Note 102.79                 Undocumented VMS Features                   79 of 80
XDELTA::HOFFMAN "Steve; VMS Engineering"             97 lines  12-MAR-1996 13:16
              -< Using SORT on seq file with non-sortable header >-
--------------------------------------------------------------------------------
Article 142299 of comp.os.vms:
Path: nntpd.lkg.dec.com!depot.mro.dec.com!pa.dec.com!news1.digital.com!decwrl!lll-winken.llnl.gov!fnnews.fnal.gov!nntp-server.caltech.edu!seqaxp.bio.caltech.edu!MATHOG
From: mathog@seqaxp.bio.caltech.edu
Newsgroups: comp.os.vms
Subject: Re: SORT, keep N lines???
Date: 8 Mar 1996 18:47:34 GMT
Organization: Biology Division, Caltech, Pasadena CA 91125
Lines: 82
Message-ID: <4hpvc6$pvd@gap.cco.caltech.edu>
References: <6205214@MVB.SAIC.COM>
Reply-To: mathog@seqaxp.bio.caltech.edu
NNTP-Posting-Host: seqaxp.bio.caltech.edu

In article <6205214@MVB.SAIC.COM>, "GWDGV1::MOELLER" <moeller@gwdgv1.dnet.gwdg.de> writes:
>re the problem posed by D.Mathog:
>
>>Often I need to sort a sequential file that has these contents
>>
>>  header
>>  :
>>  delimiter
>>  :
>>  stuff to sort
>>
<snip>
>
>A rather low-overhead solution can possibly be had from using PPFs
>(process permanent files, i.e. files opened by DCL). For example,
>
>	$ open/read INPUT {inputfile}
>	$ open/write OUTPUT {outputfile}
>	$!
>	$ {command that copies the header from INPUT to OUTPUT}
>	$!
>	$ SORT/{qualifiers} INPUT OUTPUT
>	$!
>	$ close INPUT
>	$ close OUTPUT 
>
>SORT has no problem reading the left-over lines from INPUT,
>and appending its sorted output to OUTPUT; some I/O optimizations
>may be inhibited, though.

It seems like black magic, but Wolfgang is right, this works exactly as
advertised:

$ create test.txt
abc
abcd
abcde
abcdef
1
2
3
4
5
6
7
8
9
$ open/read input test.txt
$ open/write output test_out.txt
$! move first 4 lines
$ read input string
$ write output string
$ read input string
$ write output string
$ read input string
$ write output string
$ read input string
$ write output string
$! sort the rest
$ sort/key=(pos:1,size:1,descending) input output
$ close input
$ close output
$ type test_out.txt
abc
abcd
abcde
abcdef
9
8
7
6
5
4
3
2
1

Thanks for the suggestion!

David Mathog
mathog@seqaxp.bio.caltech.edu
Manager, sequence analysis facility, biology division, Caltech