else is new), but that's another topic... From what I can tell, there is no such thing as a "local office" when it comes to software support, CSC's statements notwithstanding. The only interest of my "local office" seems to be in the administrative business of getting contracts renewed. And of course when you explain to them that the reason you haven't signed your renewal agreement is that your SPRs are disappearing into a black hole, that Colorado has no one qualified to use (never mind maintain) the product, and that Atlanta refuses to believe the product even exists ... they listen politely and then say "well, yes, but won't you renew the contract anyway?" Damn good thing it's Saturday and I'm in a good mood. {grin} Bill Mayhew Village Systems Workshop Inc PO Box 642 Natick MA 01760 617-237-0238 VAX-96 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 663.24 Comments on the SPR process 24 of 24 "Jim Littlefield" 7 lines 13-JUN-1987 20:06 -< Another comment... >- ---------------------------------------------------------------- I've had Colorado researching a problem that we encountered a couple of years ago. According to DSIN, it's still being researched! I gave up on DSIN, CSC, and SPR's a long time ago because the effort to document a problem versus the usual response just cannot be tolerated. I've got more important things to do they debug DEC's software. James Littlefield 170 Aquidneck Ave Middletown, RI 02840 (401) 849-8440 ================================================================ Note 664.1 F11BXQP Index File Extension Patch 1 of 1 "Kevin Angley" 17 lines 15-JUN-1987 17:09 -< F11BXQP Patch Bug - some further explanation >- ---------------------------------------------------------------- Don't get overly excited ... the "worse bug" that I spoke of was that the patch that I received was a very old F11BXQP (like a 4.2 version) that had the problem that append (copy) did not honor group privilege when appending. I call this "worse" because it was immediately obvious that it doesn't work (whereas the original problem was an accident waiting to happen). The bug that was reintroduced by the patch started occurring in V4.0 and V4.1 and was fixed by the mandatory update to 4.2. However, the fix didn't get checked into the mainline version of F11BXQP, so when it was rebuilt for V4.4, it got unfixed. I believe the fix in the mandatory update was Ec001 (LMP0331). If you have 4.5 F11BXQP patch from TSC, try to append to a file (without system privileges), relying on the group protection mask. VAX-97 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT TSC has since said that a patch to the patch is available. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 ================================================================ Note 665.0 PSI and DLMs with MAX WINDOW 7 No replies "Bob Tinkelman" 21 lines 3-JUN-1987 14:22 ---------------------------------------------------------------- I am having a problem with DLMs over VAX PSI. As this is really the first time I'm dealing with these, I'm probably making some obvious mistake (obvious to those who've done this before). Although I set DEFAULT WINDOW and MAX WINDOW to 7 for the X25-PROTOCOL DTE and also MAX WINDOW to 7 on the CIRCUIT, when I monitor the line, I see an X.25 call request packet without any "optional data" (or whatever that part of the packet is called which carries the negotiation information). The result is that I get stuck with the network default of a 2 packet window and very low throughput on a circuit with several satellite hops. I have the same problem trying to force a packet size of 256 rather than 128. Is there something special I need to tell DECnet or PSI to make it understand that the PPM will allow negotiation of these parameters at call-request time? Bob Tinkelman Cambridge Computer Associates, Inc. 56 Beaver Street, 3rd floor New York, NY 10004 1-212-425-5830 VAX-98 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 671.0 uVAX - Rainbow disk interchange No replies "MARTIN S WEINHOUS" 8 lines 19-JUN-1987 17:45 ---------------------------------------------------------------- Does anyone know of software that will permit the interchange of disks between the RX50s on a uVAX (running VMS) and a Rainbow (running MS-DOS). MARTIN S WEINHOUS 534 ALGONQUIN DR WARWICK, RI 02888 401 277-8311 ================================================================ Note 673.0 Tab problems with LN01 2 replies "Jamie Hanrahan" 24 lines 23-JUN-1987 15:18 ---------------------------------------------------------------- We have an LN01 on a DMF32 printer port. Device control library entries associated with form names DEFAULT and PORTRAIT switch the printer from landscape to portrait, and also specify non-default margins for both modes. Since I didn't want to bother computing tab positions (you have to do it in terms of 300ths of an inch) I just set the device to /NOTAB before starting the queue. At times, however, the print symbiont (or somebody) seems to ignore the /NOTAB characteristic of the device; tabs in the input files show up as tabs sent to the printer. I'm sure of this because the combination of non-default margins and default tab settings in the printer yields tab positions that fall BETWEEN character positions, and that's just where the characters show up! For some reason, printing something with /FORM=PORTRAIT and then going back to landscape clear things up; after that, the tabs in landscape mode work as they should. Anyone seen this? Anyone know what to do about it? I doubt it's the printer because DEC replaced the whole printer (due to mechanical problems) a while back; we had this problem with the old one too. Jamie Hanrahan VAX-99 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 673.1 Tab problems with LN01 1 of 2 "Saul Tannenbaum" 11 lines 25-JUN-1987 17:48 -< LN01 tab problems >- ---------------------------------------------------------------- I've seen it, but long, long ago. After making myself a bit crazy over it, my response was to make sure any text sent to the LN01 went thru runoff first (which expands tabs). Not a very useful general fix, I know. One thing I do recall is that DEC, at some point, thought the problem might be related to some DMF-32 problem that needed to be ECO'd. I found that interesting in that my LN01 wasn't hooked up through a DMF. You, however, might want to check your DMF ECO levels. Saul Tannenbaum Tufts University HNRC 711 Washington Str. Boston, MA 02111 (617)556-3346 ================================================================ Note 673.2 Tab problems with LN01 2 of 2 "Jack Patteeuw" 7 lines 26-JUN-1987 12:57 -< More on tabing >- ---------------------------------------------------------------- The print symbiont does not convert tabs to spaces. This is done by the driver itself (at least I'm certain of these when printing out through a terminal port). I seem to recall somewhere hearing that there was a firmware bug (one of many) in the DMF32 that could cause a problem like this so take the advice of 673.1 and have your ECO's checked. Jack Patteeuw Ford Motor Co. Electrical and Electronics Division 31630 Wyoming VAX-100 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Livonia, MI 48150 313-323-8643 ================================================================ Note 674.0 Duplicate directory file problem 2 replies "Bob Natale" 20 lines 24-JUN-1987 17:20 ---------------------------------------------------------------- Duplicate directory file-id problem We have two directory files with the same file-id: dua0:[000000]sys0.dir (10,1,0) dua0:[syse]sycommon.dir (10,1,0) (We also have a system logical, "sys$common" equivalent to dua0:[sys0.].) Our DEC h/w maintenance representative tells us that he thinks the DEC s/w representative sysgen'd a VAXcluster for us several years ago. We do not have a cluster, just a couple of DECnet'd 750s. We would really like to get rid of the second directory file, since it causes us some pain regarding backups and total disk utilization reports. Our DEC reps now say that they know how to prevent it in the future, but cannot recommend a way to fix it (short of some risky backup/restore operations). Any suggestions will be most welcome. Bob Natale Contel Business Networks 4330 East-West Hwy Bethesda MD 20814 VAX-101 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 674.1 Duplicate directory file problem 1 of 2 "Jamie Hanrahan" 58 lines 24-JUN-1987 18:51 -< It's part of standalone backup >- ---------------------------------------------------------------- DISCLAIMER: You probably already know some of the following. I'm going over it because lots of other folks don't; this issue and related ones come up again and again, so I'm covering as much as I can think of, all at once. That DEC h/w maintenance representative is guessing, and guessing wrongly. The [SYSE] directory is created when you (or someone) uses STABACKIT.COM to generate a standalone backup kit on your system disk. It uses the "aliased" directory entry for [SYS0] to avoid having to make duplicates, under [SYSE...], under files that are already present under [SYS0...]. Most VMS system disks have a few more of these: [000000]SYSMAINT.DIR is an alias for [SYS0]SYSMAINT.DIR, and if you have VAX-11 RSX installed, you'll find that [SYS0]001001.DIR is an alias for [SYS0]SYSLIB.DIR, etc. > ...it causes us some pain regarding backups and total disk > usage reports A backup save operation done in /IMAGE mode will work correctly. /IMAGE essentially goes by file headers, so the files only appear in the saveset once even though they appear in more than one directory; in a non-image backup, files in aliased directories will be duplicated in the saveset. This isn't really a problem until you do a full restore of the disk from the saveset -- the restore from the non-image saveset will yield duplicate copies of all the files in the aliased directories, while an image-mode restore from an image-mode saveset will do the right thing. There's no problem with selective restores from either type of saveset. As for disk usage... look into generating your reports from the "disk usage accounting file" that you can get from ANALYZE/DISK (which goes by file headers, and therefore contains one record per file, regardless of how many directories the file appears in), or from the DISKQUOTA utility, rather than from the output of the DIRECTORY command. VAX-102 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT (I know several system managers who have disk quotas turned on simply as a disk usage accounting tool: They give everyone huge quotas (or EXQUOTA privilege), and use the DISKQUOTA utility to get usage reports by UIC.) > We would really like to get rid of the second directory file $ SET FILE/REMOVE ddcu:[SYSE]SYSCOMMON.DIR;1 You probably know this already, but just in case...do NOT delete any of the files that appear in [SYSE.SYSCOMMON...] ! You will kill your system disk! > Our DEC reps now say that they know how to prevent it in the future. The only way to prevent it is by NOT putting a standalone backup kit on your system disk. Most of us need or want such a kit, though. Jamie Hanrahan Simpact Associates 9210 Sky Park Court San Diego, CA 92123 619-565-1865 ================================================================ Note 674.2 Duplicate directory file problem 2 of 2 "David Ambrose" 13 lines 25-JUN-1987 19:50 -< -< Duplicate directory file solution >- >- ---------------------------------------------------------------- The additional directory you have is part of the stand-alone backup. We had this problem for awhile too. Fortunately, it's easy to get rid of without graying your hairs. "SET FILE sycommon.dir/REMOVE" will remove the directory entry without deleting the file. We were able to delete everything else in the directory with no ill effects. We have two disks in our system and realised that there is no reason to have STAbackup on the system disk. However, we do keep it on all the non-system disks. If you have only one disk drive in your system, you might be well advised to retain the stand-alone backup on the system disk. David Ambrose VAX-103 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Bruning Computer Graphics (formerly Nicolet) 777 Arnold Drive Martinez, CA 94553 415/372-3439 ================================================================ Note 675.0 The fate of TECO 4 replies "JIM PALMER" 18 lines 26-JUN-1987 18:31 ---------------------------------------------------------------- AS of V4, the only compatibility mode image still shipped with the VMS kit is TECO. (It' actually a native/PDP hybrid). This presents a difficulty with the new VAX CPU's, particularly UVAX's because the RSX AME must be installed to accommodate calls to CEM$EMULATOR. I would be interested to know if TECO shall ever be re-written as a full 100% native mode utility. Any speculation? Jim Palmer American Dade Chemistry Systems 9500 Jeronimo Road Irvine, Ca. 92718-2107 (714) 458-3028 JIM PALMER 3 BROOKDALE IRVINE, CA. 92714-3338 (714) 458-3028 ================================================================ Note 675.1 The fate of TECO 1 of 4 "Larry Kilgallen" 35 lines 26-JUN-1987 21:22 -< The fates seem to be with TECO >- ---------------------------------------------------------------- Official support for TECO was announced as a bundled part of VMS effective with Version 4.0. Then another branch of DEC invented new CPUs without compatibility mode and the RSX AME (NOT bundled) hack became the only route to a working TECO on many machines. VAX-104 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT I asked about this at the Nashville "VMS Futures" session (when TECO support did not get mentioned in the official list of "futures"), and the answer was that a full native-mode TECO is planned for some future unspecified major version of VMS. (DEC was unwilling to mention the number by which even the next major version of VMS will be named, so the Pageswapper is considering running a pool in which people can guess the version number.) Of course, DEC does not guarantee to fix this particular CPU/software mismatch bug, and the "VMS Futures" session was surrounded by all sorts of caveats about "this is not a commitment", etc. There are two things in favor of this particular change making it into the next major version (by the way, I guess 5.0; everyone else can choose your own unique version number estimates in the aforementioned pool) of VMS. First, it is legally a bug fix to a previously committed feature (even though we all know that the underlying technical solution is a total replacement). Second, a native mode TECO is not an overwhelming technical challenge, such as clustering or checkpoint/restart, to name some successes and failures. The primary difficulty would seem to be getting management to agree to it, and by mentioning it in front of a couple of thousand people that certainly indicates agreement at least in principle to fixing the situation. Larry Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 ================================================================ Note 675.2 The fate of TECO 2 of 4 "Frank J. Nagy" 3 lines 27-JUN-1987 10:29 -< Future_Version == V5.0 >- ---------------------------------------------------------------- My rumor-mill-listener agrees with Larry that the "future version" mentioned for TECO is V5 (as a somewhat late Christmas present maybe?). Frank J. Nagy Fermilab PO Box 500 MS/220 Batavia, IL 60510 (312)840-4935 VAX-105 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT ================================================================ Note 675.3 The fate of TECO 3 of 4 "Kevin Angley" 3 lines 27-JUN-1987 11:45 < My pool entry: 6.9BFD-3.45*XLIV (oh, sorry .. that's ULTRIX) > ---------------------------------------------------------------- Hmmm ... rumours floating around that after VMS versions 1,2,3, and 4 .. the next may be 5. Gotta go make a phone call to Charlie Matco. Certainly this coup is worth a coffee cup. Kevin Angley 3301 Terminal Drive Raleigh, NC 27604 (919) 890-1416 * * VAX-106 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville VMS Notes from Nashville The following notes are from the VAXnotes conference entitled "VMS Q and A" run in the demonstration hall at the 1987 Spring US Chapter Symposium in Nashville. For the first time, those preregistered for the symposium were given individual accounts on the large demonstration cluster, allowing full use of VAXnotes including the NEXT UNSEEN capability to avoid previously read entries. ================================================================ Note 2.0 DCL questions 9 replies VAXFAM::ISMART 25 lines 27-APR-1987 14:38 ---------------------------------------------------------------- We are in the process of migrating from Prime to VAX/VMS computers (hurrah!). We keep assuring our doubting users that VMS is better than PRIMOS. But it is annoying when they come up with things that are easy on the Prime, that I don't know how to do on the VAX. Here are two common questions I get. Any suggestions would be most welcome. 1) Users write command files that start up third party applications, supply answers to some questions from the program, and then the command file hands over control to the user at the terminal. The application never knows that some of the input came from the command file and the rest from the terminal. VMS does not seem to allow a command file to redirect input in the middle of an application session. Is there any work-around? 2) PRIMOS has a very useful facility which allows a user to turn session logging (to a file) on and off. The only VMS equivalent seems to be to login, then SET HOST 0/LOG=filename. Not very convenient. Anything better? Thanks to anyone who replies to this. Ian Smart (username ISMART) Ford Motor Co. VAX-107 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 2.1 DCL questions 1 of 9 VAXFAM::LKILGALLEN 8 lines 27-APR-1987 15:11 -< DTM might simulate COMO files >- ---------------------------------------------------------------- 2) PRIMOS has a very useful facility which allows a user to turn session logging (to a file) on and off. The only VMS equivalent seems to be to login, then SET HOST 0/LOG=filename. Not very convenient. Anything better? In certain environments (program development) the DEC Test Manager will provide an analogous capability, but it is not so easy to use as the COMO files available on PRIMOS. ================================================================ Note 2.3 DCL questions 3 of 9 VAXFAM::MKIMURA 9 lines 27-APR-1987 15:39 -< Third Party Solution >- ---------------------------------------------------------------- While DEC does not have a convenient method of turning on session logging (to a file) besides the cumbersome SET HOST/LOG command; there are a few third party vendors which do. Precision Business Systems has a series of software called PHOTO/OBSERVER/ADVISOR which work quite well. PHOTO: Capture session to log file. OBSERVER: Observe (or spy if you will) another terminal ADVISOR: Advise (send characters to) another terminal VAX-108 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 2.4 DCL questions 4 of 9 VAXFAM::RMARSHALL 1 line 27-APR-1987 17:37 -< SESSION LOGGING >- ---------------------------------------------------------------- Try the PHOTO utility on the DECUS tapes... ================================================================ Note 2.7 DCL questions 7 of 9 VAXFAM::KCAROSSO "Kevin Carosso" 10 lines 29-APR-1987 16:26 -< Pseudo-terminal driver info (for PHOTOphiles) >- ---------------------------------------------------------------- I put the latest version of a VMS pseudo-terminal driver (originally written by Dale Moore et.al. at CMU) on this VAX SIG tape. This version fixes many major bugs. The driver finally seems to be quite stable. A pseudo-terminal provides the functionality required to write a simple PHOTO utility. I did not put a PHOTO on the submission since I'm not sure of CMU's policy with regard to distributing their PHOTO that uses the pseudo-terminal driver. (Permission to release the driver was granted some time ago). /Kevin ================================================================ Note 3.0 VIRTUALPAGECNT... Do I hear 1,000,000? 2 replies VAXFAM::JBENNETT 10 lines 27-APR-1987 14:49 ---------------------------------------------------------------- VIRTUALPAGECNT is limited to 300,000 pages under VMS 4.5. I notice that on this system (running Y4.6) the SYSGEN max is 600,000. I've got users that want that taken up to 1,000,000 pages. I know how to defeat the SYSGEN limit at boot time but I'm wondering what effects an accompanying large Page File Quota will have on the user and the system. Haven't tried it to date because I don't have an extra RA81 lying around for a huge secondary page file (but I will next week). Any thoughts? Anyone here I can talk too? VAX-109 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville - Jeff Bennett ================================================================ Note 3.1 VIRTUALPAGECNT... Do I hear 1,000,000? 1 of 2 VAXFAM::BMENGLER 4 lines 27-APR-1987 17:47 -< For what it is worth. >- ---------------------------------------------------------------- I have run with PGFLQUOTAS larger than the pagefile to make some software packages happy. As long as the software never actually tries to fault out that many pages, everything still works. A full page file hangs VMS. ================================================================ Note 3.2 VIRTUALPAGECNT... Do I hear 1,000,000? 2 of 2 STAR::CLABORN "Ernie, the sheep are " 8 lines 30-APR-1987 10:30 -< Watch your memory consumption >- ---------------------------------------------------------------- With VIRTUALPAGECNT that high, you might want to keep an eye on the amount of physical memory consumed in the system page table to map balance slots. You can get a pretty good estimate of real pages used by doing the following calculation: (VIRTUALPAGECNT * BALSETCNT) / (128 * 128) - George Claborn (VMS Performance) ================================================================ Note 4.0 Using setup files with laser printers 1 reply VAXFAM::ISMART 27 lines 27-APR-1987 14:50 ---------------------------------------------------------------- This one is a question and answer combined. If you are using HP laser printers, (or any other third party laser printers), you are probably using PRINT/SETUP files to configure the printer for different fonts/pitches/page-orientations for each print job. This causes a blank page to be ejected between each print job. VAX-110 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville The VAX manual does actually state that if your setup file contains any characters that the print symbiont does not recognize as DEC standard escape sequences, that it will do a form feed after the setup file. The work-around is to trick the symbiont into believing that the HP (or whatever) escape sequence is a valid DEC sequence by making it look like a VT240 escape sequence. All you have to do is stick at the beginning of the setup file, and at the end (I think is ASCII 154 or thereabouts. I can't remember . I also recollect that EDT refused to SPECINS ASCII chars >128, but EDT TPU would do it). The HP ignores the extra characters, but the print symbiont thinks it must be a standard DEC escape sequence, and presto, no form feed. Ian Smart Ford Motor Co. ================================================================ Note 5.0 VOLUME SHADOWING V1.2 12 replies VAXFAM::TAWINTER 5 lines 27-APR-1987 15:00 ---------------------------------------------------------------- I have just heard from my sales representative that I should not try to use volume shadowing under V1.1 but should wait for V1.2. Any comments from VMS Development? Also, is volume shadowing included in the VMS distribution kit and turned on with a key or is it on a separate distribution kit? VAX-111 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 5.1 VOLUME SHADOWING V1.2 1 of 12 VAXFAM::PCLAYTON 14 lines 27-APR-1987 15:44 -< Volume Shadowing V1.0,1.1 And Our Site >- ---------------------------------------------------------------- I have volume shadowing 1.0 up and running on my cluster and everything seems to working without problems. You must be at 4.5 or above, which can have other implications for 8xxx owners. The MAJOR problem I have with it is that in 1.0, only 4 sets are provided for per HSC and 2 members per set. That can cause problems with multi-HSC systems and automatic reboot. Version 1.1 has 7 sets per HSC so automatic reboots should be easier but the problem of balancing becomes the issue. The shadow software is built into the 4.4/4.5 kits and all you get is a key to turn it on. Note that it does not become active until the next reboot to enable the disk class driver. I have tried HSC failover and performance tests and it appears to work well. Even with SI83C disks. pclayton :-) ================================================================ Note 5.2 VOLUME SHADOWING V1.2 2 of 12 VAXFAM::RLYMAN 4 lines 28-APR-1987 09:18 -< Volume shadowing working (1.1) >- ---------------------------------------------------------------- I have version 1.1 of the shadow software running at our site, too. The only problem that we are having is trying to mount a shadowed volume set. We have received conflicting reports from DEC on whether this is supposed to work. VAX-112 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 5.3 VOLUME SHADOWING V1.2 3 of 12 VAXFAM::HCOHEN 35 lines 29-APR-1987 12:20 -< Update on Volume Shadowing Status >- ---------------------------------------------------------------- The answer in .1 is correct: 1. VAX Volume Shadowing V1.1 is stable and reliable. Digital knows of no reason that you should not be using this version of the product. 2. The VAX Volume Shadowing software is indeed distributed on the VMS kit and is enabled with the VAX Volume Shadowing key. The same key may be used with VMS V4.4 (in which case VAX Volume Shadowing V1.0 is running) and with VMS V4.5 (in which case you are running VAX Volume Shadowing V1.1). We anticipate that you will be able to use the same key with VMS V4.6 (which will then enable VAX Vol Shadowing V1.2). 3. The answer in .1 was also correct in stating that 7 shadow sets per HSC (or per pair of HSC's if the disks are dual- pathed) is supported in VAX Volume Shadowing V1.1. The old limit, with VAX Volume Shadowing V1.0 was 4 shadow sets per HSC or per pair of HSC's if the disks are dual-pathed. 4. Shadowing volumes in a volume set is supported. If you have had problems with shadowing a volume set, please file an SPR. Digital is not aware to date of problems with the shadowing of volume sets, but we will investigate. Additional information that can be supplied (either via replies to this note, or via an SPR) will be appreciated. (Note that each volume in a volume set counts against the limit of 7 shadow sets per HSC; thus, if you have a volume set comprised of 3 volumes, and you shadow each volume, then you may have a maximum of four additional shadow sets.) VAX-113 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Harriet Cohen VMS Product Manager for VAX Volume Shadowing ================================================================ Note 5.4 VOLUME SHADOWING V1.2 4 of 12 VAXFAM::DBOLTHOUSE 6 lines 29-APR-1987 13:54 -< Shadowing system disks >- ---------------------------------------------------------------- What about shadowing system disks??? I have heard this is not a good idea, but I would expect this to be one of the major uses of shadowing (i.e., redundant system disk in common system disk cluster)! David L. Bolthouse, Texas Instruments, Inc. ================================================================ Note 5.5 VOLUME SHADOWING V1.2 5 of 12 VAXFAM::APOTTER "Andrew Potter, Roc" 24 lines 29-APR-1987 14:14 -< Shadowing works fine on System disks >- ---------------------------------------------------------------- We shadow our cluster system disk with no problems. We enjoy a performance gain that allows us to use more and larger CPUs against a single system disk than would be otherwise recommended. By relocating as many write-intensive activities as possible off the system disk shadow set (PAGE+SWAP, JBCSYSQUE, SYSUAF, NETUAF...), you can gain even better performance on the common system disk. The only problem is that if you have a cold cluster shutdown, you will have to suffer through a required shadow copy operation when the first cluster system reboots. While this isn't really a problem, it slows down the system disk. This is because the shadow set is dissolved when the last system deassigns it and the primitive IO routines in the VMS BUGCHECK code must dump to the physical disk drive and hence only one member of the shadow set. Enforcing a copy operation on boot ensures you can see your crash dump. VAX-114 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville We avoid this problem by attempting never to take down all cluster members at the same time preventing the shadow set from dissolving. Another way to avoid this problem is to have at least one cluster member boot from a private system disk. Shut this system down last when you cold shutdown the cluster and boot it first on startup to allow you to do clean dismounts and mounts of the shadow set that forms the system disk for your remaining cluster members. ================================================================ Note 5.7 VOLUME SHADOWING V1.2 7 of 12 VAXFAM::KANGLEY 12 lines 29-APR-1987 14:28 -< No problems with cluster system disk shadow >- ---------------------------------------------------------------- We are using Volume Shadowing to shadow a cluster common system disk with no problems ... only "problem" was having to update all the console media. Also, documentation was particularly poor for Volume Shadowing product (it's even a funny size binder). One other thought ... our site doesn't do much during the wee hours of the night except a full disk backup which temporarily uses a dedicated scratch disk. I was thinking I could win a disk by dissolving the system disk shadow, using it as my scratch disk temporarily, then adding it back to the shadow set with a copy. Any thoughts on this? ================================================================ Note 5.8 VOLUME SHADOWING V1.2 8 of 12 STAR::STILES "When in doubt, hack." 31 lines 29-APR-1987 15:17 -< Who knows what command procs lurk in the system? >- ---------------------------------------------------------------- We are using Volume Shadowing to shadow a cluster common system disk with no problems ... only "problem" was having to update all the console media. Also, documentation was particularly poor for Volume Shadowing product (its even a funny size binder). Glad to hear that everything is working correctly. As for the documentation, what things were missing or incomplete, so we can improve it in the future? (I'm not sure I can do anything about VAX-115 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville the funny size binder, though... ;-) One other thought ... our site doesn't do much during the wee hours of the night except a full disk backup which temporarily uses a dedicated scratch disk. I was thinking I could win a disk by, dissolving the system disk shadow, use it as my scratch disk temporarily, then add it back to the shadow set with a copy. Any thoughts on this???? This should work just fine, but be aware of the windows of (mis)opportunity in such a plan -- first, be very sure not to dismount the physical member named to VMB, since this will prevent your ability to reboot in the event of catastrophe. Second, any node shutting down or crashing and then rebooting will attempt to recreate the shadow set as specified in your startup command procedure, which may very well destroy that which you just created... Also, if you have any batch or detached processes which occasionally go "looking" for disks to mount or re-mount, you should disable them (cluster-wide) first. - Mark Stiles VMS Development ================================================================ Note 5.9 VOLUME SHADOWING V1.2 9 of 12 VAXFAM::RLGRAHAM "Bob Graham, Dow Ch" 5 lines 29-APR-1987 15:27 -< breaking shadow sets to get scratch disks >- ---------------------------------------------------------------- In reference to 5.7, we regularly remove shadow set members during off hours to gain scratch disks for backups. So far we haven't had any problems with it, though we generally use volumes other than the system disk. Of course, there is some risk, if the remaining shadow set member should crash, that you will loose everything... . VAX-116 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 5.10 VOLUME SHADOWING V1.2 10 of 12 VAXFAM::PCLAYTON 32 lines 30-APR-1987 11:12 -< Question On Structure Integrity?? >- ---------------------------------------------------------------- Another item that came up during the installation and checkout of shadow sets is what happens when a shadow set membership is broken in terms of outstanding files that are open and processes that have buffers that are cached in processes throughout the cluster. My local DEC guru, under contract to us, searched through DEC internal people and came up with the following statement: "At the time of shadow set membership breakup due to a DISMOUNT, VMS searches through the process list for anyone that has open files on the shadow set and causes a process FLUSH of local/global buffers. This is to insure that the resulting dismounted volume is a 'complete' structure. This becomes critical when RMS ISAM files are open and the keys have been updated." Question To Harriet Cohen, Product Manager For Shadow Software: Is the above statement correct for current releases of the software and/or what constraints are placed on this area of concern?? What does the future hold that could cause changes to the above statement?? (understanding the problems of talking futures, comments are appreciated). Paul D. Clayton Manager Of Systems TSO Financial Corp 5 TSO Center 300 Welsh Rd. Horsham, PA. 19044 215/657-4000 Thanks in advance!! VAX-117 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 5.11 VOLUME SHADOWING V1.2 11 of 12 STAR::STILES "When in doubt, hack." 47 lines 30-APR-1987 15:49 -< The Shadow knows... >- ---------------------------------------------------------------- re .10: I will take Harriet off the hook for this one. There are two flavors of answer: If you DISMOUNT the shadow set virtual unit, (thereby "dissolving" the shadow set), the behavior is *exactly* equivalent to dismounting a non-shadowed disk volume. The outstanding I/O transactions cause further channel assignments and $OPEN calls to fail, and when all I/O in progress is completed and the transaction count on the volume goes to zero, the volume (or shadow set) is dismounted in toto, having had all buffers flushed, etc. For a shadow set, all (former) members which were just present within the shadow set are now identical and consistent individual representations of that shadow set. Re-MOUNTing the shadow set at this point will not incur any copy operations, since none are necessary. If you DISMOUNT a particular physical *member* of the shadow set, it appears to the shadowing software exactly the same as if that particular member had a hardware failure which disabled it -- no further I/O operations to that member are done, and the remaining shadow set is updated to be different should this member return or be added back into the shadow set later. If it is added back into the shadow set, it will be the target of a full copy operation to make it identical with the remaining shadow set. Think of this case as exactly the same as pulling the cables out of one of the drives. No further updates are done to that drive. This is the intended behavior, and is not subject to change. For applications which can tolerate a short interruption, a useful variation of volume shadowing is to shutdown the application, dismount the shadow set virtual unit, remount the shadow set minus one of the members, and restart the application (typical downtime less than 5 minutes). At this point the removed member can be used as an input volume for a backup, with the data in a known, consistent state. After the backup is complete, this member can be added back into the shadow set with a copy operation through MOUNT. VAX-118 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville For applications which cannot tolerate even that amount of downtime, a member can be removed from the shadow set by being dismounted in singular, and then used similarly to that just described, but the backup will be of a disk which may have inconsistencies due to incomplete writes or updates. Given some forms of database or journaling control, this may not be a problem. And of course, the shadow set itself can be used as the input for a backup (with BACKUP/IGNORE=INTERLOCK for those cases which require continued access during the backup). Given the read performance benefit of shadow sets, this may be the preferred method for some applications. - Mark Stiles VMS Development ================================================================ Note 5.12 VOLUME SHADOWING V1.2 12 of 12 VAXFAM::LKILGALLEN "Larry Kilgallen" 13 lines 30-APR-1987 17:19 -< 5.11 does not perform /RECORD >- ---------------------------------------------------------------- The problem with doing a backup to tape from a single shadow set member is that you must manually set the backup date on all files for the rest of the shadow set before allowing any access to the shadow set (if you use backup dates, which I presume is the norm). But at the time you are bringing the (n-1) shadow set members back on line, you do not know if the backup to tape will work or not (sorry, DEC hardware folks, but it's true). Therefore you must instead make a list (preferably using a computer) of the current files (including file id) and do the backup date setting only after the backup on the removed member is successful. That way you can avoid setting an incorrect backup date on files which were not at that spot on the disk when the removal was done. VAX-119 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 6.0 TU81 problems. No replies VAXFAM::CSMITHJR 24 lines 27-APR-1987 15:01 ---------------------------------------------------------------- Has anyone else encountered any problems with the TU81 or TU81-plus tape drive? We have a COBOL application that writes to tape. The program does its own end of reel processing, and works well on other tape drives, including the TU78, TU77, TE16, TA78, and TA81. When the program detects an end of volume condition, the program closes the file and requests that the user mount the next tape. When the next tape is mounted, the user replies to a prompt saying that the tape is mounted, and on all but the TU81, things go along as expected - data is written to the next reel, and so forth, and so on. On the TU81, however, the tape is mounted, and the program is told to continue. The program is doing SOMETHING...it continues to use CPU, and seems to think it's writing data; however, the tape does not move. At this point, we suspect there may be a device driver incompatibility involved. Has anyone else seen any similar problems? Thanks Charles T. Smith, Jr. National Computer Systems ================================================================ Note 7.0 Bug reports 4 replies VAXFAM::APOTTER 27 lines 27-APR-1987 15:01 ---------------------------------------------------------------- There is a DCL Bug that still exists in 4.6. This procedure has data in it (delineated by $deck and $eod) that is never executed which is my intent for this example. The problem is that the word "subroutine" in data is mis-interpreted by the label processing code and results in the statement "goto end" to fail. VAX-120 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville This was SPR'd and Colorado said that was the intent. We don't buy that. $ write sys$output "This should not error" $ goto end ! this statement fails $ create test.dat $ deck this is data that has the word subroutine in it $eod $ write sys$output "Should not see this" $end: $ write sys$output "it worked" ================================================================ Note 7.1 Bug reports 1 of 4 VAXFAM::APOTTER 32 lines 27-APR-1987 15:11 -< VMOUNT ignores DEVICE protections. >- ---------------------------------------------------------------- Here is another feature...(bug?) I want to protect a tape drive from user access. So I do the following: $ SET DEVICE/ACL=(ID=[*,*],ACCESS=NONE) MFA0: (command succeeds) I then log in to a normal (non privileged) user account and: $ ALLOCATE MFA0: (attempt to allocate protected device) %SYSTEM-F-NOPRIV, no privilege for attempted operation. (Device protection works fine for allocate) I then try mounting the device (without ALLOCATING it) VAX-121 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville $ MOUNT/FOR MFA0: %MOUNT-I-MOUNTED, mounted on MFA0: (The mount succeeds ignoring the Device ACL.) The problem is that VMOUNT ignores DEVICE protection when accessing a mountable device. This is NOT to be confused with VOLUME and FILE SYSTEM protections which are properly enforced. This problem was SPR'ed and Colorado called back with a non-answer, saying that this was the intent. Any Comments? _ Andrew Potter _ Rochester Institute of Technology ================================================================ Note 7.4 Bug reports 4 of 4 VAXFAM::VMSDEV 24 lines 30-APR-1987 17:36 -< RE: Mount ignores device ACLs - bug, fixed >- ---------------------------------------------------------------- RE: .1 $ SET DEVICE/ACL=(ID=[*,*],ACCESS=NONE) MFA0: (command succeeds) I then log in to a normal (non privileged) user account and: $ ALLOCATE MFA0: (attempt to allocate protected device) %SYSTEM-F-NOPRIV, no privilege for attempted operation. (Device protection works fine for allocate) I then try mounting the device (without ALLOCATING it) $ MOUNT/FOR MFA0: %MOUNT-I-MOUNTED, mounted on MFA0: (The mount succeeds ignoring the Device ACL.) This was a bug, and has been fixed (sorry, I don't recall the which version - I think it was fixed in 4.4). Sorry about any inconvenience this problem may have caused. VAX-122 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Hai Huang VMS Development ================================================================ Note 8.0 CI/local clusters 6 replies VAXFAM::JSANDERS 6 lines 27-APR-1987 15:38 ---------------------------------------------------------------- I currently have a VAXcluster with one VAX-11/780 and two VAX-11/750's on two HSC-50's. I have a big MicroVAX coming in May and want to use it as a member of the network team. Can I use the VAX-11/780 as both a CI cluster member and a boot node for a Local Area Cluster via the Ethernet ? ================================================================ Note 8.1 CI/local clusters 1 of 6 VAXFAM::USER 1 line 27-APR-1987 15:49 -< can't get there from here >- ---------------------------------------------------------------- No. You will not be able to do that until Version 5.0 of VMS. ================================================================ Note 8.2 CI/local clusters 2 of 6 VAXFAM::GKOHLS "Gary ap Kohls" 6 lines 27-APR-1987 16:06 -< not yet >- ---------------------------------------------------------------- Not until some as-yet-unannounced version of VMS (5.0). There are some sessions this week on what are called mixed-mode VAXclusters based upon both CI and NI (Ethernet). It was alluded to in the VMS Futures session and the slides will probably make it into the Pageswapper. VAX-123 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 8.3 CI/local clusters 3 of 6 VAXFAM::RCRITZ 4 lines 27-APR-1987 16:26 -< careful >- ---------------------------------------------------------------- Re .1 and .2: You are correct that it is being worked on. It is, as yet, unannounced and therefore is not committed for V5.0. ================================================================ Note 8.4 CI/local clusters 4 of 6 VAXFAM::DGARBER 6 lines 27-APR-1987 16:28 -< Mixed Clusters >- ---------------------------------------------------------------- That is what the discussion on "mixed clusters" is about. Currently you cannot boot a MicroVAX from an HSC50-based disk. This limitation will probably be removed in V5 of VMS, but there is no firm commitment from DEC to do this (still, it is very likely). David Garber - Northrop Research Center ================================================================ Note 8.5 CI/local clusters 5 of 6 VAXFAM::CSMITHJR "Charles T. Smith, " 4 lines 27-APR-1987 16:30 -< But it would be nice... >- ---------------------------------------------------------------- It would certainly be a useful thing to have. Boot devices are not really a big issue; having the ability to share the disk farm among all processors is. VAX-124 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 8.6 CI/local clusters 6 of 6 VAXFAM::TGARDNER 9 lines 28-APR-1987 10:17 -< RSM might suffice >- ---------------------------------------------------------------- True that you cannot mix NI-based and CI-based clusters, but you get some of the NI cluster functionality from the Remote System Manager (RSM) package. Chiefly, you get semi-automatic software updates for the MicroVAX plus the ability to share output print queues. (I'm not sure of the details of this, but I'm told it works). RSM is also fairly cheap. It's not a cluster, but it does help. T. ================================================================ Note 9.0 Auto-refresh of terminal on resume 3 replies VAXFAM::GKOHLS "Gary ap Kohls" 11 lines 27-APR-1987 16:02 ---------------------------------------------------------------- I would like to have my application programs repaint the screen when a user switches back to a terminal server session running my application program. I already have a refresh function but it requires the user to hit the refresh key. Is there any relatively simple method of having the program observe that the session has been resumed and trap through an AST (or some other procedure) to a subroutine which does the refresh? The only suggestions that have been made thus far involve modifications to TTDRIVER or LTDRIVER which is not really my cup of tea... . In advance - thanks for your comments. VAX-125 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 9.1 Auto-refresh of terminal on resume 1 of 3 STAR::KENNEY 28 lines 29-APR-1987 15:30 -< One possible work-around >- ---------------------------------------------------------------- I am afraid that, short of patching the LAT driver or the terminal server software, no direct way exists to do what you want. You can, if you have a VT2xx or VT3xx series terminal, use a User Defined Key (UDK) to do this. 1) When a person logs into the LAT have them define a standard switch character. 2) Have your application load the UDK with the switch character and the refresh character. 