[prev in list] [next in list] [prev in thread] [next in thread] 

List:       majordomo-workers
Subject:    Re: Visible post patch for 1.94.4
From:       Dave Wolfe <dwolfe () risc ! sps ! mot ! com>
Date:       1999-01-20 15:04:48
[Download RAW message or body]

[ Kyle R VanderBeek writes: ]

> Well, it hasn't been tested much, but here is a solution to the problem
> I've been having concerning UCE relaying through my lists.  It's a small
> patch to resend and config_parse.pl that adds the "sublist_names"
> string_array to the list configuration (defaulting to just "listname").

Please upload this to ftp://sol.ccsf.cc.ca.us/incoming/ and notify Joe
Jah <jjah@cloud.ccsf.cc.ca.us> that you're requesting it be placed in
the Majordomo optional patches area.

I have one suggestion for the patch as well: when the default is
computed, if the list is a digest append the parent list name to the
digest list name to form the default sublist entries rather than simply
set it to the parent list name. I've seen many digest subscribers reply
to the digest rather than the parent list, even when the Reply-To
header is set (there are a lot of broken mailers out there). I believe
the usual setup is to alias the digest list name to the parent list to
accommodate replies that way. E.g.:

+        'sublist_names',       '#! local($TEMP) = $list;
+                                    if ($list =~ /-digest$/) {
+                                       $TEMP =~ s/-digest$//;
+                                       "$TEMP\001$list";
+                                    } else {
+                                       $list;
+                                    }',

There should probably also be a note in the config variable description
about escaping regex meta-characters, since your code doesn't make any
attempt to do that itself (i.e. regexes can be used for sublist items
the way you have it coded). Why anyone would want to do that or have
listname addresses containing regex meta-characters would certainly be
unusual, but it's always the assumptions and boundaries that get you.

An introductory note with the patch wouldn't hurt either. Something
about having no delusions about how easy it is to spoof this fix, when
adding new config variables to Mj one needs to send a 'writeconfig'
command to all lists to update the config files, and even though this
is an extension, the default is to enable it, so one must explicitly
disable it (remove the value) in list config files where it's not
wanted.

> I should probably also look at patching digest, too, but that can come
> later.

What do you think needs to be done in digest? If anyone has digest
receiving messages that haven't successfully gone through resend,
spamming is the least of their problems.

-- 
 Dave Wolfe

[prev in list] [next in list] [prev in thread] [next in thread]