[Mailman-Users] If your list uses Topics, I need your help

[Mark Sapiro]

In Mailman 2.1.8 ane prior, there are several problems with Topics.

The help says that you can put keywords one per line in a topic regexp.
This does not work. Also, the Topics regexp is compiled in 'verbose'
mode which means that spaces which are intended to be significant in
the regexp must be inside a bracketed 'character class' or preceded by
a backslash, and hash mark ('#') which is not in a 'class' or preceded
by a backslash introduces a comment which is not part of the regexp.
This too is not mentioned in the help for Topics.

These things have caused problems for people wanting to use Topics.
Some people have figured out how to create topic regexps that work,
but it can be a tedious process to do this.

I propose to change this for Mailman 2.2. My thought is that topic
regexps will no longer be compiled in 'verbose' mode and that
multiline topic regexps will be joined with the OR operator ('|') when
compiled for use so that a topic regexp of, for example,

 one
 two
 three

will actually be equivalent to

 one|two|three

and will match a Subject: or Keywords: header/line containing any one
or more of the three words 'one', 'two' or 'three'.

The problem comes about with lists that have topic regexps which
currently span multiple lines and possible contain comments. An
example might be

 one|    # This is the first word in the list
 two|    # and the second
 three   # and the last, any one of which will match.

Clearly, I can't just join these lines with '|' and compile in
non-verbose mode as this would include the comments and
non-significant white space in the resulting pattern.

My proposal is to convert existing topic regexps to an equivalent,
single line, non-verbose regexp as the list is upgraded when the
instalation upgrades to 2.2, This would result in the above being
changed to

 one|two|three

which would be saved as the regexp for that Topic.

My question (the help I'm soliciting) is does anyone have a problem
with this? To the best of my ability to code it, the Topic would
continue to function exactly as before. The only visible change would
be that the contents of the Regexp box for that topic would change
from the multiline form with comments to a single line without
comments.

So, do you currently have any topic regexps on multiple lines and/or
with comments, and if so would you be disturbed if they were converted
to a single line with the comments and insignificant white space
stripped?

Also, if you have such, and the regexps are more complicated than the
above example, can you send the multiline regexp(s) to me for testing
(off list with subject "Mailman topic regexp" if you don't want to
post them).

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan