RELEASE NOTES FOR VERSION 3.1
		    =============================

1. Data Structures
------------------
The data structures were re-written so that (hopefully) a significant
improvement in the speed of parsing and making will be observed. Many
strings that were malloc'ed on the fly are now character arrays on the
stack. I'm pretty sure this is more efficient than lots of repetitive
calls to malloc(). Where I was uncertain about the maximum length of a
string, I left it as a dynamically allocated string.

2. Default Rules
----------------
The handling of default rules has been completely revised...
    o Rules are added to the head of the default rules list, so that
      if a redundant default rule is encountered it supersedes the
      old copy, since lists are searched from head to tail.
    o Multiple default rules for a given target are allowed. This
      means that a .OBJ may depend on .C OR .MAR OR .CLD. Just list
      each rule separately. The first rule that matches the target AND
      has an existing matching prerequisite is 'made'. If no matching
      rule is found that has a prerequisite that exists, the first
      rule (if any) that matched the target is used.

3. Target Line syntax
---------------------
Target lines have a new extension to the syntax. The allowable
delimiters for targets and prerequisites are whitespace (old) and
commas (new). This allows one to define a symbol as a list of object
modules separated by commas, so that the symbol may be used in both
the target line and the link command. The colon separating the targets
from the prerequisites must still be followed by at least one
whitespace character if there are any prerequisites.

4. Symbol Definitions
---------------------
Additional definitions of a symbol supersede the old value. All whitespace
at the beginning of the symbol definition is eaten.

5. Continuation Lines
---------------------
Continuation lines are handled in a much cleaner manner. The allowable
continuation characters are the backslash '\' and the hyphen '-'. Upon
digesting an input line, both are converted to a backslash. When
target lines are parsed, all backslashes are converted to spaces, so
it is important not to continue a line in the middle of a target name.
When DCL command lines are added to the output file, backslashes are
converted to hyphens and the line is broken after the hyphen.

6. INCLUDE Files
----------------
The command 'INCLUDE <file>' is interpreted as a directive to read the
specified file and parse it. The word 'INCLUDE' must be all caps and
must not be preceded by any whitespace or it will be interpreted as a
DCL command line.