3) Now when users want to switch sessions they hit (for instance) shift F6 and it sends the switch character and a CTRL/W. There are no changes that can be made to the terminal driver to help. It has no knowledge of a session being switched. This knowledge is private to the terminal server and the LAT driver. Forrest Kenney VMS Development PS. To define F6 to do what you want using CTRL-\ as the switch character, send this escape sequence to the terminal when your application starts UP: ESCP1;1|17/1C17ESC\ PPS. VT1xx series terminals will ignore this escape sequence. VAX-126 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 9.2 Auto-refresh of terminal on resume 2 of 3 N8700::TLASKO 3 lines 30-APR-1987 14:39 -< another option >- ---------------------------------------------------------------- This is just a note that you don't have to do this with a VT3xx series terminal because it provides complete backing store and state for two separate sessions.... ================================================================ Note 9.3 Auto-refresh of terminal on resume 3 of 3 STAR::KENNEY 7 lines 30-APR-1987 15:32 -< VT330/340 maybe, maybe not >- ---------------------------------------------------------------- I don't mean to argue with Tim but the refresh of multiple sessions depends upon a number of things. It may or may not provide what you want. The bottom line is that there is no guaranteed way to make this work every time under all possible configurations. Forrest Kenney VMS Development ================================================================ Note 11.0 Reverse chronological directories 4 replies VAXFAM::DANIELSMITH "DANIELSMITH" 8 lines 27-APR-1987 16:10 ---------------------------------------------------------------- Is there a way to produce a directory listing sorted in any other order than alphabetically by name? For example, reverse chronological? Or... Is there some profound philosophical reason I don't yet understand why this is not an obvious good idea? (RT-11 has had it since 1979 or so... .) VAX-127 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 11.1 Reverse chronological directories 1 of 4 VAXFAM::SHARROD 11 lines 28-APR-1987 10:46 -< Sounds good to me too >- ---------------------------------------------------------------- It seems like a good idea to me too. As noted in note 11.0, the RT-11 DIRECTORY command has all sorts of sorting options. So why not VMS too? There is an option on the TOPS-10 DIRECTORY command that produces a file that can be sorted using the sort utility. A similar option for VMS would be suitable. Scott Harrod Kildeer, Illinois 60047 ================================================================ Note 11.2 Reverse chronological directories 2 of 4 VAXFAM::DGORDON "Doug Gordon - Vande" 7 lines 28-APR-1987 12:20 -< DIRECTORY/OUT=filespec >- ---------------------------------------------------------------- If all you want is a file that can be sorted, use DIRECTORY/NOHEADER/NOTRAILER/OUTPUT=filespec and run that through SORT. Most of what you want can be done in a DCL command procedure, although it's liable to be slow. --Doug Gordon DEC Internal Software Services VAX-128 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 11.3 Reverse chronological directories 3 of 4 VAXFAM::CRHODE "Chris Rhode (Lockhee" 9 lines 28-APR-1987 13:56 -< More on DIRECTORY/OUT=filespec >- ---------------------------------------------------------------- In particular, DIRECTORY/NOHEADER... displays dates in a format unsuitable for sorting (e.g. sort in reverse chronological order by modify date). You can parse up the resulting output (ick!) and use the F$CVTIME lexical function to convert the dates into a sortable format (1986-01-01 instead of 01-JAN-1986, some such...)...yes, it's slow. I suppose you could use F$FILE_ATTRIBUTES instead of parsing up output from DIRECTORY/NOHEADER... -Chris ================================================================ Note 11.4 Reverse chronological directories 4 of 4 VAXFAM::LFRESINSKI 4 lines 29-APR-1987 16:04 -< Sorting doesn't cut it... >- ---------------------------------------------------------------- These options are a lot of overhead. TOPS-20 offered reverse chronological and several people are trying to reinvent the wheel. It seems like DEC should provide DIRECTORY/SELECT=REVERSE or some such which would be much faster. ================================================================ Note 12.0 Accounting/Physical ports 5 replies VAXFAM::DBOLTHOUSE 12 lines 27-APR-1987 16:16 ---------------------------------------------------------------- Can anyone tell me if ACCOUNTING will record the physical port for a login session (in the case of LT or VT devices) in version 4.6 (or a future, unspecified release...which may or may not occur...). It seems it should, now that the $QIO hooks are available to get the information easily... VAX-129 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville This is useful in the case of having expensive output devices for which we may want to charge connect time. It's tough to get that info now. David L. Bolthouse Texas Instruments ================================================================ Note 12.3 Accounting/Physical ports 3 of 5 VAXFAM::DTHURY "Denny Thury, Texas " 11 lines 28-APR-1987 11:08 -< Which physical terminal do you want? >- ---------------------------------------------------------------- Which physical port do you want?? a. The one first logged in on? b. The one when logging off? c. or something in the middle? Remember that with Virtual Terminals, the user can change the physical port, yet maintain the same virtual port! So if you were planning on accounting, and I was bent on "cheating", I'd log on to a "cheap" terminal, disconnect, then log back on the the "expensive" terminal! ================================================================ Note 12.4 Accounting/Physical ports 4 of 5 VAXFAM::GASCOTT "Greg Scott" 8 lines 29-APR-1987 13:21 -< a session record for each session >- ---------------------------------------------------------------- It should write a session record for each terminal session. Suppose a user logs in on TXB0, gets detached from carrier loss, gets back on to TXA5. Three session records should be written (one for the TXB0 time, one for the "detached" time, and one for the TXA5 time). That way you can do whatever accounting you want, based on whatever physical resources the user has consumed. VAX-130 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 13.0 Need better cluster support! 13 replies VAXFAM::AHUNT 21 lines 27-APR-1987 16:39 ---------------------------------------------------------------- When is DEC planning to add support for commands such as SHOW SYSTEM/NODE=node SHOW USER/NODE=node This has become a very sore point with our users. I find logging into another machine to look at the status of a job to be annoying. Since we have a machine in a cluster dedicated to BATCH (a nice feature of clusters) and interactive use turned off, the users really find this to be a problem. I am aware of ways to work around the problem using DECnet etc. However, I feel this would be a convenient but (Developers am I wrong?) somewhat easy feature to implement. My understanding was that the cluster software had the hooks to add such features. What do others think? Alan Hunt Ford Motor Co. ================================================================ Note 13.1 Need better cluster support! 1 of 13 VAXFAM::LKILGALLEN "Larry Kilgallen" 4 lines 27-APR-1987 17:25 -< Sure, it would be nice... >- ---------------------------------------------------------------- I agree that it would be convenient, but disagree that it would be easy to implement. DEC has access to SCS for a transport mechanism, but that just makes it possible, not easy. VAX-131 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 13.2 Need better cluster support! 2 of 13 VAXFAM::RMARSHALL 4 lines 27-APR-1987 17:49 -< Remote "SHOW SYSTEM" >- ---------------------------------------------------------------- The FINGER utility on the DECUS tapes does what you ask. Of course, it uses DECnet and it must be installed on all systems that you wish to "FINGER", but it beats the hell out of logging in to another node. ================================================================ Note 13.3 Need better cluster support! 3 of 13 VAXFAM::BMENGLER 1 line 27-APR-1987 17:56 -< a way to roll your own. >- ---------------------------------------------------------------- Implementing a remote DECnet task is an easy way to accomplish this. ================================================================ Note 13.4 Need better cluster support! 4 of 13 N8700::SSIMM 19 lines 28-APR-1987 09:51 -< TELL.COM >- ---------------------------------------------------------------- A DCL command procedure called TELL.COM has been floating around for a while. It dates back to Tim Halverson (of the Maynard crew). TELL.COM allows any DCL command (that doesn't do screen formatted I/O) to be executed remotely. Commands that TELL.COM works with include SHOW SYSTEM, SHOW NETWORK, DIRECTORY, etc. TELL has two modes -- interactive (as if you had SET HOST) and a one-shot command mode. The procedure is rather on the complex side. TELL.COM is patterned after the NCP TELL command. (I realize this is NOT what the original note was looking for...) VAX-132 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Stephen Simm General Electric Consulting 518-458-2409 ================================================================ Note 13.7 Need better cluster support! 7 of 13 VAXFAM::AHUNT 11 lines 28-APR-1987 10:55 -< A little more info! >- ---------------------------------------------------------------- When I wrote the note I was aware of several work-arounds most using DECnet. However, our feeling is that DECnet is putting lots of overhead into a process that is simple and frequent. Having a job login and swapping in a process is a lot of work just to check on the cpu runtime of a batch job. I am also aware of "permanent netservers" to reduce the overhead but you still pay the DECnet overhead. DEC was able to implement the SYS$BRKTHRU system service for communication purposes so it would seem that the ground work for communications between systems through the cluster has been provided. ================================================================ Note 13.8 Need better cluster support! 8 of 13 VAXFAM::AHUNT 4 lines 28-APR-1987 10:58 -< Some additional tidbits! >- ---------------------------------------------------------------- I should note that this is a large cluster (8800, 8700, 2 8600's, a lonely 780) with several hundred users and 100-200 concurrent users. Having them pounding DECnet does not seem like a logical solution. VAX-133 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 13.9 Need better cluster support! 9 of 13 VAXFAM::JHANRAHAN 4 lines 28-APR-1987 14:04 -< DECnet should be okay >- ---------------------------------------------------------------- There is widespread disinformation around about the amount of `overhead' required to do things via DECnet. I suggest you try it and measure it before assuming it'll be `too slow'. ================================================================ Note 13.10 Need better cluster support! 10 of 13 VAXFAM::WBAKER 20 lines 29-APR-1987 11:17 ---------------------------------------------------------------- It's usually not DECnet that is slow, but rather a symptom of how your logins are setup to allow NETWORK logins. The key to this seems to be making sure you bypass all your INTERACTIVE login and possibly you BATCH login stuff and just let DECnet into the system without all the setting of symbols etc. I.e., check the mode of the process, if it's NETWORK then EXIT. This will help greatly, because if you've set up proxies on your user accounts between nodes, then those people will go through their normal login procedure which includes all the stuff they put in as well as what the SYSTEM MANAGER adds in the way of initializing the process. If they specify that they want to login via the default nonprivileged DECnet account, this should not be a problem. Even with a proxy in place, they can login to the default nonprivileged DECnet account by indicating the node with double quotes where the username password normally is used. (ie: type node""::0=tell ) Once you have the login problem solved, if the linkup between nodes is still slow, then you possibly need to try to tune your network for the general use by its users. You know best on that. VAX-134 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 13.11 Need better cluster support! 11 of 13 VAXFAM::RLGRAHAM "Bob Graham, Dow C" 11 lines 29-APR-1987 11:26 -< doing it over DECnet >- ---------------------------------------------------------------- As an addition to 13.10, one tactic that can be helpful is to force your "TELL" command to use the default DECnet account by specifying the null string for the access control. This increases the chance of a NETSERVER process being around when the user gives the command. In addition, you can define the logical name NETSERVER$TIMEOUT to be the time that NETSERVER processes hang around. If you do this as a process logical for the default DECnet account, then it will only affect those processes. On our cluster, we've found that 2 hours is a good time. There is almost always a mail message or cluster SHOW command done by someone in that time period. ================================================================ Note 13.12 Need better cluster support! 12 of 13 VAXFAM::JHANRAHAN 7 lines 29-APR-1987 13:20 -< Our NETSERVERs don't always get reused. >- ---------------------------------------------------------------- That reminds me of something... on our small network (8200, 750, MicroVAX II, Ethernet connections) the NETSERVERs that hang around are not reused if the next inbound connect follows too quickly on the heels of the last disconnect. For instance, if I TYPE node::file1, TYPE node::file2, TYPE node::file3, then do a SET HOST to node:: and do a SHOW SYSTEM, I'll find I've created three NETSERVERs. This seems excessive. What's going on? VAX-135 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 13.13 Need better cluster support! 13 of 13 VAXFAM::PBECK 9 lines 29-APR-1987 13:30 -< Non-zero loop execution >- ---------------------------------------------------------------- NETSERVER invokes the requested DECnet object by doing LIB$DO_COMMAND or LIB$RUN_PROGRAM, which exits NETSERVER.EXE. There is a loop inside NETSERVER.COM that brings it back into the EXE when the object has exited; only after NETSERVER.EXE is invoked again will NETACP pass another link to this server (since it is NETSERVER.EXE that issues the QIO to NETACP to request the link). This takes non-zero time, and if another link shows up in that interval, it creates a new process. ================================================================ Note 16.0 HSC50 format retry failures 2 replies VAXFAM::DGARBER 4 lines 27-APR-1987 16:48 ---------------------------------------------------------------- We recently upgraded from version 2.50 to version 3.00 on HSC50 micro-code (TU58's). Since the upgrade, initializing "virgin" tapes (on drives that are on the HSC50) produces numerous "formatter retry failures" on the HSC50 console. Is this normal, i.e., a feature? David Garber - Northrop Research ================================================================ Note 16.1 HSC50 format retry failures 1 of 2 VAXFAM::PCLAYTON 19 lines 27-APR-1987 17:15 -< HSC Format Errors On HSC50 And 70 >- ---------------------------------------------------------------- We had the same problem. The error occurs only during the writing of what appears to be the tape labels. I have had numerous conversations with our local office and they finally came up with a patch to the 3.00 HSC code. You have to tell the local office to cough it up and install it. VAX-136 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Note that this is for people who are using the HSC backup to make the tapes NOT VMS backup. A similar problem also appears when you have a tape that was used in doing an HSC backup and then try to use the tape to do 'normal' VMS work. You have to override the 'access,owner' on the mount. Paul D. Clayton TSO Financial Corp Manager Of Systems :-) ================================================================ Note 16.2 HSC50 format retry failures 2 of 2 VAXFAM::YCHEUNG "Yin Cheung" 3 lines 28-APR-1987 09:45 -< Get the patch >- ---------------------------------------------------------------- Yes, the patch to V3.00 should help. In addition to easing the formatter retry problem, this patch supposedly also supports "reverse retry." ================================================================ Note 17.0 DUAL SESSIONS ON VT340 TERMINALS 3 replies VAXFAM::AHAMPTON 3 lines 27-APR-1987 16:57 ---------------------------------------------------------------- How do you create dual sessions on the new VT340 terminals? I checked HELP and the latest "NewFeatures" topic included information on version 4.4 instead of 4.6. VAX-137 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 17.1 DUAL SESSIONS ON VT340 TERMINALS 1 of 3 VAXFAM::GISETT 3 lines 28-APR-1987 12:36 -< MUST USE SSU UTILITY >- ---------------------------------------------------------------- You have to use the SSU utility. I saw a demo of this in the TERMINAL PRODUCTS area. I tried it on the user terminals, but unfortunately, it appears that a terminal has to be set "disconnectable" for this to work. ================================================================ Note 17.2 DUAL SESSIONS ON VT340 TERMINALS 2 of 3 VAXFAM::GISETT 4 lines 29-APR-1987 14:34 -< DEPENDS UPON WHICH COMPUTER YOU USE >- ---------------------------------------------------------------- It appears that the ability to use the session utility depends upon which member of the VAXFAM clus (here at the Symposium) that you get connected to. To determine whether your machine supports SSU, enter SHO DEVICE TD*. If TD* devices are there, then your cluster member will support SSU. ================================================================ Note 17.3 DUAL SESSIONS ON VT340 TERMINALS 3 of 3 N8700::TLASKO 10 lines 30-APR-1987 14:42 -< more information on how to set it up >- ---------------------------------------------------------------- Once you determine whether it is there (as .2 describes), then hop into Set-Up mode, select the Global Set-Up screen and make sure that: Dual Terminal is "enabled" and Terminal Comm Ports is set to "Sessions on Comm1" Get out of set-up and type the command "$ ssu enable". The terminal should flash a DONE message if this is successful. VAX-138 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Tim TBU Engineering ================================================================ Note 20.0 Project Accounting in VMS 5.0? 8 replies VAXFAM::TFRERICKS 1 line 27-APR-1987 17:36 ---------------------------------------------------------------- Anybody heard anything about project accounting in VMS 5.0? ================================================================ Note 20.1 Project Accounting in VMS 5.0? 1 of 8 VAXFAM::DTHURY "Denny Thury, Texas I" 6 lines 28-APR-1987 11:15 -< Not according to VMS developers >- ---------------------------------------------------------------- This question was raised at the VMS FUTURES session Monday. The speaker (the VMS Project Manager) stated that NO action was taken on project accounting for next major release of VMS! ================================================================ Note 20.2 Project Accounting in VMS 5.0? 2 of 8 VAXFAM::MSHANNON 5 lines 28-APR-1987 12:30 -< Project Accounting with PAAS >- ---------------------------------------------------------------- What about PAAS? (The Project Accounting system of subsort by Battelle?) It looks like a pretty useful system, but I'd like to know if anyone else has had experience with it. --Marc ================================================================ Note 20.4 Project Accounting in VMS 5.0? 4 of 8 VAXFAM::SKALLICK 8 lines 29-APR-1987 14:03 -< another possibility >- ---------------------------------------------------------------- Experience has taught me that PACS is a very good project accounting package, as long as you don't have to do anything fancy. It is simple and easy to learn. Consequently, however, there is not an abundance of functionality. Quantum RS (by CIS) is also very good, and much more robust. However, Quantum is also cumbersome, complex, and it tends to start making VAX-139 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville assumptions about what you meant to do, etc. I would recommend looking at PACS first, but if you need something more capable, get Quantum (and a new person to support it). ================================================================ Note 21.0 Enhanced PRINT/REMOTE in VMS 5.0? 9 replies VAXFAM::TFRERICKS 2 lines 27-APR-1987 17:37 ---------------------------------------------------------------- Anybody heard anything about enhancements to PRINT/REMOTE in VMS 5.0? ================================================================ Note 21.1 Enhanced PRINT/REMOTE in VMS 5.0? 1 of 9 VAXFAM::APOTTER "Andrew Potter, Roch" 7 lines 28-APR-1987 09:26 -< User Modified Symbiont >- ---------------------------------------------------------------- No news about a better print/remote, but Kevin Carosso has a very nice user-modified print symbiont that will probably do what you want. I think it is better than the one that came on the latest SIG tape (San Francisco) because it doesn't require privileges on the remote printing node. I think he is here at DECUS. ================================================================ Note 21.2 Enhanced PRINT/REMOTE in VMS 5.0? 2 of 9 VAXFAM::GEVERHART 3 lines 29-APR-1987 15:40 -< Coming on VAX Spring '87 Tape >- ---------------------------------------------------------------- John Osudar has submitted a remote print symbiont (actually two of them) to the Spring '87 VAX tape. Be on the lookout for them. VAX-140 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 21.3 Enhanced PRINT/REMOTE in VMS 5.0? 3 of 9 VAXFAM::KCAROSSO "Kevin Carosso" 13 lines 29-APR-1987 16:20 -< How to get NETSMB >- ---------------------------------------------------------------- Our (not "my", since I started it but someone else got stuck with finishing it!) NETSMB works quite nicely - certainly does the job for us. I'm submitting it to this VAX SIG tape. If you'd like a copy earlier and you have some sort of Internet access, you can reach Chris Yoder, who inherited the code, at: chris@engvax.scg.hac.com (if your site does domains right) or: chris%engvax@oberon.usc.edu (if your site doesn't) or: scgvaxd!engvax!chris (if only UUCP paths work) My address is as above, with username "kvc". /Kevin ================================================================ Note 21.4 Enhanced PRINT/REMOTE in VMS 5.0? 4 of 9 VAXFAM::VMSDEV 5 lines 29-APR-1987 18:25 -< Not in the next major release >- ---------------------------------------------------------------- There are no enhancements to PRINT/REMOTE planned for the next major release of VMS. However, Digital is investigating ways to provide a more general solution to the remote printing problem in the future. Jim Krycka, VMS Development VAX-141 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 21.6 Enhanced PRINT/REMOTE in VMS 5.0? 