Mailman 3 January 2003 - Mailman-Developers

Shared Memory?
by Bryce Tate 23 Jan '03

23 Jan '03

Does Mailman use shared memory? Thanks, BT

1 0

post-only list?
by Marc Perkel 23 Jan '03

23 Jan '03

Hi, I'm in the process of testing/upgrading to Mailman 2.1. In Mailman 2.0.x someone had a patch that would let me create a list called "post-only" and members of that list could post to all lists. I use this feature a lot. So - does this patch work/exist in 2.1 - or - is there some other master white list feature that I can use? Marc Perkel Systems Admin Electronic Frontier Foundation

1 0

Announce: LDAP member adaptor Version 2.1
by Martin Whinnery 21 Jan '03

21 Jan '03

Short announcement. I've made available version 2.1 of an LDAP membership adaptor for mailman. Given an LDAP server with groups, it allows you (with some restrictions) to populate list membership dynamically from LDAP. It will recurse groups within groups. It will handle both groups with email addresses or distinguished names as attributes. It is available at http://webserver.offal.homelinux.org/ If you're running mailman and have an LDAP service, please try this and tell me what you think. Or how it broke. Martin Whinnery martin.whinnery(a)offal.homelinux.org

1 0

Re: [Mailman-Developers] Re: [Spambayes] spambayes fronting a mailing list?
by Donn Cave 21 Jan '03

21 Jan '03

Quoth barry(a)python.org (Barry A. Warsaw): ... | False positives get caught on the admindb screen, so you approve and | train them in one action. Although I never saw any false negatives, I | think the way to handle these will be to add a -spam address that | people can send messages to. If the list admin sends it then it gets | spam trained. If not, the list admin will have a chance to decide | whether to spam train it or not. I thought of that, but had some doubts. I'd strike the ``if the list admin sends it'' part, but maybe other sites are OK with admin passwords or whatever it takes to make that work. But mainly I'm not sure I see how an average admin is going to re-send something and make it look like the original - I'd hate to accidentally train a spam filter to reject all mail from the admin! Donn Cave, University Computing Services, University of Washington donn(a)u.washington.edu

1 0

Re: vaca invites
by Tom Neff 20 Jan '03

20 Jan '03

One idea would be to put a spam-style filter on the inbound invite-confirm mail processor, so that phrases like "out of the office" could be caught. Thanks to Barry for moving the sourceforge tracker to the new list!

1 0

Multipart/Alternative mangling
by Brad Mettee 20 Jan '03

20 Jan '03

Pardon the intrusion, but..... I've searched the archives but haven't found any reference to this problem, so here it is. We just installed v2.1 of mailman and are seeing some odd things in a few versions of Outlook and Outlook Express related to header/footer info as added by mailman. Mailman 2.0 used to eat the footer when it was a plain/html mixed e-mail, now the problem is that an e-mail composed in plain & HTML text, originally sent out as Multipart/Alternative (M/A), gets turned into Multipart/Mixed (M/M) and some versions of Outlook and OE display the footer ok, others turn it into an attachment, and yet others display it AND turn it into an attachment. I know where the code is to add the header/footer info, and understand how it works, and do know (from viewing a direct download of e-mail via telnet to pop3 server) that mailman is handling it exactly the way it was written to do it. The problem is that wrapping a new MIME type (M/M) around the original M/A type, and then adding the headers & footers to the new M/M data is causing problems with how it's interpreted by the different e-mail programs. I have a simple solution, but don't know how to implement it. I've tried inserting new sections into the original M/A data (one new plain, one new html), but that causes loss of data in Eudora, as well as not showing up at all in Outlook 2002. The raw e-mail via telnet looks exactly like I'd expect it to, so I'm sure it's the e-mail reader that's mangling the results. Here's my idea on how to address the problem. From what I've seen, every single mail program (and I've had people mail me from 5 diff ones so far) uses content-type: M/A for text/plain & text/html messages. What I want to do is test for multipart/alternative, append plain text header/footer to the text/plain section, and append modified header/footer with html changes to the text/html section. I'm very new to python, and haven't a clue as to how to modify the contents of each section of the multi-part data. Here's some mocked up code that shows what I think needs to be done to fix the problem. at about line 106 of Mailman/Handlers/Decorate.py # 03-01-18 BTM - added to handle HTML type multi-part information elif msg.get_type() == 'multipart/alternative': # The next easiest thing to do is prepend the header and append # the footer to each subpart of both text/plain & text/html # here's where I'm fuzzy on walking the list of subparts msgplain = msg.getpart('plain') frontsep = endsep = '' if header and not header.endswith('\n'): frontsep = '\n' if footer and not oldpayload.endswith('\n'): endsep = '\n' newplain = header + frontsep + msgplain + endsep + footer # again, no clue on setting it back into the msg msg.set_part('plain', newplain) if header: header = '' + header.replace('\n', ' \n') + '' if footer: footer = '' + footer.replace('\n', ' \n') + '</html>' msghtml = msg.getpart('html') # I'm aware that the case of <HTML> varies in diff e-mail apps, not sure how to overcome it (can replace be case insensitive?) if header: newhtml = msghtml.replace('<html>', '<html>\n' + header) if footer: newhtml = newhtml.replace('</html>', footer + '\n</html>') msg.set_part('html', newhtml) wrap = 0 # 03-01-18 BTM - end of additions I know this won't fix 100% of the problems with mixed message types, I just found out this evening that Eudora 4.3.2 doesn't attach my signature in HTML if I attach a file. The main body of the e-mail is HTML, but the signature is in plain text, and the whole msg is M/M. If I don't attach a file, the entire message (including signature) is done in HTML & plain text just fine, and sent as M/A. I'm sure there will be some cases where it just doesn't show up on the destination mail program, but this should make it work more often than it currently does. I've figured out python syntax pretty well, just can't figure out what methods of what objects to use to get/set the pieces of the multipart/alternative that need alteration. I'm not a python programmer, and before Saturday had never used it before, so my knowledge is limited to what I found on python.org in the reference docs. If no-one wants to solve this directly, can someone point me in the right direction (maybe with code snippets, or source lines of code already in mailman) on how to get the contents of the text/plain (or text/html) sections of the mime data? Does this make any sense? If not, clue me in, please! (if this doesn't go through to the list intact, I'll re-send with the code as an attachment). Thanks, Brad Mettee Baltimore, MD