6 of 9 N8700::TFRERICKS 4 lines 30-APR-1987 12:52 -< Another ONE >- ---------------------------------------------------------------- We are looking at MAGIC ONE by Lee Leahy (?). Just to add to the list of places to fill this need. Troy Frericks (515)294-1420 ================================================================ Note 22.0 DISKQUOTA Errors 13 replies VAXFAM::TFRERICKS 9 lines 27-APR-1987 17:42 ---------------------------------------------------------------- There are still disk quota errors in version 4.5 of VMS. User disk quotas are reported incorrectly in many cases. Both high and low. ANALYZE shows this happens every day. Also, I change disk quotas and they seem to go back to the original values. This happens about once a week. I have heard complaints about this before, thought it would be taken care of by now. Anyone know anything about it? ================================================================ Note 22.1 DISKQUOTA Errors 1 of 13 VAXFAM::CSMITHJR "Charles T. Smith, " 5 lines 27-APR-1987 17:44 -< One way >- ---------------------------------------------------------------- One way to clean things up is to run anal/disk/repair/confirm on the disk every week or two. Not only does this clean up quota problems, but lost files and many other problems out on the disk. VAX-142 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 22.6 DISKQUOTA Errors 6 of 13 VAXFAM::JPRECIADO 8 lines 28-APR-1987 11:27 -< DEC Knows already >- ---------------------------------------------------------------- From a user who has experienced this problem and talked to the appropriate DEC developer I can say that DEC is aware of the problem and regards it as a high priority to fix. It's not an easy ball of wax to crack. DEC has told me that when they fix the problem they will NOT wait for a VMS release but will distribute it as a patch, image, etc., and retrofit where necessary to solve it. Be patient (I know it's difficult). They know it's something that they must solve soon. ================================================================ Note 22.11 DISKQUOTA Errors 11 of 13 VAXFAM::CRHODE "Chris Rhode (Lockhe" 17 lines 29-APR-1987 13:58 -< Re: How do I fix it? >- ---------------------------------------------------------------- Under limited experience with ANALYZE/DISK/REPAIR, I've found that I have to do a REBUILD command inside DISKQUOTA, -then- run ANALYZE/DISK/REPAIR. If you skip the DISKQUOTA REBUILD, ANALYZE/DISK/REPAIR asks some cryptic questions about what to do with the case where the quota file doesn't match what ANALYZE/DISK finds, and the repair in that case generally does not do what you want. I always privately allocate/mount disks I am doing this on so that the two can't get out of sync between commands/while commands are being run. Also you should do ANALYZE/DISK before doing ANALYZE/DISK/REPAIR to find problems with directory files (corrupted directory files, FOO.DIR;2 etc.) and fix those, because they seem to prevent ANALYZE/DISK/REPAIR from completing otherwise. If you still can't get the disk quotas to line up with actual usage after trying this, all I can suggest is that you SPR the problem. From what I've heard, ANALYZE/DISK/REPAIR is the final solution to the problem. VAX-143 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 23.0 DCL SUBROUTINE bug?!?!?! 6 replies VAXFAM::USER 18 lines 27-APR-1987 17:48 ---------------------------------------------------------------- I have found a problem with DCL SUBROUTINEs in VMS V4.4: The only way to get SUBROUTINEs to work consistently is to put them at the beginning of the DCL command procedure (i.e., before any statements that refer to them). If you do not do this, sometimes the the subroutines will execute correctly and sometimes they won't (the probably is consistent in that if it does(n't) work it will continue to (not) work until you edit the file and change the code. I cannot find this documented anywhere and it worries me. Any comments? Barry Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 ================================================================ Note 23.1 DCL SUBROUTINE bug?!?!?! 1 of 6 N8700::SSIMM 9 lines 28-APR-1987 09:55 -< It is in there somewhere >- ---------------------------------------------------------------- I ran into the same requirement -- strange things happen otherwise. It IS listed as a requirement in the manuals. I'm not sure where I saw the note -- possibly the release notes? Stephen Simm General Electric Consulting 518-458-2409 VAX-144 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 23.2 DCL SUBROUTINE bug?!?!?! 2 of 6 VAXFAM::HSPAIN 6 lines 28-APR-1987 10:48 -< Tell us it's not true... >- ---------------------------------------------------------------- I didn't know of the positional requirement for subroutines in DCL procedures. Do you mean that I will need to move all subroutines to the top of the file? (what about the second subroutine?) Harrison Spain McDonnell Douglas ================================================================ Note 23.3 DCL SUBROUTINE bug?!?!?! 3 of 6 VAXFAM::BMULLALLEY "Barry Wallis" 10 lines 29-APR-1987 13:52 -< It is true... >- ---------------------------------------------------------------- The only reliable way I have found to use DCL subroutines is to put the non-subroutine portion after all the subroutines. DCL does work as documented in that it will fall through the subroutines without executing them. Barry L. Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 VAX-145 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 23.4 DCL SUBROUTINE bug?!?!?! 4 of 6 VAXFAM::SSIMM 10 lines 29-APR-1987 15:02 -< And It Is Enforced, Too... >- ---------------------------------------------------------------- If I don't have every one of those pesky subroutines at the top I've gotten many weird errors -- one of which blew me entirely off the system. (It came back with an oddball error message followed by the USERNAME: prompt.) Steve Simm GE Consulting Services 518/458-2409 ================================================================ Note 23.5 DCL SUBROUTINE bug?!?!?! 5 of 6 STAR::SCHULTZ "Harold W. Schultz - " 10 lines 30-APR-1987 08:02 -< This problem will be fixed >- ---------------------------------------------------------------- The intent of subroutines was to allow them to be defined anywhere within the command procedure. However, a problem was recently discovered when the subroutine being called is defined later in the command procedure. The current workaround is to define all subroutine procedures at the beginning of the file. This problem will be fixed as soon as possible to allow subroutines to be defined anywhere in the procedure. Harold Schultz VMS Development VAX-146 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 23.6 DCL SUBROUTINE bug?!?!?! 6 of 6 VAXFAM::LDILL 10 lines 30-APR-1987 09:42 -< How about this one? >- ---------------------------------------------------------------- I haven't had any problems calling subroutines that were placed at the bottom of the procedure, but I did run into all kinds of trouble if you put a comment (!) on the same line as the SUBROUTINE declaration statement. You'll get all kinds of various errors, from removing your process to continually calling the first subroutine no matter which subroutine was actually wanted. I'm now running v4.5 and have found all my problems to be fixed. Leon Dill Hennepin County, Minnesota ================================================================ Note 24.0 Accounting logical name 4 replies VAXFAM::USER 10 lines 27-APR-1987 17:58 ---------------------------------------------------------------- I've experienced problems on my 2-node VAXcluster when using the system wide logical name "accountng" to redirect the VMS Accountng.dat file. After a crash and reboot, VMS doesn't always honor the logical name and sometimes puts the accounting data back into Sys$Manager. I've made sure I assign the "accountng" log-name early in the boot process. Is this a known bug? (particularly on 8800s?) R. Smith Raytheon Co. Sudbury MA 617 440-3564 VAX-147 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 24.1 Accounting logical name 1 of 4 VAXFAM::LKILGALLEN "Larry Kilgallen" 2 lines 27-APR-1987 18:03 -< It is certainly "supposed to" work >- ---------------------------------------------------------------- It was "fixed" in VMS V2.0. ================================================================ Note 24.2 Accounting logical name 2 of 4 VAXFAM::JMCBRIEN "Jeff McBrien - 3M" 4 lines 28-APR-1987 11:41 -< Documentation? >- ---------------------------------------------------------------- In the current documentation, VMS 4.4, I was unable to locate any documentation to the effect of defining a logical for ACCOUNTNG. I have however noticed the same problem. ================================================================ Note 24.3 Accounting logical name 3 of 4 VAXFAM::VMSDEV 3 lines 29-APR-1987 18:31 -< noted, but not soon >- ---------------------------------------------------------------- I believe this is an undocumented "feature" of the Job Controller. This is "on the list" to provide in a future release, but don't look for it in the infamous "next major release". ================================================================ Note 24.4 Accounting logical name 4 of 4 VAXFAM::PRANKIN "Pat Rankin" 4 lines 30-APR-1987 11:21 -< try brute force >- ---------------------------------------------------------------- You might try the following sequence in your systartup $ set accounting/disable $ define/system/exec accountng 'whatever' $ set accounting/enable VAX-148 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 25.0 Alternate host for disks 5 replies VAXFAM::RCSMITH 7 lines 28-APR-1987 09:51 ---------------------------------------------------------------- Does anyone know how to specify the "alternate host" for an HSC-based disk. We have some RA81s on HSCs which show up differently with a SHOW DEC/FULL. Some show primary host and alternate host; some just say primary host. I'm mounting them all the same way (I think). R. Smith Raytheon Co. ================================================================ Note 25.1 Alternate host for disks 1 of 5 VAXFAM::PCLAYTON 14 lines 28-APR-1987 10:37 -< Option On Using HSC Disks And Mounting >- ---------------------------------------------------------------- I have had similar problems. The case seems to be in making sure both port buttons are pushed and then waiting some 'unspecified' time frame BEFORE doing the MOUNT. The problem as I see it at this time is that the UCB is not updated at any specific time. We have had cases where at boot time, devices where not known to any HSC. The DEC remedy is to do a F$GETDVI an see it the device exists first. If it does, then mount it, otherwise wait for a while and then retry. My other grip is that no ability is available to specify which HSC is to be used for the mount!! Paul D. Clayton Manager Of Systems TSO Financial VAX-149 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 25.2 Alternate host for disks 2 of 5 VAXFAM::GWALROD 4 lines 28-APR-1987 11:08 -< Alloclass >- ---------------------------------------------------------------- Have you tried using the alloclass of the HSC and mounting using the alloclass. Or if you want to mount on a preferred HSC specific the hscnode$DUxx, if that HSC can't access that drive, failover will to the other HSC. This is assuming the disk is dual pathed. ================================================================ Note 25.3 Alternate host for disks 3 of 5 VAXFAM::JPRECIADO 25 lines 28-APR-1987 11:37 -< Let's tell DEC that we want this >- ---------------------------------------------------------------- There is presently no way to specify a preferred host path in the VMS mount command. I have asked DEC for this in the VMS futures talk this week. I want the ability to say use a preferred host to get to this disk if that host is available. Otherwise, use a host that offers access to that disk via the appropriate allocation class. Be aware that the MSCP protocol itself does not dynamic load balancing for the same reasons that it does not allow dynamic dual porting. The HSC also needs exclusive access to the disk in order to maintain real-time positioning data for seek optimization. What I'm after is the ability to do some static load balancing at MOUNT time. My experience has been that VMS will use the first used HSC for all access to disks (and tapes) connected to the same HSC pair (allocation class). The only way to steer it is to manipulate the drive port buttons before and after the mount. Something along the line of the following mount switch would be nice: $ MOUNT $255$DUA7: DISK/SYSTEM/PREFERRED=(TWEEDL) where TWEEDL is the name of the host I would like it to go through if available. If that host is not available of cannot get to the specified disk, then I'd settle for an informational message saying that the mount used an alternate path to the disk. VAX-150 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville How about it DEC (and users) ? ================================================================ Note 25.4 Alternate host for disks 4 of 5 VAXFAM::GLESTER 5 lines 29-APR-1987 11:21 -< Syntax of /PREFER switch >- ---------------------------------------------------------------- I like the syntax. Would /PREFER take an ordered list. I.e. the first path is my first choice and the second path my second choice (and maybe with some future software & hardware a third, fourth, etc.)? The second choice now would just be for clarity in my startup file. ================================================================ Note 25.5 Alternate host for disks 5 of 5 STAR::STILES "When in doubt, hack." 33 lines 29-APR-1987 14:36 -< Here a disk, there a disk, ... >- ---------------------------------------------------------------- re: alternate path disk names: You should always use the allocation class form of a disk name, when a disk has a non-zero allocation class. This allows path decisions to be made regardless of current topology, including recabling drives in the event of (semi-permanent) HSC or K.sdi failure. In particular, shadow set virtual units have NO alternate path information kept about them, and the use of the node$ddcu: form could lead to the inability to access a shadow set should a failover occur. VMS goes out of its way to attempt to resolve the node$ddcu: format when possible, but it is guaranteed that $alloclass$ddcu: will always work. re: no alternate path name for some disks: This is due to the disk not being "available" to the alternate HSC at the time VMS made a connection to it -- most likely due to being mounted on the other HSC. There is a mechanism in MSCP to determine access path relationships, which will eventually find these alternate paths. However, that mechanism was broken in V4.4, so occasionally disks will appear not to have a valid alternate path even after the system has been operational for quite some time. Not to worry, though -- the failover mechanism used by the disk class driver will search all available controllers to "find" a drive when necessary, so dual-path VAX-151 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville dynamic failover continues to operate. (And once this has occurred, the alternate path information is kept up-to-date correctly). Note that dismounting the disk will sometimes cause the information to be updated, and popping/pushing port buttons a few times almost always will. re: mount/preferred_path We hear you. This would make a fine SIR topic. - Mark Stiles VMS Development ================================================================ Note 26.0 DECType, forms & queues 1 reply VAXFAM::JNUNNALLY 2 lines 28-APR-1987 10:03 ---------------------------------------------------------------- Is it possible to specify a form type and queue when printing from within DECTYPE? ================================================================ Note 26.1 DECType, forms & queues 1 of 1 VAXFAM::GLESTER 2 lines 29-APR-1987 11:23 -< Use multi queues >- ---------------------------------------------------------------- You can set up multi queues to same device (easy if a LAT printer) with different defaults. Otherwise I don't know. ================================================================ Note 27.0 SETUAI and passwords 15 replies VAXFAM::JNUNNALLY 2 lines 28-APR-1987 10:04 ---------------------------------------------------------------- Is it possible to use the system service SETUAI to set a user's password using plaintext (or must it be encrypted first?) VAX-152 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 27.1 SETUAI and passwords 1 of 15 VAXFAM::CRHODE "Chris Rhode (Lockhee" 8 lines 28-APR-1987 10:10 -< SETUAI - somewhat unrelated issue >- ---------------------------------------------------------------- On a somewhat unrelated issue...I had heard that there were a lot of inconsistencies in the documentation/implementation of $GETUAI; the person documenting these problems said therefore that they didn't trust $SETUAI...has anybody else had (good/bad) experience with $GETUAI/$SETUAI that they can tell us about? advTHANKSance, -Chris ================================================================ Note 27.2 SETUAI and passwords 2 of 15 VAXFAM::WBAKER 1 line 28-APR-1987 10:30 -< setuai and password >- ---------------------------------------------------------------- system does the encryption. you pass it plaintext. ================================================================ Note 27.3 SETUAI and passwords 3 of 15 VAXFAM::JMCBRIEN "Jeff McBrien - 3M" 5 lines 28-APR-1987 11:45 -< Record size changes >- ---------------------------------------------------------------- My experience with the $SETUAI facility is that it changes the record size of the record that you touch. If you EVER intend to use the documented user area of the UAF this may cause big problems. I hope this is fixed in a future version of VMS (We currently are running 4.4). VAX-153 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 27.4 SETUAI and passwords 4 of 15 VAXFAM::TFRERICKS 13 lines 29-APR-1987 09:12 -< BUG and YOU ENCRYPT. >- ---------------------------------------------------------------- This service has a bug as it is related with group privilege. The problem is that (in pre-4.6) if a user has group privilege, this service thinks they have sysprv. There is a patch (we have installed it at Iowa State University as all of our instructors have/need grpprv. As far as the encryption goes, THE PROGRAMMER NEEDS TO DO THE ENCRYPTION. I didn't want to bother, so I set 0's when I need to change a password for a user using this service. The user is told to then logon and change it immediately. Needless to say, MOST of the password changes I do are done with authorize. Troy Frericks (515)294-1420 ================================================================ Note 27.5 SETUAI and passwords 5 of 15 VAXFAM::APOTTER "Andrew Potter, Roc" 38 lines 29-APR-1987 11:53 -< Actually USING $SETUAI >- ---------------------------------------------------------------- We have found $SETUAI incomplete for adding new accounts. In the interest of making as VMS version independent code as possible, we are using it in our new account authorization package. (call it a callable authorize subroutine library) Here is what we have to do to add an account: 1) Read the DEFAULT UAF record with RMS 2) Change the USERNAME in that Record to the username you wish to add 3) Write the record (with RMS) 4) Using HPWD.MAR (Typed in from the fiche) encrypt a password. 5) Use $SETUAI to set all the necessary account attributes VAX-154 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville (including the encrypted password) 6) Using the other security services, we emulate the behavior of Authorize in granting and maintaining the UIC Identifiers for the account. (Create GROUP identifier from ACCOUNT field if necessary, add USERNAME identifier) 7) Finally we call the XQP QIO interface to add a disk quota (not part of authorize, but desirable in our application) Deleting the account is also difficult but doable. We have to manually delete the Identifiers for USERNAME and possibly GROUP and remove the record with RMS. We have not had any problems with this approach, (other then the known GRPPRV bug.) but are hoping that DEC will finish out these services in a supported callable interface. - Andrew Potter - Rochester Institute of Technology - Rochester NY ================================================================ Note 27.6 SETUAI and passwords 6 of 15 VAXFAM::JHANRAHAN 1 line 29-APR-1987 13:10 -< Adding accounts via sys service calls >- ---------------------------------------------------------------- I'd like to see the code for that in the Pageswapper. ================================================================ Note 27.7 SETUAI and passwords 7 of 15 VAXFAM::LKILGALLEN "Larry Kilgallen" 1 line 29-APR-1987 13:29 -< The Pageswapper does not solicit "code" >- ---------------------------------------------------------------- The Symposium tape is a good place for program exchange. VAX-155 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 27.8 SETUAI and passwords 8 of 15 VAXFAM::RMARSHALL 9 lines 29-APR-1987 14:24 -< Changing password via SETUAI >- ---------------------------------------------------------------- I have used SETUAI to change passwords, but it took me awhile to get it right. The system does the encryption, but as I remember (I did this a long time ago) I had to do something with the SALT, too. Somehow that messed me up and it took me a few hours to figure out what I was doing wrong... Bob Marshall Lockheed (408)756-5737 ================================================================ Note 27.9 SETUAI and passwords 9 of 15 VAXFAM::BPICKARDRIC 5 lines 29-APR-1987 14:31 -< Details of SETUAI and (SALT or SEED). >- ---------------------------------------------------------------- It isn't clear to me how to write the encrypted password using SETUAI. I believe that there is a SEED which is available via GETUAI and SETUAI. Is that seed an arbitrary argument to HPWD? Is there not a address in VMS to which one can branch in order to user the current version of HPWD? ================================================================ Note 27.10 SETUAI and passwords 10 of 15 VAXFAM::SSIMM 8 lines 29-APR-1987 15:04 -< Does it or Doesn't It? >- ---------------------------------------------------------------- OK, I give up... Does it or doesn't it? (Hash the password...) Are there any tricks to getting it to hash it? Steve Simm GE Consulting Services 518-458-2409 VAX-156 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 27.11 SETUAI and passwords 11 of 15 STAR::PIPER "Derrell Piper - VAX/VM" 14 lines 29-APR-1987 15:39 -< Nope >- ---------------------------------------------------------------- OK, I give up... Does it or doesn't it? (Hash the password...) Are there any tricks to getting it to hash it? $SETUAI does not hash the password. The UAI$_PASSWORD item code in these services returns/sets the quadword password field in the UAF file. This quadword is the result of the system hashing function (which takes as one of its inputs a SALT value for the username/password being encrypted.) Unfortunately there are no tricks to access the system hashing function (HPWD). (Although as was stated in an earlier reply, if you must have it you can copy it from the µfiche.) ================================================================ Note 27.12 SETUAI and passwords 12 of 15 N8700::TFRERICKS 4 lines 30-APR-1987 12:59 -< Page on Fiche >- ---------------------------------------------------------------- From memory, it is on page 258h of the fiche. It is cryptic to type in, prone to errors (typos) so test it. Troy Frericks (515)294-1420 VAX-157 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 27.13 SETUAI and passwords 13 of 15 N8700::TFRERICKS 12 lines 30-APR-1987 16:24 -< Encryption Algorithm >- ---------------------------------------------------------------- HAY DEC... Since the code for HPWD is copyrighted,,, How about "fixing" SETUAI so it could do the encryption when passed the SALT and USERNAME (of course) and "RAW_PASSWORD". Or a system service that would return the quad word when the username and salt are passed. Something... Anything... (except a reference to the fiche!). Troy Frericks (515)294-1420 ================================================================ Note 27.14 SETUAI and passwords 14 of 15 VAXFAM::STOLLBOM "Cullen Tollbom, Ba" 4 lines 30-APR-1987 17:51 -< $SETUAI inadequate >- ---------------------------------------------------------------- I had several instances where I need to update/write in the UAF. $SETUAI was primitive enough that I tossed the whole idea of using and developed RMS routines (with full file share and record locking of course). This works. ================================================================ Note 27.15 SETUAI and passwords 15 of 15 STAR::PILANT "L. Mark Pilant" 10 lines 30-APR-1987 20:44 -< $SETUAI inadequate?...Tell me why >- ---------------------------------------------------------------- RE: Problems with $SETUAI/$GETUAI Submit your problems/bugs/suggestions via SPRs or contact the folks at the TSC (if you have the appropriate support). These suggestions do make their way back to us. Software development involves feedback. If we never hear VAX-158 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville anything is wrong, it'll never improve. - Mark, VMS Development ================================================================ Note 28.0 locating default directories 8 replies VAXFAM::JNUNNALLY 3 lines 28-APR-1987 10:05 ---------------------------------------------------------------- Is it possible to determine the default directory of another user? If it is necessary to follow the links through the PCB, where does one look? ================================================================ Note 28.5 locating default directories 5 of 8 VAXFAM::SSIMM 12 lines 29-APR-1987 14:33 -< Interprocess AST >- ---------------------------------------------------------------- Finding out the current default directory requires (as far as I know) an inter-process AST -- kernel mode -- to go over to the target process and look this information up. Both Jim Precidio (sp?) and Bruce Ellis can help you with this sort of thing. Bruce did a PSS on internals and gave out some code that could be twisted around to perform this. Steve Simm GE Consulting 518/458-2409 ================================================================ Note 28.6 locating default directories 6 of 8 VAXFAM::BMULLALLEY "Barry Wallis" 9 lines 29-APR-1987 15:58 -< I think I have a routine to do it? >- ---------------------------------------------------------------- I may have some code back at work that does this. If I can get on a terminal I will look at it for you. Otherwise, you can get in touch with me after the symposium. VAX-159 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Barry L. Wallis Fleetwood Enterprises, Inc. 3125 Myers Street Riverside, CA 92523 (714)351-3682 ================================================================ Note 28.7 locating default directories 7 of 8 STAR::STILES "When in doubt, hack." 20 lines 29-APR-1987 16:36 -< Behind door #3... >- ---------------------------------------------------------------- The magic location you wish to find is PIO$GT_DDSTRING, a counted string containing the current default directory string, including brackets. Since this is a VMS defined location, its address may change per (major) release. Since it is a P1 location (per process), you will have to use the special-kernel AST mechanism to do this via program -- if you're just interested for a particular process, you can use SDA, and do: $ ANALYZE/SYSTEM SDA> SET PROCESS/INDEX=x ! where x is the process of interest SDA> EXAMINE PIO$GT_DDSTRING SDA> EXAMINE ! keep examining till it's all gone... If what you are looking for is a program method, I suggest you contact the person in .6 -- getting the special-kernel ASTs right the first time is tricky... - Mark Stiles VMS Development VAX-160 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 29.0 SMG TERMTABLE entries 2 replies VAXFAM::JNUNNALLY 4 lines 28-APR-1987 10:07 ---------------------------------------------------------------- We have had trouble adding a definition to the SMG TERMTABLE. The definitions don't seem to 'take'. Is there an example somewhere that we could follow? Any suggestions? We would like to add a definition for a foreign terminal. ================================================================ Note 29.1 SMG TERMTABLE entries 1 of 2 N8700::RLGRAHAM "Bob Graham, Dow Ch" 18 lines 28-APR-1987 10:33 -< getting a new TERMTABLE to go >- ---------------------------------------------------------------- One of the basic problems with the SMG termtable as it is implemented now is that you have to reboot the system for the new definitions to "take". That's because the termtable is mapped as a global section at boot time. Theoretically you could remove it and install the new one on line, but if you did, all the processes that are currently mapped to it would die. Not nice. The easiest way to test a new termtable is to create it and then define it as your process private table. I forget the exact logical name but I think it's something like SMG$TERMLOC. I do remember that the logical name should point to the directory only, not the file name. Anyway, it's documented in the RTL manual where SMG is discussed. Once you have the termtable working as a private table, then you can copy it to SYS$SYSTEM and reboot. Bob Graham Dow Chemical, La. Division VAX-161 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 29.2 SMG TERMTABLE entries 2 of 2 VAXFAM::PRANKIN "Pat Rankin" 2 lines 28-APR-1987 12:36 -< Process-specific Termtable >- ---------------------------------------------------------------- The logical name is TERM$TABLOC and defines the directory where SMG should look to find TermTable.Exe. ================================================================ Note 31.0 Printer Queue Problem 3 replies VAXFAM::JNUNNALLY 3 lines 28-APR-1987 10:10 ---------------------------------------------------------------- Why should a printer queue drop a /BLOCK_LIMIT value when a simple STOP/QUEUE/RESET followed by START/QUEUE/FORM=xxx is performed in order to change the forms on the printer? ================================================================ Note 31.3 Printer Queue Problem 3 of 3 STAR::GUZMAN "Ray Guzman -- VMS Engi" 4 lines 29-APR-1987 15:22 -< The fix is in >- ---------------------------------------------------------------- This problem has been corrected in V4.6. The work-around is to always specify the block limit when doing an INITIALIZE/QUEUE, START/QUEUE, or SET QUEUE. ================================================================ Note 34.0 BACKUP: Which tape next? 14 replies VAXFAM::USER 10 lines 28-APR-1987 10:26 ---------------------------------------------------------------- Is there a way to make BACKUP recognize a tape label? Specifically, can BACKUP be made to recognize a set of pre-labeled tapes. In backing up our files, we rely on the operators to mount the correct tape. Is there someway to tell BACKUP to only use a certain series of tapes during this backup session? VAX-162 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Howard de St. Germain US Army TRAC White Sands Missile Range New Mexico. ================================================================ Note 34.4 BACKUP: Which tape next? 4 of 14 VAXFAM::RCRITZ "Richard Critz" 6 lines 28-APR-1987 11:15 -< in the future >- ---------------------------------------------------------------- RE .0: The BACKUP futures talk on Monday evening indicated that BACKUP will use the $MOUNT system service for all requests with the attendant tape label processing. This support is in some future release of VMS. ================================================================ Note 35.0 BACKUP: Which volume on which drive. 2 replies VAXFAM::LCROWDER 6 lines 28-APR-1987 10:43 ---------------------------------------------------------------- Is there a way to determine which tape volume number is mounted on which tape drive? There must be a something defined in the system somewhere that knows that info. What I wish to do is track the number of volumes during backup. The idea is know which tape drive holds the last volume after an image backup. That way, I could know which tape drive to begin the next disk backup upon. ================================================================ Note 35.1 BACKUP: Which volume on which drive. 1 of 2 VAXFAM::WBAKER 6 lines 28-APR-1987 10:55 -< volume number on tape volumes. >- ---------------------------------------------------------------- when the system initializes the next volume of a volume set it uses the first four characters of the first volume, and then adds its own number 01, 02, 03, etc... until done. perhaps just ask the system to do a f$getdvi on each device and report the label on the tape to you, then parse out the last two numbers and see which one is the highest number. VAX-163 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 35.2 BACKUP: Which volume on which drive. 2 of 2 VAXFAM::LKILGALLEN "Larry Kilgallen" 13 lines 29-APR-1987 13:40 -< Use DISMOUNT/NOUNLOAD >- ---------------------------------------------------------------- Then remount all available tapes to find their labels. If you have initialized all tapes in advance to FIXDEC, then the one with a label you specified to Backup is the last one to which it has written. Alternatively, if you run a pseudo-labeled tape shop, initialize all tapes to their proper value (but without files). The tape whose first file is not the null place holder has data on it. Using these techniques requires you ALLOCATE the drives (unfortunately) because there is a window of vulnerability during the time the tape is dismounted. ================================================================ Note 37.0 DEFAULT ACLs don't work on remote nodes 1 reply VAXFAM::HSPAIN 7 lines 28-APR-1987 10:53 ---------------------------------------------------------------- Default ACL protection on remote nodes is bypassed. We use this scheme to assure that files all have correct protections even when created from other nodes in the network. This stopped working around V4.4. Has this been repaired in 4.6? Harrison Spain McDonnell Douglas VAX-164 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 37.1 DEFAULT ACLs don't work on remote nodes 1 of 1 VAXFAM::STOLLBOM "Cullen Tollbom, Ba" 5 lines 30-APR-1987 17:58 -< Answer to Protect Problem >- ---------------------------------------------------------------- I had the same question. Not only does it ignore ACLs, it also ignores the SOGW bits. VMS Development told me the problem was fixed in VMS V4.6. It broke in V4.3, I believe. Something about optimizing wild card searches and they broke the $XABPRO part of it. ================================================================ Note 42.0 LAVC booting concerns 5 replies VAXFAM::DBOLTHOUSE 9 lines 28-APR-1987 14:05 ---------------------------------------------------------------- Is it possible to do a conversational boot from a LAVC member (i.e., not the boot node). If so, I think there is an interesting problem VMS Development needs to address; would rather not post details. If not, what level of confidence can I have in this mechanism? adTHANKSvance David L. Bolthouse Texas Instruments Inc. ================================================================ Note 42.1 LAVC booting concerns 1 of 5 VAXFAM::RPETERSEN 4 lines 28-APR-1987 14:12 -< LAVC booting concerns >- ---------------------------------------------------------------- My understanding on conversational boots for a non boot node was that it was an option that you need to specify when adding a satellite node. VAX-165 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 42.2 LAVC booting concerns 2 of 5 VAXFAM::GKOHLS "Gary ap Kohls" 5 lines 29-APR-1987 12:24 -< Toggle conversational? >- ---------------------------------------------------------------- R. Petersen is exactly correct. The cluster manager set this with the satellite configuration command procedure. I haven't looked into how this is changed after the satellite is created as it is a clear security hole but is very useful to have when dealing with boot problems on the satellite. ================================================================ Note 42.3 LAVC booting concerns 3 of 5 VAXFAM::JHOUGHTON 6 lines 29-APR-1987 14:18 -< Satellite node conversational boot flag exists >- ---------------------------------------------------------------- Look at SATELLITE_CONFIG.COM. There is a SYSGEN parameter which allows/disallows conversational boot (it is one of the non-descriptively-named parameters (PE3 comes to mind) which is set/reset in response to the "allow conversational boots" question, and was mentioned in the release notes that came with the field-test version. ================================================================ Note 42.4 LAVC booting concerns 4 of 5 VAXFAM::RPETERSEN 5 lines 30-APR-1987 10:00 -< Toggle conversational? >- ---------------------------------------------------------------- I am sure that it is a sysgen parameter that can be changed. Off the top of my head though I cannot remember which one. Check your doc I guess, or see Ray Kaplan who gave an excellent talk on sysgen yesterday. VAX-166 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 42.5 LAVC booting concerns 5 of 5 VAXFAM::TGARDNER 5 lines 30-APR-1987 10:44 -< It really works >- ---------------------------------------------------------------- Indeed, the conversational boot option can be disabled when the satellite is configured. I didn't really believe it at first, but it does work as advertised. (Thank God... otherwise, the LAVc would be nearly useless in most environments). ================================================================ Note 43.0 TCP/IP under VMS 7 replies N8700::DLANDERSON 6 lines 29-APR-1987 09:22 ---------------------------------------------------------------- Can anyone help with some info/discussion on available software products supporting TCP/IP under VMS for the new BI DEC Ethernet interface? We are bringing an 8350 into an environment where the only terminal access will be via non-DEC Ethernet terminal servers working under TCP/IP. Any other discussion on TCP/IP for VMS will be welcome in the replies to this note. Thanks. ================================================================ Note 43.1 TCP/IP under VMS 1 of 7 VAXFAM::JCHRISTENSEN 4 lines 29-APR-1987 11:06 -< Try WIN/VX >- ---------------------------------------------------------------- We are using WIN/VX from The Wollongong Group on our VAX 8500 with DEBNT Ethernet controller. There is a patch required on the VMS 4.5 ETDRIVER, but once we got that figured out, everything has been working OK. VAX-167 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 43.2 TCP/IP under VMS 2 of 7 VAXFAM::JHANRAHAN 2 lines 29-APR-1987 13:12 -< Also try NRC >- ---------------------------------------------------------------- Network Research Corp.'s FUSION 3.0 will work with the DEBNT and its replacement. (Either it does now, or it will very soon.) ================================================================ Note 43.3 TCP/IP under VMS 3 of 7 VAXFAM::MSHANNON 15 lines 29-APR-1987 13:32 -< CMU has TCP/IP too (CMU/Tektronix) >- ---------------------------------------------------------------- Another available TCP/IP is CMU's modified Tektronix code. It is available (I think free, or at least the cost of distribution) to anyone with a Tektronix TCP/IP license. [Tektronix, I believe, has very special provisions for this to educational institutions.] To find out more about obtaining CMU's Tek TCP/IP (it's a really good product - it runs well on my uVAX-I :-) ), send mail (the real USPS kind) to: CMU/Tektronix TCP/IP Distribution Carnegie-Mellon University Computing Center 4910 Forbes Avenue Pittsburgh, PA 15213 --Marc ================================================================ Note 43.4 TCP/IP under VMS 4 of 7 VAXFAM::DBOLTHOUSE 26 lines 29-APR-1987 17:11 -< TCP/IP experiences >- ---------------------------------------------------------------- Re: TCP/IP. We looked at 4 vendors: EXCELAN, INTERLAN, NRC (Fusion Network Software), and TWG (The Wollongong Group). The first two are hardware/software solutions; the second two are software only. The Tek/CMU distribution wasn't available when I did the evaluation (8 months ago now). VAX-168 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville The hardware/software solutions are inherently more stable (less kernel mode code to crash your VAX since much work is done down on the board) and are inherently much faster (TELNET echoing is done down on the board). However, they don't do much for you if you're running on a VAXBI-based machine and you don't have the UNIBUS adapter board. We have NRC's FNS on one machine; within 3 days of installing the software, we had a kernel mode crash. (ACCVIO in the FNS code, R2 was blown...) This is typical behavior, in our experience. On the other hand, when I did the evaluation, I was also able to crash TWG software. The hardware/software solutions were more robust. In NRC's defense, they attempt to give you support. I have heard dozens of complaints against TWG regarding inadequate support, both on the Internet, and even a couple of disgruntled users here this week. In short, take your time, tread carefully, and exercise those evaluation copies! ================================================================ Note 43.5 TCP/IP under VMS 5 of 7 VAXFAM::DBOLTHOUSE 8 lines 29-APR-1987 17:13 -< Last note extended... >- ---------------------------------------------------------------- I shouldn't have finished the previous note so quickly. We have had NRC's FNS for a couple of years; that wasn't clearly 'noted'. Also, I forgot to tell you who I am... David L. Bolthouse Texas Instruments, Inc. 214-952-2059 Internet: bolthouse%mcopn1@ti-eg.com VAX-169 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 43.6 TCP/IP under VMS 6 of 7 VAXFAM::KCAROSSO "Kevin Carosso" 42 lines 29-APR-1987 17:19 -< Useful TCP/IP tips >- ---------------------------------------------------------------- Yeah, the CMU stuff works quite well and the price is right! One thing to watch out for. The initial note on this topic mentioned that the primary interest in TCP/IP was remote (TELNET) terminal sessions. You should be aware that some TCP/IP/TELNET implementations are not particularly suited to efficient remote terminal connections. A TELNET connection into a VAX/VMS host running the CMU code will induce a lot of overhead processing as characters to/from the terminal session flow out into/out of the user's process, through a pseudo-terminal, through the TELNET server process, through the TCP/IP network ACP, and finally through the network device. Added to the overhead induced by the flow of information through so many entities is the compute cost of the TELNET/TCP/IP protocol layers. I suspect that Wollongong's networking software is a bit more optimized, so the performance hit may be smaller. You might have other problems with Wollongong. Some users are less than ecstatic with the product. Another approach that might work for you is a TELNET/TCP/IP on a board. This offloads some of the processing from the VAX and may make remote terminal sessions via TELNET gentler to the system. Excelan, MICOM, and CMC have products that may do this (well, Excelan does for sure...) Again, there may be other reasons this approach is not viable, you'll have to make the determination. If you would like more information on TCP/IP options for VAX/VMS, or any other system, contact the Network Information Center of the Stanford Research Institute. I'm sorry I do not have addresses or phone numbers, but you should be able to track down SRI on your own and work from there. The SRI NIC maintains an extensive list of TCP/IP implementations, called the Vendor's Guide, which they will happily send to you. VAX-170 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Of course, in addition to the Vendor's Guide, the SRI NIC maintains documentation on the entire Internet protocol suite if you would like to roll-your-own! /Kevin Carosso kvc@engvax.scg.hac.com Hughes Aicraft Co. ================================================================ Note 43.7 TCP/IP under VMS 7 of 7 VAXFAM::WTHOMSON 12 lines 30-APR-1987 17:00 -< I have MICOM and Wollongong >- ---------------------------------------------------------------- We had a lot of problems getting non-transparent communications to work with MICOM/INTERLAN .. their TELNET and FTP are OK though.. We currently have both their HW/SW and also Wollongong (although we just received it). The Wollongong SW looks 'spiffier' has the FINGER function that looks real neat.. Otherwise haven't had time to evaluate it.. Bill Thomson Martin Marietta Baltimore Aerospace (301) 682-1643 ================================================================ Note 44.0 Shareable Image...Can it be done? 4 replies VAXFAM::WWEISSBORN 34 lines 29-APR-1987 10:35 ---------------------------------------------------------------- I have a user with the following request. Any help would be appreciated. The need is to be able to produce a sharable image with the following characteristics. 1. Developed using an application library with a locally supplied interface to that library. 2. Shareable with other code which DOES NOT HAVE ACCESS to the modules in the application library. The external code VAX-171 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville should only have access to the entry point supplied by the interface. (all direct access to the library modules by external code would be expressly forbidden.) Example. External object library Local Object library ----------------------- -------------------- Contains modules A-Z Main routine XYZ which calls A,D,J,and Q from the external library the entry points for the external library (A,D,J,Q) are "invisible" and cannot be used for development purposes by some other user. Can it be Done???????? Bill Weissborn Schlumberger Well Services 12770 Coit Rd. Suite 210 Dallas, Tx. 75251 (214) 980-7924 ================================================================ Note 44.1 Shareable Image...Can it be done? 1 of 4 VAXFAM::GLESTER 19 lines 29-APR-1987 11:43 -< Here's how to do it >- ---------------------------------------------------------------- Yes, watch the hands they never leave the keyboard! 1) Your local lib should contain dummy routines that give an error (or take other action) when called. Same name as the "hidden" routines. 2) Your local shareable lib needs and "owned" variable that says if this is the first time it is being called, or have an initialization routine (your choice). 3) On initialization or first time call a special initialization routine that does the following "magic": VAX-172 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville a) Uses LIB$FIND_IMAGE_SYMBOL to find routine names in the "forbidden" lib. b) Stores addresses from (a) in "owned" data structure. c) Sets initialize done flag 4) Your interface routines then call the "hidden" routines with a CALLG (or LIB$CALLG) using the address in (3b) Have fun. ================================================================ Note 44.3 Shareable Image...Can it be done? 3 of 4 VAXFAM::SSIMM 23 lines 29-APR-1987 14:45 -< CALLS (Rx) >- ---------------------------------------------------------------- The C (or perish the though, MACRO) can be passed in the address of a function to call -- NOT the NAME. The routine can then call the routine directly. In MACRO this looks something like this: PUSHL data PUSHAL moredata CALLS (R4) ; R4 holds the address of the routine ; to call. LIB$FIND_IMAGE_SYMBOL performs the "link" at run time. The above technique performs the link at link time. The only major difference between the two is that LIB$FIND_IMAGE_SYMBOL requires a couple of strings and the CALLS requires an address. DEC uses the CALLS type passing in some of the system services (LIB$PUT_OUTPUT is usually the default) and LIB$FIND_IMAGE_SYMBOL is used in the SET HOST/DTE mechanism. Steve Simm GE Consulting Services 518/458-2409 VAX-173 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 44.4 Shareable Image...Can it be done? 4 of 4 VAXFAM::GLESTER 3 lines 30-APR-1987 09:12 -< Thanks for the correction >- ---------------------------------------------------------------- Most correct. I could not remember the name at the time so took a best guess. It did not sit quite right so when I had a minute I was going to check the name. ================================================================ Note 47.0 Incorrect page count on laser printer o/p 7 replies VAXFAM::RGROMAN 10 lines 29-APR-1987 12:06 ---------------------------------------------------------------- We are using a QMS 800 laser printer on our cluster. On certain types of jobs (UNIRAS graphics output/rasterized images) the page count that is entered into the VMS accounting record is high, by a factor of 100 or so. We believe this is because it is counting each rasterized line as a line of text. Does anyone else have this problem? Has anyone solved this problem? Thanks Bob Groman Woods Hole Oceanographic Inst. ================================================================ Note 47.1 Incorrect page count on laser printer o/p 1 of 7 VAXFAM::ASORRELL "Al Sorrell, Westi" 23 lines 29-APR-1987 13:18 -< A little light - maybe a solution >- ---------------------------------------------------------------- Yes - the difficulty can also be seen using MEC's MASS11 Draw - Here's the source of the problem -- The codes which drive this printer (QUIC) are of the form ^Uxxxx:yyyy which defines a "move pen-up to location x,y". Most "smart" packages will try to buffer up a bunch of the commands in a buffer up to 255 bytes long, then output it as a single record followed by a . VAX-174 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville If you have your printer defined with a page length of, say 66, accounting thinks that for every 66 pairs you have just generated 1 PAGE of output and you get billed accordingly. We just discovered this a few days before coming down here. One thought (not yet tried!!) is to define a form with 0 or maybe 65535 lines/page and tell users to use this form for graphics output if they don't want excessive billing. Then use another form for "standard" printout. Al ================================================================ Note 47.3 Incorrect page count on laser printer o/p 3 of 7 VAXFAM::JARTLEY "Judy Artley, Rayth" 13 lines 29-APR-1987 14:59 -< Pages printed = f(gets and puts)? >- ---------------------------------------------------------------- We are going to start charging for pages printed and noticed that jobs PRINTed/NOFEED have very low page counts. (This includes the MASS-11 print jobs.) Does anyone have a rule of thumb for relating the number of GETS and PUTS (as recording in the accountng record) to the number of pages on various printers? We are thinking of doing regression analysis. Our sites have LN03's, diablo, line printers, HP plotters and zeta plotters. -- Judy Artley Raytheon Co. 2-2-421 200 Nickerson Road Marlboro, MA ================================================================ Note 47.4 Incorrect page count on laser printer o/p 4 of 7 VAXFAM::CRHODE "Chris Rhode (Lockhe" 16 lines 29-APR-1987 16:54 -< Low/Incorrect page counts in ACCOUNTING >- ---------------------------------------------------------------- PRINT/PASSALL definitely records only one page printed regardless of the number of pages actually printed. Third-party products may use PRINT/PASSALL without your knowing it (I didn't, until I had to figure out why print counts were so low). I can't see that VMS could be made any more intelligent in this case because there is no clear way to define when a page is VAX-175 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville complete if you are passing data straight through and the printer itself defines end of page. Perhaps smart printers that can send back a "pages printed" count? Or some intelligent software on the VAX side that can interpret stuff sent through it to determine where the physical end of page will be? Ick. Sigh. The bottom line seems to be...don't trust printer accounting... you could assign departmental printers (via queue ownership and protection) and use the page counter on the printer to charge usage, but I know that route is unacceptable for many sites. ================================================================ Note 47.6 Incorrect page count on laser printer o/p 6 of 7 VAXFAM::ASORRELL "Al Sorrell, Westi" 12 lines 30-APR-1987 10:02 -< >- ---------------------------------------------------------------- GETs & PUTs and file sizes, etc. really aren't gong to give much of a clue. In fact, I think we've bumped into a problem that right now admits of no real good solution. Fundamentally, there is no way of differentiating a 500 block file of vectors which outputs 1 page from one that outputs 500 pages, at least not if there are no embedded s. I almost wonder if users of graphics etc. which require more time in the printer (esp Postscript!) maybe shouldn't be charged more - thus maybe the "surcharge"is not unjust in itself. My users, of course, think otherwise. Al ================================================================ Note 47.7 Incorrect page count on laser printer o/p 7 of 7 VAXFAM::RLGRAHAM "Bob Graham, Dow C" 21 lines 30-APR-1987 10:28 -< when it doubt, write it yourself >- ---------------------------------------------------------------- There is a way to do this, but you won't like it. That is to write a user-written (NOT user-modified) print symbiont. The reason the user-modified symbiont route can't be used is because the one DEC supplied routine that you can't replace is the FORMAT routine, which is where the page and other statistics are gathered. VAX-176 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville I have written a multi-stream, user-written symbiont to support various plotters that we have. One of the side effects was that since the symbiont understood when a new page was put into the plotter, it could keep accurate statistics. This wasn't the reason for writing the symbiont, but it was a nice side effect. I'd be glad to put my symbiont on the VAX SIG tape, but I have a lot of trouble with Dow to get permission to do things like that. Maybe at the next symposium... Bob Graham Dow Chemical PO BOX 400 Bldg 2503 Plaquemine LA 70765 504/389-1689 ================================================================ Note 49.0 Why is padding rec'd for transfer vectors? 6 replies VAXFAM::JHANRAHAN 15 lines 29-APR-1987 13:48 ---------------------------------------------------------------- In the directions for creating a shareable image (if I remember right it's in the LINKER manual, but I'm not sure), DEC recommends that padding space be left at the end of the transfer vectors `to allow room for future expansion'. I know how the transfer vectors work, I know how the fixup vectors work, etc., and I can't for the life of me figure out why you would need to leave such padding space, except maybe if it was a based shareable image. Is this a remnant from an early version of the doc., or am I missing something somewhere? VAX-177 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 49.1 Why is padding rec'd for transfer vectors? 1 of 6 VAXFAM::SSIMM 12 lines 29-APR-1987 14:52 -< Room to Remove a Vector >- ---------------------------------------------------------------- I've padded space IN the transfer vectors to allow me to stick in a MOVL #SS$_NORMAL,R0 followed by a RET to allow me to yank out various routines without relinking. As for padding at the end I've never bothered and have been supporting many versions of the same shareable for quite a while. (DEC's wording in the manual puzzled me, too.) Steve Simm GE Consulting Services 518/458-2409 ================================================================ Note 49.2 Why is padding rec'd for transfer vectors? 2 of 6 N8700::SLIONEL "Steve Lionel - DEC " 36 lines 30-APR-1987 11:19 -< Pad for downwards compatibility >- ---------------------------------------------------------------- The text you mention may have dated from VMS V1 days, when there were more restrictions on shareable images (in fact, you could really only have one shareable image, since if you had more than one, you would have to relink if one of them changed its size). My memory is a bit fuzzy on this. However, there is still a good reason to pad at the end of a vector, though not just with zeroes. Consider the frequent complaint that images linked on version V4.4 (say) of VMS won't run on V4.2, because the global section match ids of the RTL shareable images have changed. The GSMATCH is changed when the vector is changed (added to). If this wasn't done, and your program, linked on V4.4, called the new routine, you would get unpredictable (and potentially VERY nasty) results if you ran it against the V4.2 RTL that didn't have the new vector where your program expected. VAX-178 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville While our recommendation for those who want to support multiple releases is to link against the shareable images of the earliest VMS version you want to support, another way is to pad your transfer vector with dummy entries - ones which will call an error routine (that may perhaps signal "unimplemented routine called") if one of the new entries is called on an older version of the shareable image. Then, when you want to add a new routine, replace one of the dummy entries with a real one for your routine. If you do this, you can avoid changing the GSMATCH id except in cases where the vectors are truly incompatible (if you run out of dummy entries, for instance). Several of the VMS-supplied shareable images are using this technique to reduce the frequency of GSMATCH changes. More details are provided in the "VAX/VMS Guide to Creating Modular Procedures". Steve Lionel Ex VMS RTL developer ================================================================ Note 49.3 Why is padding rec'd for transfer vectors? 3 of 6 N8700::RLGRAHAM "Bob Graham, Dow Che" 6 lines 30-APR-1987 13:16 -< Thanks DEC >- ---------------------------------------------------------------- Yeah!! I'm glad that someone at DEC is doing something to cut down on the frequency of GSMATCH updates. Thanks much. Now if we could only get certain third party folks, !@#%@$#^$ SAS in particular, to quit using "BASED" images, we could save a lot of time and problems for us poor systems folk. VAX-179 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 49.4 Why is padding rec'd for transfer vectors? 4 of 6 VAXFAM::JHANRAHAN 12 lines 30-APR-1987 14:22 -< I'd prefer GSMATCH error msgs, myself >- ---------------------------------------------------------------- I too am glad to see that DEC is trying to cut down on the frequency of GSMATCH updates. However, I disagree that padding the transfer vector with vectors to dummy routines that just RET is particularly safe. If you call a routine that you expect to do something, and it just RETs, that could be just as nasty as calling a routine whose transfer vector doesn't exist at all (since many programmers don't check return status)... Meanwhile, my answer for avoiding RTL GSMATCH troubles is to ship .OLB files and link on the target system. I note that DEC does this with certain products. ================================================================ Note 49.5 Why is padding rec'd for transfer vectors? 5 of 6 VAXFAM::SLIONEL "Steve Lionel - DEC" 11 lines 30-APR-1987 16:07 ---------------------------------------------------------------- Re: .4 I did not suggest padding with routines that just return, but rather routines that signal errors. The problem with shipping objects and linking on-site is that you no longer are in control of what your program's environment is, and thus are less confident of how your program will behave in such an environment. I consider this an inadvisable technique. Steve VAX-180 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 49.6 Why is padding rec'd for transfer vectors? 6 of 6 VAXFAM::KGLOSSOP 6 lines 30-APR-1987 17:58 ---------------------------------------------------------------- Note that it is possible to have an entry point that will signal a "Not yet implemented" error (or something similar) regardless of how it is called (i.e. xSB, JMP, BRx, CALLx). The trick is to make the entry mask an executable instruction. I don't have a sequence handy, but I can look it up and post it if anyone is interested. ================================================================ Note 51.0 PCA and DTM 5 replies VAXFAM::WTKING 7 lines 29-APR-1987 14:12 ---------------------------------------------------------------- Does anyone have any practical experience with DEC's PCA, performance/coverage analyzer, or DTM, DEC test manager? How does it work with large programs generating lots of output files which include timestamps? ================================================================ Note 51.1 PCA and DTM 1 of 5 VAXFAM::MDICKAU 15 lines 29-APR-1987 16:09 ---------------------------------------------------------------- You should probably come by the L&T campground and talk to me or Walt Carrell (PCA project leader). Without knowing what you are doing, all I can say is that PCA can handle large programs, although there may be some overhead because it needs to watch various pieces of your program, and the bigger the program, the more pieces to watch... DTM does not care how many files you have, although you may run into practical limits of RMS with many files in a single directory. Still, DTM gives you the ability to specify where files are to be kept, allowing you to easily split up the files should sheer number produce performance problems. VAX-181 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Martin Dickau DEC/Test Manager Development ================================================================ Note 51.2 PCA and DTM 2 of 5 VAXFAM::ASORRELL "Al Sorrell, Westin" 9 lines 29-APR-1987 16:27 -< PCA's pretty neat! >- ---------------------------------------------------------------- < Note 51.0 by VAXFAM::WTKING > |Does anyone have any practical experience with DEC's | PCA, performance/coverage analyzer, or DTM, DEC test | manager? We've used PCA for quite a while on some relatively large Ada programs and a few Fortran programs. We've had very good luck with it. Al ================================================================ Note 51.3 PCA and DTM 3 of 5 VAXFAM::GLESTER 9 lines 30-APR-1987 09:24 -< y >- ---------------------------------------------------------------- I've used PCA on several programs (including a DB compiler). It is very easy to use with a high level language and not to bad with MACRO (all you have to do is filter out the ZERO locations). The programs we use it with are generally LARGE (for some reason they seem to be the ones that eat system resources) and have had no problems. If you can use DEBUG to debug the program PCA seems to be able to fit. I'm giving a session LT079 in the Jefferson Room Friday at 15:30. The title is experiences with VAXSET (which PCA is part of). VAX-182 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 51.4 PCA and DTM 4 of 5 VAXFAM::M_KOWALSKI 18 lines 30-APR-1987 13:15 -< SPSS experience >- ---------------------------------------------------------------- I've used DTM on a procedure of SPSS-X which was fairly small, ~10,000 Fortran source lines, 100 routines. I had about 40 tests for the code which generated about 10 blocks of output apiece. I was completely satisfied with how DTM did its job, although the documentation was a bit difficult to get through the first time. Integration with CMS was extremely nice! DTM is now being integrated by our Quality Assurance guys into the total product test environment. I don't have anything to report yet. They are planning to use it and PCA to rework our test library. Call us in about 3 months for further info or watch this topic at the next symposium. Be forewarned that use of the VAXset tools for software development will significantly change your CPU usage...and not for the good! But the flip side is that your developers and code will greatly benefit! ================================================================ Note 51.5 PCA and DTM 5 of 5 VAXFAM::M_KOWALSKI 5 lines 30-APR-1987 13:16 -< contact for spss >- ---------------------------------------------------------------- Oops! To contact us: Mark Kowalski SPSS Inc 402 Washington Northfield, MN 55057 VAX-183 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 54.0 analyze/disk question 5 replies VAXFAM::TKENNEDY 12 lines 29-APR-1987 15:23 ---------------------------------------------------------------- I have a problem (I think) with analyze/disk. I have a 780 with a single RA81 disk. anal/disk/repair reports 'incorrect directory back link' for one of the SYS... directories (I can't remember which). It then implies that this was corrected by not giving any error message. On the next attempt, it still shows as an error. Is this a consequence of running it on a mounted system disk? Thanks in advance for your help, Terry Kennedy St. Peter's College Jersey City, NJ ================================================================ Note 54.1 analyze/disk question 1 of 5 VAXFAM::RLGRAHAM "Bob Graham, Dow C" 10 lines 29-APR-1987 15:42 -< possible file name aliases >- ---------------------------------------------------------------- If you have the VAX/RSX product on your system, then you will get this "error" because that product creates aliases for some of the SYS directories { [1,2] ==>> [SYSLIB] and [1,54] ==>> [SYSEXE] }. Repeated running ANAL/DISK/REPAIR just flops the back links between the named and numbered directories. If you don't have VAX/RSX, then you may want to look for some other directories which have aliases. These could have been created by some privileged user using the SET FILE/ENTER command. If not needed, then you can remove the alias with the SET FILE/REMOVE command. VAX-184 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 54.2 analyze/disk question 2 of 5 STAR::STILES "When in doubt, hack." 6 lines 29-APR-1987 16:08 -< probably not important >- ---------------------------------------------------------------- VMS also uses alias directories and files. Very likely it is the backlink for SYSBOOT (which is entered in multiple directories) that it is complaining about -- if so, this message can safely be ignored. - Mark Stiles VMS Development ================================================================ Note 54.3 analyze/disk question 3 of 5 VAXFAM::RFAUST "Rich Faust, E. I. du" 5 lines 29-APR-1987 17:29 -< another disk/analyze question... >- ---------------------------------------------------------------- I get an error message when trying to run analyze /disk on an RA81 user disk on an 8550 which is not in a cluster. The message says something about some kind of caching preventing the operation. I realize this is vague, but the problem may be simple enough that it has an answer... ================================================================ Note 54.4 analyze/disk question 4 of 5 VAXFAM::PRANKIN "Pat Rankin" 6 lines 30-APR-1987 11:06 -< if it works, don't fix it! >- ---------------------------------------------------------------- re: .0 & .1 Installing stand-alone backup also produces additional links. Despite the warning from ANALYZE, this is not a problem. If you are unsure enough to be asking this question, you should definitely NOT use SET FILE/REMOVE or DELETE on any operating system generated directories or files. VAX-185 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 54.5 analyze/disk question 5 of 5 VAXFAM::TKENNEDY 7 lines 30-APR-1987 17:01 -< Thanks for help >- ---------------------------------------------------------------- Thank you all for your replies. I DO have VAX RSX, but had the problem before i installed it. It seems to be the non-problem mentioned in the second reply of a system-generated alias. Thanks again, Terry Kennedy ================================================================ Note 56.0 VMS Mail 5 replies VAXFAM::JOSBORN 4 lines 29-APR-1987 16:11 ---------------------------------------------------------------- I want to write a foreign mail handler for VMS Mail. By that I mean the handler "MYHANDLER" that is called when I send mail to user MYHANDLER%USERNAME. Has anyone done this or seen any documentation on how to it? ================================================================ Note 56.1 VMS Mail 1 of 5 VAXFAM::LKILGALLEN "Larry Kilgallen" 9 lines 29-APR-1987 16:53 -< Several have done it but no documentation >- ---------------------------------------------------------------- Joiner Associates have done it for their JNET product and Kevin Carosso has, I believe, done it for the VVCP/UUCPVMS/name- of-the-week project Jamie Hanrahan is organizing. I have looked at the fiche some years ago and it did not seem like it would be hard. Beware, however, of the notice that DEC gave Monday of callable mail in some future version of VMS. The environment might change. On the other hand, it might get easier. At present I believe DEC supplies no documentation on this capability. VAX-186 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 56.2 VMS Mail 2 of 5 VAXFAM::WBAKER 4 lines 29-APR-1987 17:29 -< MRMANAGER >- ---------------------------------------------------------------- buy message router. then use logicals to do the function to send through the MRGATE. you may still have to write a program for taking the data from MRGATE to your other system, but at least how to do it is documented in the manuals. ================================================================ Note 56.3 VMS Mail 3 of 5 VAXFAM::LKILGALLEN "Larry Kilgallen" 2 lines 29-APR-1987 17:32 -< Callable message router costs extra (Q*733) >- ---------------------------------------------------------------- And the cost might be judged considerable compared with writing a VMSmail interface yourself. ================================================================ Note 56.4 VMS Mail 4 of 5 VAXFAM::KCAROSSO "Kevin Carosso" 29 lines 29-APR-1987 17:39 -< VMS foreign mail interface >- ---------------------------------------------------------------- I wrote one of these a long time ago and put it on the Spring '85 VAX SIG tape. It's called UUCP_MAIL. No, UUCP is not provided on the tape, that's AT&T code. What I did put on the tape was my foreign protocol code. You can take my code as a model and write your own handler for your own network (or whatever). I have put an updated version of UUCP_MAIL as a submission to this VAX SIG tape. Coming soon to a LUG near you! Another approach is to get a more general mail system to which protocol dependent modules can be more easily added and add your code to that. PMDF is a general-purpose RFC-822 compliant mail system for VAX/VMS. PMDF uses the foreign protocol hook into MAIL since MAIL remains the user interface. With PMDF, however, the MAIL hack is done for you and you add your code to the other end. PMDF is distributed for the cost of distribution with full sources (in Pascal). Contact Ned Freed at Harvey Mudd College (714)621-8006 for more information. By the way, PMDF already VAX-187 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville supports a wide range of mail delivery mechanisms: store-and-forward DECnet, general SMTP, BITNET using Jnet, SMTP on TCP/IP using Wollongong, Tektronix, CMU, and others, telephone dial-up (for el-cheapo network mail), and more so you may not even have to add a module. There is some effort under way to link together interested PMDF sites into a large dial-up network similar to what the UNIX world has had with UUCP. Sites with access to other large networks could serve as gateways to the rest of the world. /Kevin Carosso kvc@engvax.scg.hac.com Hughes Aircraft Co. ================================================================ Note 56.5 VMS Mail 5 of 5 VAXFAM::RCRITZ "Richard Critz" 2 lines 30-APR-1987 10:41 ---------------------------------------------------------------- Refer to the futures slides from Monday. There is some help indicated therein. ================================================================ Note 57.0 LOCK MANAGER, re: value block 3 replies VAXFAM::FKRIKORIAN 69 lines 29-APR-1987 16:30 ---------------------------------------------------------------- The VMS lock manager provides a feature called the "value block". This feature allows a value to be associated with a lock when that lock is released from certain modes, so that the value can be obtained by another process on acquisition or conversion of the lock. In a sense, the value acts as a very small (sixteen byte) shared memory area, access to which is controlled by the lock it belongs to. When a process owns a lock in certain modes, that process is considered to be in control of the lock value. This is the case because the lock value will be modified when the process releases the lock. If the process terminates abnormally, due to a system crash (not $delprc), the return status will indicate that the last lock owner did not set a value. It appears from experimentation that the lock value in this case remains at its previously-set value. VAX-188 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville When a processor fails, the return status should indicate any locks that were owned in exclusive mode by processes on that processor, and the lock values should remain at their previously-set values. However, experimentation shows that although the lock values do remain at their previously-set values, the return status indicates locks other than those expected, and seems to indicate locks owned in ANY mode by the failed processor that are not also owned in a lock-setting mode (exclusive or protected write) by another processor. QUESTIONS 1. What value is in the lock value for a lock if a process terminates while holding the lock in exclusive mode? Is it guaranteed to be the last value set, or might it be some earlier value, or might it be garbage? (We expect it to be the last value set). 2. Under what circumstances is the invalid-lock-value warning status returned when a processor fails? (We expect it only if the lock was held in a lock-setting mode by a process on the failed processor). 3. When the warning flag is set, what is in the lock value? (We expect the lock value to be guaranteed to contain the last value that was successfully set, even if it was set by the failed processor while holding the lock and the lock has not been released in the meantime). 4. When the warning is not returned, what is in the lock value? (We expect the same guarantee as in 3.). MOTIVATION We are using lock values to protect disk blocks, roughly as described in the system services documentation. The lock value contains data that describes the current state of the block. Typically, a processor changes the lock value to indicate that it has done something to the block, and then at some later point performs some other different disk operation (such as forcing out a log) to indicate that those changes are committed. We need a guarantee that no process on a different CPU can see the subsequent disk operation without also seeing the lock value change. For example, suppose that the following sequence of events occurs (the lock value initially contains 1, and the disk block initially contains "X"): VAX-189 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville set lock value to 2 write "Y" in a disk block We need to know that if the processor that did this fails, and if some other processor executes: read disk block read lock value that there is no way the processor can see "Y" in the disk block and 1 in the lock value. ================================================================ Note 57.3 LOCK MANAGER, re: value block 3 of 3 STAR::SNAMAN 113 lines 30-APR-1987 22:01 ---------------------------------------------------------------- QUESTIONS 1. What value is in the lock value for a lock if a process terminates while holding the lock in exclusive mode? Is it guaranteed to be the last value set, or might it be some earlier value, or might it be garbage? (We expect it to be the last value set). << While it is always a previous value, it is not necessarily the last value set. It is the most recent value that can be found among the surviving nodes. >> 2. Under what circumstances is the invalid-lock-value warning status returned when a processor fails? (We expect it only if the lock was held in a lock-setting mode by a process on the failed processor). << Currently the invalid value block status is returned whenever no locks are present with a mode more restrictive than CR mode during the final stages of the lock database reconstruction. VAX-190 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Think of it this way. If, after the database is rebuilt but before any new locks are granted, there exists a lock held at CW, PR, PW, or EX mode, then there could not have been another lock held at PW or EX mode before the lock database was rebuild which is no longer present. This is because the lock would have been incompatible with the locks presently granted on the resource. Therefore the value block is considered valid. >> 3. When the warning flag is set, what is in the lock value? (We expect the lock value to be guaranteed to contain the last value that was successfully set, even if it was set by the failed processor while holding the lock and the lock has not been released in the meantime). << It will be some previous value but may *not* be the last value set. A sequence number is used to identify the latest copy of the value block that can be found and it will be the one that is used. The latest copy available after a cluster state transition may not be the most recent value set since the processor that contains the most recent value set may no long be a member of the VAXcluster. >> 4. When the warning is not returned, what is in the lock value? (We expect the same guarantee as in 3.). << If the invalid status is not returned, then the value will be the latest value set. Note, that this is true only for as long as there are locks on the resource. If all locks on the resource are dequeued, then the value contained in the value block is gone. In this case, the value returned for the first subsequent $ENQ will be zero indicating that this is a new resource as far as the lock manager is concerned and the status will be valid. >> VAX-191 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville MOTIVATION We are using lock values to protect disk blocks, roughly as described in the system services documentation. The lock value contains data that describes the current state of the block. Typically, a processor changes the lock value to indicate that it has done something to the block, and then at some later point performs some other different disk operation (such as forcing out a log) to indicate that those changes are committed. We need a guarantee that no process on a different CPU can see the subsequent disk operation without also seeing the lock value change. For example, suppose that the following sequence of events occurs (the lock value initially contains 1, and the disk block initially contains "X"): set lock value to 2 write "Y" in a disk block We need to know that if the processor that did this fails, and if some other processor executes: read disk block read lock value that there is no way the processor can see "Y" in the disk block and 1 in the lock value. << This is the case assuming that the $DEQ or convert operation that updates the value block is completed before the write operation is initiated and invalid value block status is not returned on the next $ENQ before the read operation. If invalid value block status is returned then the block may have been updated and the value block may contain a 1 so the value block contents should be ignored. Have you considered the possibility that the value block may be updated but the disk block may not? For example the node or process may fail after the $DEQ that changes the value block and before the write that updates the disk block completes. VAX-192 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville >> Sandy Snaman VAX/VMS Development Group ================================================================ Note 58.0 RENAME BUG 7 replies VAXFAM::RSUPPLEE "Rick Supplee" 17 lines 29-APR-1987 17:18 ---------------------------------------------------------------- Are any of the DEC developers aware of the RENAME/SINCE=TODAY having problems with files that are currently open. I first found this problem when learning SPM. I was using an example SPM program out of SYS$EXAMPLES I believe it was SPM$MANAGER. It does a RENAME/SINCE=TODAY *.CAP;* *.DAT;* However; an error occurs on this command since there is a .CAP file currently open for data collection. I have done a similar check on this version of VMS Y4.6 and it still occurs. The system I have at home is VMS 4.4. I am able to do a DIRECTORY/SINCE=TODAY *.CAP why not RENAME/SINCE=TODAY? Is someone aware and will there be a fix?????????????????????? I found this out just before coming to DECUS so I have not called TSC. Thanks, Rick ================================================================ Note 58.1 RENAME BUG 1 of 7 N8700::SLIONEL "Steve Lionel - DEC -" 7 lines 30-APR-1987 11:26 ---------------------------------------------------------------- DIRECTORY uses a special interface to get the file header info without opening the file - RENAME uses RMS to open the file if you ask it to select files based on creation date, etc. This is an interesting suggestion - thanks. Steve Lionel VAX-193 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 58.2 RENAME BUG 2 of 7 VAXFAM::RSUPPLEE "Rick Supplee" 4 lines 30-APR-1987 12:56 -< TYPO IN 58.0 >- ---------------------------------------------------------------- The qualifier that is really being used is not /SINCE=TODAY it is /BEFORE=TODAY. I was in a hurry when I typed in 58.0. The problem is still the same in that a fatal error message is given because of the way RENAME must look at files. Sorry for the typo. ================================================================ Note 58.3 RENAME BUG 3 of 7 N8700::RLGRAHAM "Bob Graham, Dow Che" 5 lines 30-APR-1987 13:19 -< a question about how DIRECTORY does it >- ---------------------------------------------------------------- In response to 58.1 --- Is the special interface that DIRECTORY uses merely undocumented, or is it { mildly, moderately, extremely } dangerous to attempt to use? ================================================================ Note 58.4 RENAME BUG 4 of 7 VAXFAM::SLIONEL "Steve Lionel - DEC" 13 lines 30-APR-1987 13:36 -< QIOs? >- ---------------------------------------------------------------- Re: .3 I am not sure, but I think it is merely $QIOs to the XQP to read the file header. You could look at the fiche for details, or perhaps someone else who has looked at this can comment. Actually, there have been suggestions in the past to give RMS the ability to retrieve information about a file without opening it. (For example, $DISPLAY.) If this capability were added, many VMS utilities and user programs could easily obtain this information safely. I think I'll ask for this again - I can think of some of my own code that would benefit. VAX-194 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Steve ================================================================ Note 58.5 RENAME BUG 5 of 7 STAR::PILANT "L. Mark Pilant" 11 lines 30-APR-1987 15:25 -< Who's afraid of the Q-I-O...Q-I-O >- ---------------------------------------------------------------- RE: How DIRECTORY gets information DIRECTORY uses QIOs directly to the ACP to get information about a file available to the local node (disk or tape). This interface is documented in part 1 of the I/O User's Guide. RMS is used to get the information when the file is on a remote node, or it is necessary to obtain information about the file's structure (e.g., indexed files). - Mark, VMS Development (Also author of DIRECTORY) ================================================================ Note 58.6 RENAME BUG 6 of 7 VAXFAM::PSCOPELLITI "Pat Scopelliti" 5 lines 30-APR-1987 16:37 -< wildcard renames don't work much either >- ---------------------------------------------------------------- Here's another unusual rename problem... if you have a bunch of files called t1.t, t2.t, t3.t and you want to change the name to s1.t, s2.t, s3.t you might be tempted to use the command RENAME T%.T S%.T Well.. it don't do it.. DEC.. HELP? VAX-195 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 58.7 RENAME BUG 7 of 7 VAXFAM::PSCOPELLITI "Pat Scopelliti" 1 line 30-APR-1987 16:39 -< Others do it... >- ---------------------------------------------------------------- by the by.. RSTS/E does it real well... ================================================================ Note 59.0 LNM$GROUP_xxxxxx Creation 7 replies VAXFAM::LDILL 12 lines 30-APR-1987 09:25 ---------------------------------------------------------------- How can I create group logical name tables (normally during system startup) from the startup process. In other words, without having to submit batch jobs, or run detached processes under different UIC's, which will then have the system create the tables for me. This method works OK, but it's pretty slow. I can create tables using system services, and name them LNM$GROUP_xxxxxx, with all the proper attributes, but not with the "group" attribute. Any ideas where to look for documentation on the data structures used for the tables? Leon Dill Hennepin County, Minnesota ================================================================ Note 59.2 LNM$GROUP_xxxxxx Creation 2 of 7 N8700::SSIMM 19 lines 30-APR-1987 11:01 -< Here's How.. >- ---------------------------------------------------------------- Something along the lines of the following will work: RUN/UIC=[x,y]/INPUT=logicals-to-be-defined.COM - /OUTPUT=log-file.LOG SYS$SYSTEM:LOGINOUT This fires up a detached process (running DCL) that reads lines ("$ DEFINE xyz pdq") from the input file. Since it is running under the correct UIC (for the group) the group table will be created for you if it does not exist. This is one of the mechanisms described in the release notes. VAX-196 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Do NOT use the SET UIC method... It won't work right with the new logical name tables... Stephen Simm GE Consulting Services 518-458-2409 ================================================================ Note 59.3 LNM$GROUP_xxxxxx Creation 3 of 7 N8700::APOTTER "Andrew Potter, Roche" 7 lines 30-APR-1987 11:13 -< Defining Group logical names >- ---------------------------------------------------------------- You can define logical names into ANY group in SYSTARTUP .com by using the following syntax: $ define/table=LNM$GROUP_XXXXXX LOG_NAME EQUIV_NAME works like a champ for me. ================================================================ Note 59.4 LNM$GROUP_xxxxxx Creation 4 of 7 N8700::SSIMM "Stephen Simm, GE Cons" 10 lines 30-APR-1987 11:21 -< Is the Table There? >- ---------------------------------------------------------------- I tried the /Table=LNM$GROUP_XXXXXX trick under previous versions and it didn't work if the logical name table hadn't already been created. (LOGINOUT creates it when it can't find it) This might have changed in a more recent release -- I'm using the RUN method from the release notes and haven't bothered to go back and check. Stephen Simm 518-458-2409 VAX-197 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 59.5 LNM$GROUP_xxxxxx Creation 5 of 7 VAXFAM::GISETT 6 lines 30-APR-1987 12:52 -< DEC WON'T LET YOU DO IT >- ---------------------------------------------------------------- I looked into the fiche about a year ago attempting to do something similar. I found that the code to set the (I believe) GROUP attribute was modified to specifically not allow users to do it themselves. This was in the create logical name table system service. I think LOGINOUT uses the brute force method of building and linking the command blocks without calling a system service. ================================================================ Note 59.6 LNM$GROUP_xxxxxx Creation 6 of 7 VAXFAM::APOTTER "Andrew Potter, Roc" 18 lines 30-APR-1987 12:56 -< More on group tables... >- ---------------------------------------------------------------- If a process has not been created in the target group yet you may have to create the group table Since the Group table is a kernel mode table you will need to do this with the $CRELNT system service. Basically the call to $CRELNT would do the following dcl command If it existed $create/name_table/kernel/own=[xxxxxx,177777] - /table=lnm$system_directory /protection=(s:rwed,o:r,g:r,w) - lnm$group_xxxxxx where xxxxxx is the Octal group. This is not hard to do with $CRELNT if you like item lists.... ;-) VAX-198 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 59.7 LNM$GROUP_xxxxxx Creation 7 of 7 N8700::LKILGALLEN "Larry Kilgallen" 6 lines 30-APR-1987 13:44 -< Creating from SYSTARTUP >- ---------------------------------------------------------------- In order to create a kernel mode group logical name table, use the RUN/UIC=[55,0] SYS$SYSTEM:LOGINOUT.EXE mechanism, but use NLA0: for /INPUT. After ensuring a slight delay (possibly by similar commands for other groups) use DEFINE/TABLE=LNM$GROUP_etc directly from SYSTARTUP.COM. This avoids having your definitions scattered around various command files. ================================================================ Note 60.0 Rdb V2.2 and RDML/PASCAL problems 2 replies VAXFAM::DCRIM 33 lines 30-APR-1987 09:36 ---------------------------------------------------------------- Having lots of fun with Rdb V2.2 RDML for PASCAL? (version 4.5 VMS and v3.5 of PASCAL) Problems you may encounter: 1) RDML will occasionally generate "BEGIN" blocks with no "END". 2) The "RDB$PORT_FIELD_n" (where n is numeric) pascal data type is incorrect and will cause a run time failure in most cases. You must prefix the range -32768..32767 with the [WORD] attribute. Yes, this does mean editing all of the .PAS files from RDML!! 3) RDML will generate "....data type conflict..." when a Rdb field of type text is used in an RSE with with a PASCAL character constant. This was not true with RDBPASCAL. 4) If multiple PASCAL modules invoke the database, one must invoke with the [global] attribute all others must invoke with the [EXTERNAL] attribute. RDBPASCAL allowed [external] for all modules. I know this should be obvious!!! VAX-199 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville 5) You must also change your linking procedures if you use RDML. See the documentation. V2.0.2.1 and 2.2 of Rdb: If you have been encountering problems with the .SNP file growth.... like it grows until the disk is full...you might want to go hammer on the dm people in suite 6536. The apparent work-around is to define the % extend value in the multi-volume clause for the snapshot file to 1. I would appreciate knowing how many of you out there are experiencing these problems. Considering the problems with V2.2, we do not believe it is worth installing? ================================================================ Note 60.1 Rdb V2.2 and RDML/PASCAL problems 1 of 2 VAXFAM::IMUSSER 30 lines 30-APR-1987 10:40 -< more on rdb 2.2/pascal use >- ---------------------------------------------------------------- Interesting comments....I was about to go back and install the 2.2 that has been on my desk for a month, now I am not so sure. I was anxious to start using the Db$Key which heretofore has not worked in the imbedded rdo-pascal environment. Use of that feature would not be as necessary if I could do something like RecVar := S.* where Recvar is typed off of an inherited cdd type and s in the context var. (is this possible?). What I now have in the code is needless duplication of simple moves from the context var into local variables. If the db$key worked, I could at least do this in a separate routine. Also, what with the lse support of compile/review, where does this leave the pre-compilers? No one wants to look at the generated pascal when repairing the bug. This is already a problem for us in our non rdb-precompiler programs since we have our own precompiler for code insertion which is run before pascal. Tool integration is nice, but when it is too tightly coupled, there is less and less room for user customization/alternate methods using the base level tools separately -- another item in this vein is the lse support for cms replace/fetch/insert/.... These are functions we have protections built around and are not to be used off the cuff. VAX-200 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Anyway, we only have one small rdb/pascal application (using it as a test bed). I'd be interested in any other problems you have with the 2.2 stuff, especially the pre-compiler. Ivan E. Musser Shell Oil WRC D-2162 P.O. Box 1380 Houston Tx 77001 713-493-8635 ================================================================ Note 60.2 Rdb V2.2 and RDML/PASCAL problems 2 of 2 VAXFAM::KRODWELL 51 lines 30-APR-1987 16:03 -< Thank you... >- ---------------------------------------------------------------- Thank you for your concerns regarding RDML. If you have further suggestions/concerns please come to the DMS SIG suite 6536. The problems reported are known or have been noted and will be addressed in a future version. The RDML precompiler's intent is to flush out the problems which exist in the old Pascal precompiler. There are problems with the new precompiler which we are actively resolving. Rdbpascal and RDML were both shipped with the V2.2 kit for this reason. It is recommended that you do make the transition to RDML if possible. Some responses to .0 below. RE: .1 of this note, RDB$DB_KEY support does exist in Rdbpascal and RDML. Development is aware of the complex problem surrounding LSE's support of COMPILE/REVIEW. There is currently no support in any of the precompilers for the use of VARIABLE := S.* neither is there a GET statement. Please continue to inform us of any other problems that you encounter through SPRS. Karen Rodwell Rdb/VMS Development RE: .0 VAX-201 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville 1) There is a problem with V2.2 RDML BEGIN/END where start_trans has an on error clause. 2) There are known problems with WORD datatypes in RDML. 3) RDML will generate "....data type conflict..." when a Rdb field of type text is used in an RSE with with a PASCAL character constant. This was not true with RDBPASCAL Not sure on this one, will be reported to engineering. 4) If multiple PASCAL modules invoke the database, one must invoke with the [global] attribute all others must invoke with the [EXTERNAL] attribute. RDBPASCAL allowed [external] for all modules. I know this should be obvious!!! This is true of all preprocessors...Handles required for multiple dbs. 5) You must also change your linking procedures if you use RDML. See the documentation. The linking procedures have been changed because the library shouldn't be in starlet as was formerly done (Rdbpascal). ================================================================ Note 61.0 FILLM? 3 replies VAXFAM::MCHALENBURG 7 lines 30-APR-1987 09:57 ---------------------------------------------------------------- We are having some confusion with the FILLM quota. We recently ran a BASIC program that opened a file on channel 99. This was the first file it opened explicitly. When the FILLM quota was set to 20, we received the error "Cannot open file" or some such. When the FILLM was set to 40, everything worked fine. VAX-202 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 61.1 FILLM? 1 of 3 VAXFAM::KRANDELL 9 lines 30-APR-1987 10:54 -< HOPE TO SOLVE YOUR CONFUSION >- ---------------------------------------------------------------- FILLM DOES NOT HAVE ANYTHING TO DO WITH THE UNIT NUMBER YOU USED TO OPEN THE FILE. YOU HAVE TO REMEMBER THAT THE FILLM IS A MAXIMUM LIMIT ON THE TOTAL NUMBER OF FILES YOU CAN HAVE OPEN AT ONCE, BUT THIS NUMBER INCLUDES ALL OF THE RUN/TIME LIBRARY FILES, SHAREABLE IMAGES, ETC. THAT ARE NEEDED TO RUN YOUR PROGRAM (COUNTS AS ONE TOO). YOU COULD LOG INTO ANOTHER TERMINAL AND USE SHOW PROCESS/RMS TO SHOW ALL OF THE FILES YOU HAVE OPEN, BUT SINCE YOU HAVE SOLVED YOUR PROBLEM ANYWAY, IT MIGHT BE A WASTE OF TIME. ================================================================ Note 61.2 FILLM? 2 of 3 N8700::SSIMM 9 lines 30-APR-1987 11:03 -< CHANNELCNT >- ---------------------------------------------------------------- The "special" SYSGEN parameter CHANNELCNT? controls the number of channels available to a process. It defaults to 128. Some of the channels are reserved for use by the LIB$GET_LUN modules. MC SYSGEN SHOW/SPECIAL ================================================================ Note 61.3 FILLM? 3 of 3 VAXFAM::SLIONEL "Steve Lionel - DEC" 19 lines 30-APR-1987 13:42 -< LIB$GET_LUN has nothing to do with channels >- ---------------------------------------------------------------- Re: .2 I'm sorry, but the comment about channels reserved to LIB$GET_LUN is incorrect. LIB$GET_LUN simply arbitrates use of a set of BASIC and FORTRAN file unit numbers, which, as stated in .1, are not directly related to channels. VAX-203 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Things that consume FILLM are shareable images that are not installed /OPEN, message sections (ditto), your SYS$INPUT, SYS$OUTPUT and SYS$ERROR files, etc. A FILLM of 20 is low - 40-50 is more appropriate. You can reduce dependence on FILLM by installing images (both shareable and executable) as /OPEN (in addition to /SHARED if you want). CHANNELCNT is indeed a limit on the number of channels a process can have, but you normally don't need to change it. Only if you boost FILLM very far might you need to adjust CHANNELCNT. Steve ================================================================ Note 66.0 Large WSMAX a problem? 3 replies VAXFAM::TGARDNER 25 lines 30-APR-1987 10:59 ---------------------------------------------------------------- I recently acquired a 3rd party application program that wants a WSQUOTA of 8000, and PGFLQUO of at least 12000. With a 4.5Mb 11/750, such quotas are not possible. I am willing to put up with a lot of paging (the application runs monthly at 2am) but am not sure how to configure other things. I set WSQUO to about 2000 and WSEXT of 6300 (the best I could do, given the hardware & software). Of course, this meant setting WSMAX to 6300 as well. I heard in a session this week, that setting WSMAX to this (relatively) high percentage of the total memory on the system may present some problems with VMS (specifically when the job is swapped, VMS might extend pool and create a situation where the job could not be swapped back in). Presuming that this exact situation is not going to occur (since other jobs won't be on the system at the same time) am I likely to see other problems with VMS. Because I want to be able to run this particular application in batch after hours, I don't want to have to boot up a specially configured system, as this would once again require me to be on site to do the job, thus losing most of the benefit of the batch job in the first place. Thanks in advance. Tim Gardner/Genigraphics Corp. VAX-204 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 66.1 Large WSMAX a problem? 1 of 3 VAXFAM::CRHODE "Chris Rhode (Lockhee" 9 lines 30-APR-1987 11:22 -< Don't know about WSMAX but watch out for PGFLQUO >- ---------------------------------------------------------------- I can't offer any comments on WSMAX, but if the application program allocates large chunks of dynamic memory, you may need that large PGFLQUO. This is because PGFLQUO is an upper limit on the total amount of dynamic memory you can ask for, and because when you grab (say) a 5K-page chunk of dynamic memory, it socks it into the page file (it doesn't get paged into physical memory until actually referenced). This was extremely frustrating for me because we have LOTS of physical memory and not much disk space for large page files. ================================================================ Note 66.2 Large WSMAX a problem? 2 of 3 VAXFAM::SLIONEL "Steve Lionel - DEC" 10 lines 30-APR-1987 13:46 ---------------------------------------------------------------- Another problem is that if you set WSMAX too high, you may prevent VMS from expanding its pooled memory - you'll get the error "pool expansion failure" if it wants to allocate more LRPs, etc. I believe the rule is to keep WSMAX lower than the number of fluid pages on your system (those not "allocated to VMS" in the SHOW MEMORY display). (You should allow at least 200 pages difference - there's another parameter in the calculation that I don't remember offhand.) Steve VAX-205 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 66.3 Large WSMAX a problem? 3 of 3 VAXFAM::VMSDEV "John C. Hallyburton" 19 lines 30-APR-1987 18:18 -< No problem -- go for it! >- ---------------------------------------------------------------- .. I heard in a session this week, that setting WSMAX to this (relatively) high percentage of the total memory on the system may present some problems with VMS (specifically when the job is swapped, VMS might extend pool and create a situation where the job could not be swapped back in). I was the speaker for that session, and what you heard was not what I said. (At least, not what I *thought* I said). In order to _PREVENT_ that kind of deadlock, sometimes VMS will not allow pool to expand. Occasionally, customers will set WSMAX large not realizing that pool expansion failure can result, hence AUTOGEN's up-and-coming warning message if WSMAX is set maybe big enough to where this could possibly be a problem. Bottom line: at 2AM, pool expansion failure is probably the least of your concerns. There should be no problem with a large WSMAX. John ================================================================ Note 67.0 FORTRAN or C File Copy 5 replies VAXFAM::SFEHL "Appaloosa Breeder" 13 lines 30-APR-1987 11:05 ---------------------------------------------------------------- I would like to be able to make a copy of an existing file from FORTRAN-77 or VAX C. This is to do away with some lengthy and slow DCL procedures. The problem I always run into, is that the file attributes are not identical. Since we have a working solution to our problem (the DCL procedure) I have not spent a lot of time with this. I am trying to copy the following types of files: a. ASCII files (e.g. .FOR, .C, .MAR). b. Object files. c. Executable files. VAX-206 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville What is the fundamental principal I am missing? ================================================================ Note 67.1 FORTRAN or C File Copy 1 of 5 N8700::SSIMM "Stephen Simm, GE Cons" 13 lines 30-APR-1987 11:11 -< Brute Force Approach >- ---------------------------------------------------------------- The brute force method is to spawn (or LIB$SPAWN) the COPY command. A refined version of this kludge has a spawned process reading from a mailbox that you write successive COPY commands into. (Until you write in an EOF) As for writing an actual COPY routine in your program, it isn't that difficult -- you've already run into the biggest hitch. (The file attribute differences...) Stephen Simm 518-458-2409 ================================================================ Note 67.5 FORTRAN or C File Copy 5 of 5 VAXFAM::SLIONEL "Steve Lionel - DEC " 3 lines 30-APR-1987 13:47 ---------------------------------------------------------------- I think callable CONVERT will do what you want. See the "Utility Routines Reference Manual". Steve ================================================================ Note 70.0 Phantom files 1 reply N8700::CROTHAUSER 12 lines 30-APR-1987 11:30 ---------------------------------------------------------------- Phantom files A user running a program on a MicroVAX II creates scratch files on a VAXcluster member using DECnet FAL links. The user types ctrl/c to abort (really suspends!) the program. VAX-207 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville We reboot the VAXcluster member after some system work. A SHOW DEVICE indicates that the scratch files are still taking up disk space. Analyze/disk/repair lists the files without a version numbers. Why did this happen? How can we get rid of the scratch files? ================================================================ Note 70.1 Phantom files 1 of 1 VAXFAM::PCLAYTON 8 lines 30-APR-1987 13:20 -< Reclaim File Space >- ---------------------------------------------------------------- Run ANAL/DISK/REPAIR to get the files into the SYSLOST directory and then delete them. Paul D. Clayton TSO Financial Corp Manager Of Systems :-) ================================================================ Note 71.0 WIN software in cpu loop 3 replies VAXFAM::RGROMAN "Bob Groman, Woods " 12 lines 30-APR-1987 13:01 ---------------------------------------------------------------- We had troubles with the WIN software from The Wollongong Group hanging when a user set host to another VAX, started up TELNET and the original process was killed. (I think I may have this backwards but the idea was that the node running TELNET didn't realize it should go away.) The result was a cpu spinning its wheels, racking up lots of DIO's and BIO's. Has anyone else seen this problem? Anyone have a solution? Thanks Bob Groman Woods Hole Oceanographic Woods Hole, MA 02543 617/548-1400 x2409 VAX-208 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 71.1 WIN software in cpu loop 1 of 3 STAR::KENNEY 29 lines 30-APR-1987 16:06 -< You need to get WIN/VX fixed >- ---------------------------------------------------------------- This is known and can be seen with other applications besides WIN/VX. Presently the only known fix is to teach the application not to do this. Not knowing exactly how WIN/VX does TELNET I will make some guesses about what is happening and suggest a fix. 1) WIN/VX does a asynchronous single character read from the keyboard most likely in passthru mode. 2) In the AST routine it requeues the read and then passes the data just read out over the network. It is a sure bet that they do no error checking. If they do this they have a very tight loop that keep the PRIORITY relatively high. Now what happened is that when the DECnet link collapsed it goes through the following steps: 1) Delivers all HANGUP AST's especially the critical one for DCL to log you out. 2) If completes all current I/O's with a reason of SS$_HANGUP. I think that is the status used. It absolutely uses an error status. 3) It will complete all QIO's queued to the network Pseudo Terminal with this same error. What you need to get WIN/VX to do is a hangup AST on the DECnet line. When it sees this AST is should shut down operations and exit ideally with a reason of SS$_HANGUP. Forrest Kenney VMS Development VAX-209 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 71.2 WIN software in cpu loop 2 of 3 STAR::KENNEY 4 lines 30-APR-1987 17:45 -< Some other who do this >- ---------------------------------------------------------------- Just remembered some code that does this. 1) KERMIT sorry do not know the version number 2) SET HOST/DTE all V4.x versions ================================================================ Note 71.3 WIN software in cpu loop 3 of 3 VAXFAM::RGROMAN "Bob Groman, Woods H" 3 lines 30-APR-1987 17:56 -< Thanks for reply >- ---------------------------------------------------------------- Thanks for the information. We will pass it back to The Wollongong Group. ================================================================ Note 75.0 BACKUP directory structures 2 replies VAXFAM::SFEHL "Appaloosa Breeder" 17 lines 30-APR-1987 14:34 ---------------------------------------------------------------- Can the BACKUP utility backup files in one type of directory structure and restore to a different structure? This is what I want to do: Source Destination ---------------------------- ----------------------- [support.cel.300000.renold1] [cms.cel.300000.renold1] At present we copy all the files from the source directories to a dummy that looks like the destination directories we need. This is slow and uses disk space unnecessarily. The documentation leads me to believe that what I want to do can only be done the way I am doing it. VAX-210 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 75.1 BACKUP directory structures 1 of 2 VAXFAM::CGETZ 11 lines 30-APR-1987 15:35 -< Just a try >- ---------------------------------------------------------------- Using a user level directory [MYDIR] and a subdirectory of [.JUNK] and other nested subdirectories I just tried this: BACKUP [.JUNK...] [.TEST...] This renamed the directory JUNK to TEST while all other names were preserve. The command was issued at the [MYDIR] level. I don't know if this works at the [0,0] index level. ================================================================ Note 75.2 BACKUP directory structures 2 of 2 STAR::STILES "When in doubt, hack." 47 lines 30-APR-1987 16:12 -< BACKUP can copy directories >- ---------------------------------------------------------------- re: < Note 75.0 by VAXFAM::SFEHL "Appaloosa Breeder" > Can the BACKUP utility backup files in one type of directory structure and restore to a different structure? This is what I want to do: Source Destination ---------------------------- ----------------------- [support.cel.300000.renold1] [cms.cel.300000.renold1] Yes. BACKUP, when running in disk-to-disk mode without a /SAVESET qualifier, is essentially a COPY command that understands directory hierarchy. Example 1: $ BACKUP indisk:[support.cel.300000.renold1] - $_ outdisk:[cms.cel.300000.renold1] will copy the bottom level directory, as you would expect, and create the necessary upper directories as it goes. VAX-211 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Example 2: $ BACKUP indisk:[support...] - $_ outdisk:[cms...] will reproduce the entire directory tree from [support...] to [cms...] Example 3: $ BACKUP indisk:[support...] - $_ outdisk:[cms] will copy all the files from the entire directory tree [support...] into the single directory [cms], not preserving the directory structure (in other words, like COPY would do). Various extra subdirectory specifications in the input or output specifications should work. This capability is described in the BACKUP Utility Reference Manual. If the documentation is not clear, please mail in the postpaid Reader's Comment form at the back of the manual and describe how it could be better. - Mark Stiles VMS Development ================================================================ Note 76.0 COMPAQ 286 keypad redefine 1 reply VAXFAM::SFEHL "Appaloosa Breeder" 14 lines 30-APR-1987 14:41 ---------------------------------------------------------------- I use a COMPAQ DESKPRO 286 with DECnet-DOS. What I would like to do is redefine the keypad keys KP2, KP4, KP6, KP8 to act as the arrow keys for DCL. I have tried to use the DEFINE KEY command with disappointing results. Instead of recalling the last command when I hit the up-arrow (KP8) the cursor moved up the CRT one row. Now this may have applications but that is not what I want. VAX-212 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Does anyone know how I can solve this problem? The COMPAQ supposedly looks like a VT102 to the VAX and I use a SET TERM /DEVICE=VT102. ================================================================ Note 76.1 COMPAQ 286 keypad redefine 1 of 1 STAR::KENNEY 15 lines 30-APR-1987 15:44 -< Cannot be done with define key >- ---------------------------------------------------------------- You need to teach the COMPAQ to send the correct escape sequences DEFINE key happens at much to high a level to do command recall and editing. The sequences you need are: Arrow direction Sequence .break up esc[A down esc[B right esc[C left esc[D Forrest Kenney VMS Development ================================================================ Note 77.0 TK50's creating MUA4224: devices 3 replies VAXFAM::BMCCORMICK "Bob McCormick" 13 lines 30-APR-1987 16:05 ---------------------------------------------------------------- Oh about a year and a half ago (maybe longer?!) when I got my first (one of the first) TK50 tape drives I found a problem that resulted in a MUA4224: device being created. I suspect that its causes by things like pressing the little red button in on the tape drive when you're trying to do something like a MOUNT... I reported it back to Colorado, but have heard nothing since then. I know of many of our systems that this still seems to occur on. Anyone else seen this (annoying) bug? VAX-213 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville OH, buy the way, when we see it we reboot the system. We've had weird things happen using the tape drive after you see the creation of the MUA4224: device. ================================================================ Note 77.1 TK50's creating MUA4224: devices 1 of 3 STAR::STILES "When in doubt, hack." 6 lines 30-APR-1987 16:17 -< known hardware problem >- ---------------------------------------------------------------- This is a problem in older versions of the TK50 microcode. Have Field Service update the TK50 to the current ECO levels and this problem should disappear. - Mark Stiles VMS Development ================================================================ Note 77.3 TK50's creating MUA4224: devices 3 of 3 VAXFAM::TKENNEDY 15 lines 30-APR-1987 17:14 -< How to get the update >- ---------------------------------------------------------------- The problem is caused by an attempt to mount or initialize the device while the drive is still trying to get it's act together. New EPROMS on the controller is supposed to fix it. The latest and greatest EPROM part numbers are 23-208E5-00 and 23-209E5-00 (set of 2), and a board with these ROMS in it should be at Rev. E1 or later. Let me clarify - changing the ROMS IS the E1 rev, and is not a REQUIREMENT for rev E1 to install ROMS. This should be a freebie if you have a DEC maintenance contract, otherwise call DEC Unlisted Parts at 1-603-884-5001 to find out the 'EQ kit' number so you can do it yourself. Terry Kennedy St. Peter's College VAX-214 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville ================================================================ Note 78.0 XQP cache usage - how can you tell? No replies VAXFAM::WFLEISCHMAN 8 lines 30-APR-1987 17:20 ---------------------------------------------------------------- I would like to determine the amount of paged pool being used for XQP-cached data structures e.g. file headers, directory blocks, etc. I've tried running ANALYZE/SYSTEM with SHOW POOL/SUMMARY/PAGED and I don't see any specific for this. Does it just show in the category "UNKNOWN"? The non-paged pool entries show up as "VCA", but no entries for paged pool are shown as VCA. ================================================================ Note 79.0 A Plug for PAAS No replies VAXFAM::STOLLBOM "Cullen Tollbom, B" 26 lines 30-APR-1987 17:35 ---------------------------------------------------------------- From the developer of PAAS... I'm glad to see the interest in PAAS. There are a number of improvements I intend to make to the package as of yet (DECUS filled my head with some new ideas...). It has been used at my site for about a year now (after 2 complete rewrites). It took that long because it evolved as VMS evolved, and I couldn't get FULL functionality out of it until VMS V4.3. The other thing to remember is that it does NOT modify VMS, although it does modify user processes. So even if DEC still doesn't implement project accounting (grrr!) under VMS V5.0 PAAS should continue to be implementable. I have about 500 user accounts on my 11/780 and 200 projects being shared by them. It has worked out well. One thing to remember, however, is that this adds an entire new dimension to managing your VAX sites, one which I think is very powerful having used it. All you who are interested should see the sources/documentation in the DECUS SIG/Library by Fall DECUS '87. It should be a boon to those of you like me who have to conform to Government security requirements. VAX-215 PAGESWAPPER - August 1987 - Volume 9 Number 1 VMS Notes from Nashville Cullen Tollbom Battelle Pacific Northwest Laboratory P.O. Box 999 Richland, WA 99352 VAX-216 PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX System SIG Committee List VAX System SIG Committee List As of June 24, 1987 CHAIR (CORE) Susan T. Rehse Lockheed Missiles & Space Co. O/19-50, B/101, P.O. Box 3504 Sunnyvale, CA 94088-3504 VICE-CHAIR (CORE) WORKING GROUP COORDINATOR Ross Miller Online Data Processing, Inc. N 637 Hamilton Spokane, WA 99202 SYMPOSIA COORDINATOR (CORE) Jack Cundiff Horry-Georgetown Technical College P.O. Box 1966 Conway, SC 29526 COMMUNICATION COORDINATOR (CORE) Don Golden Shell Oil Company Westhollow Research Center P.O. Box 1380, Room D2158 Houston, TX 77251-1380 LIBRARIAN Joseph L. Bingham Mantech International 2320 Mill Road Alexandria, VA 22314 LUG COORDINATOR (CORE) Dave Schmidt Management Sciences Associates 5100 Centre Avenue Pittsburgh, PA 15232 VAX-217 PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX System SIG Committee List SYSTEM IMPROVEMENT REQUEST (CORE) Mark Oakley Battelle Memorial Institute 505 King Avenue Columbus, OH 43201-2669 SYMPOSIA COORDINATOR, ASSISTANT David Cossey Computer Center Union College Schenectady, NY 12308 PRE-SYMPOSIUM SEMINAR COORDINATOR HISTORIAN Jeff Jalbert J C C P.O. Box 381 Granville, OH 43023 PRE-SYMPOSIUM SEMINAR COORDINATOR (ACTING) June Baker Computer Sciences Corporation 6565 Arlington Boulevard Falls Church, VA 22046 COMMUNICATIONS, ASSISTANT David L. Wyse Professional Business Software 3680 Wyse Road Dayton, OH 45414-2539 VOLUNTEER COORDINATOR Elizabeth Bailey 222 CEB Tennessee Valley Authority Muscle Shoals, AL 35661 NEWSLETTER EDITOR Lawrence Kilgallen Box 81, MIT Station Cambridge, MA 02139-0901 VAX-218 PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX System SIG Committee List SESSION NOTES EDITOR Ken Johnson Meridien Technology Corp. P.O. Box 2006 St. Louis, MO 63011 CAMPGROUND COORDINATOR Kirk Kendrick Shell Oil Company 333 Highway G, MS D-2146 Houston, TX 77082-8892 PAST CHAIR Marge Knox Computation Center University of Texas Austin, TX 78712 ADVISORS: -------- Joseph Angelico U.S. Coast Guard Detachment National Data Buoy Center NSTL Station, MS 39529-6000 Art McClinton Mitre 1820 Dolley Madison Blvd. McLean, VA 22102 Al Siegel Battelle Memorial Institute 505 King Avenue Columbus, OH 43201-2693 WORKING GROUPS: -------------- COMMERCIAL Robert Boyd GE Microelectronics Ctr. P.O. Box 13409, MS 7T3-01 Research Triangle Park, North Carolina, 27709-3049 VAX-219 PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX System SIG Committee List FIELD SERVICE Dave Slater Computer Sciences Corporation 6565 Arlington Blvd. Falls Church, VA 22046 INTERNALS Carl Friedberg Seaport Systems, Inc. 165 William Street, 9th Floor New York, NY 10038-2605 LARGE SYSTEMS INTEGRATION Leslie Maltz Stevens Institute of Technology Computer Center Hoboken, NJ 07030 LIBRARY Glen Everhart 25 Sleigh Ride Road Glen Mills, PA 19342 MICROVAX Ray Kaplan Pivotal, Inc. 6892 East Dorado Court Tucson, AZ 85715-3264 (602) 886-5563 MIGRATION AND HOST DEVELOPMENT VAXINTOSH Jim Downward KMS Fusion Incorporated P.O. Box 156 D Ann Arbor, MI 48106 MULTIPROCESSOR Eugene Pal U.S. Army CAORA (ATORCATC) Fort Leavenworth, KA VAX-220 PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX System SIG Committee List NETWORKS Bill Hancock P.O. Box 13557 Arlington, TX 76094-0557 REAL-TIME PROCESS CONTROL Dennis Frayne McDonnel Douglas 5301 Bolsa Aveneu Huntington Beach, CA 92646 Larry Robertson Bear Computer Systems 56512 Case Avenue North Hollywood, CA SECURITY C. Douglas Brown Sandia National Labs Division 2644 P.O. Box 5800 Albuquerque, NM 87185-5800 SYSTEM MANAGEMENT Steve Tihor 251 Mercer Street New York, NY 10012 VAXCLUSTER Thomas Linscomb Computation Center University of Texas Austin, TX 78712 PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Submission Form INPUT/OUTPUT Submission Form A SIG Information Interchange Please reprint in the next issue of the Pageswapper If this is a reply to a previous I/O, which number? ________ Caption: ______________________________________________________ Message: ______________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ Contact: Name _______________________________________________________ Address ____________________________________________________ ____________________________________________________________ ____________________________________________________________ ____________________________________________________________ Telephone ____________________________ Signature _____________________________ Date ________________ Mail this form to: Larry Kilgallen, PAGESWAPPER Editor Box 81, MIT Station, Cambridge, MA 02139-0901, USA To register for on-line submission, dial (in the United States): (617) 262-6830 and log in with the username PAGESWAPPER. PAGESWAPPER - August 1987 - Volume 9 Number 1 INPUT/OUTPUT Submission Form Tear out or photocopy reverse to submit an I/O item Larry Kilgallen, PAGESWAPPER Editor Box 81, MIT Station Cambridge, MA 02139-0901 USA PAGESWAPPER - August 1987 - Volume 9 Number 1 System Improvement Request Submission Form System Improvement Request Submission Form Page 1 of _____ ________________________________________________________________ Submittor: Firm: Address: Phone: ________________________________________________________________ How to write an SIR: Describe the capability you would like to see available on VAX systems. Be as specific as possible. Please don't assume we know how it's done on the XYZ system. Justify why the capability would be useful and give an example of its use. If you wish, suggest a possible implementation of your request. ________________________________________________________________ Abstract (Please limit to four lines): ________________________________________________________________ Description and examples (use additional pages if required) PAGESWAPPER - August 1987 - Volume 9 Number 1 System Improvement Request Submission Form Tear out or photocopy reverse to submit an SIR Mark D. Oakley Battelle Columbus Division Room 11-6-008 505 King Avenue Columbus, Ohio 43201-2369 USA PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX Systems SIG Fall 1987 SIR Ballot VAX Systems SIG Fall 1987 SIR Ballot DECUS membership number __________________ (six digits) Our site uses the following VAX cpus (check all that apply) 8700/8800 ___ 8600/8650 ____ 8500/8550 ____ 8300/8200 ____ 11/780,11/782,11/785 ____ 11/750 ____ 11/730,11/725 ____ MicroVAX I,II ____ MicroVAX 2000, VAXstation 2000 ____ We use VAXes in the following applications(Check all that apply) Business EDP ____ Software Development ____ Education ____ Computer Science Research ____ Data Acquisition/Control____ CAD/CAM ____ Service Bureau ____ Hardware Development ____ Scientific/Engineering ____ Office Automation ____ Telecommunications _____ Other __________________________ I support the following as the most important System Improvement Requests. (List from zero to fifteen SIR's): -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- -------- I oppose the following SIR's as detrimental. (List from zero to five SIR's): -------- -------- -------- -------- -------- Mail to: Mark D. Oakley Battelle Columbus Division Room 11-6008 505 King Avenue Columbus, OH 43201-2693 USA To be counted, your ballot must be received by October 30. PAGESWAPPER - August 1987 - Volume 9 Number 1 VAX Systems SIG Fall 1987 SIR Ballot Tear out or photocopy reverse to vote on SIRs Mark D. Oakley Battelle Columbus Division Room 11-6008 505 King Avenue Columbus, Ohio 43201-2693 USA