1 0

M/A mangling - oops
by Brad Mettee 20 Jan '03

20 Jan '03

I accidently sent to the list with the wrong return address, I will re-send with the right one so it gets posted directly. Thanks for your patience, Brad Mettee Baltimore, MD

1 0

Multipart/Alternative mangling
by Brad Mettee 20 Jan '03

20 Jan '03

Pardon the intrusion, but..... I've searched the archives but haven't found any reference to this problem, so here it is. We just installed v2.1 of mailman and are seeing some odd things in a few versions of Outlook and Outlook Express related to header/footer info as added by mailman. Mailman 2.0 used to eat the footer when it was a plain/html mixed e-mail, now the problem is that an e-mail composed in plain & HTML text, originally sent out as Multipart/Alternative (M/A), gets turned into Multipart/Mixed (M/M) and some versions of Outlook and OE display the footer ok, others turn it into an attachment, and yet others display it AND turn it into an attachment. I know where the code is to add the header/footer info, and understand how it works, and do know (from viewing a direct download of e-mail via telnet to pop3 server) that mailman is handling it exactly the way it was written to do it. The problem is that wrapping a new MIME type (M/M) around the original M/A type, and then adding the headers & footers to the new M/M data is causing problems with how it's interpreted by the different e-mail programs. I have a simple solution, but don't know how to implement it. I've tried inserting new sections into the original M/A data (one new plain, one new html), but that causes loss of data in Eudora, as well as not showing up at all in Outlook 2002. The raw e-mail via telnet looks exactly like I'd expect it to, so I'm sure it's the e-mail reader that's mangling the results. Here's my idea on how to address the problem. From what I've seen, every single mail program (and I've had people mail me from 5 diff ones so far) uses content-type: M/A for text/plain & text/html messages. What I want to do is test for multipart/alternative, append plain text header/footer to the text/plain section, and append modified header/footer with html changes to the text/html section. I'm very new to python, and haven't a clue as to how to modify the contents of each section of the multi-part data. Here's some mocked up code that shows what I think needs to be done to fix the problem. at about line 106 of Mailman/Handlers/Decorate.py # 03-01-18 BTM - added to handle HTML type multi-part information elif msg.get_type() == 'multipart/alternative': # The next easiest thing to do is prepend the header and append # the footer to each subpart of both text/plain & text/html # here's where I'm fuzzy on walking the list of subparts msgplain = msg.getpart('plain') frontsep = endsep = '' if header and not header.endswith('\n'): frontsep = '\n' if footer and not oldpayload.endswith('\n'): endsep = '\n' newplain = header + frontsep + msgplain + endsep + footer # again, no clue on setting it back into the msg msg.set_part('plain', newplain) if header: header = '' + header.replace('\n', ' \n') + '' if footer: footer = '' + footer.replace('\n', ' \n') + '</html>' msghtml = msg.getpart('html') # I'm aware that the case of <HTML> varies in diff e-mail apps, not sure how to overcome it (can replace be case insensitive?) if header: newhtml = msghtml.replace('<html>', '<html>\n' + header) if footer: newhtml = newhtml.replace('</html>', footer + '\n</html>') msg.set_part('html', newhtml) wrap = 0 # 03-01-18 BTM - end of additions I know this won't fix 100% of the problems with mixed message types, I just found out this evening that Eudora 4.3.2 doesn't attach my signature in HTML if I attach a file. The main body of the e-mail is HTML, but the signature is in plain text, and the whole msg is M/M. If I don't attach a file, the entire message (including signature) is done in HTML & plain text just fine, and sent as M/A. I'm sure there will be some cases where it just doesn't show up on the destination mail program, but this should make it work more often than it currently does. I've figured out python syntax pretty well, just can't figure out what methods of what objects to use to get/set the pieces of the multipart/alternative that need alteration. I'm not a python programmer, and before Saturday had never used it before, so my knowledge is limited to what I found on python.org in the reference docs. If no-one wants to solve this directly, can someone point me in the right direction (maybe with code snippets, or source lines of code already in mailman) on how to get the contents of the text/plain (or text/html) sections of the mime data? Does this make any sense? If not, clue me in, please! (if this doesn't go through to the list intact, I'll re-send with the code as an attachment). Thanks, Brad Mettee PC HotShots, Inc. Baltimore, MD -> Let us bring out the *Power* of your PCs. <- -> Custom Business Software Solutions since 1991 <- visit http://www.pchotshots.com for information about our company.

1 0

Re: [Mailman-Users] "Invite" vs. autoresponders
by barry＠python.org 19 Jan '03

19 Jan '03

[I'm moving this discussion to mailman-developers. -BAW] >>>>> "SB" == Stonewall Ballard <sb.list(a)sb.org> writes: SB> I've recently discovered that vacation autoresponders will SB> subscribe recipients to Mailman lists when they get "invited". Dang. This is because the From address contains the confirmation cookie encoded in the address. This might kill this idea for ease-of-use confirmations. SB> This is not good. It makes the feature dangerous if you've SB> promised to not subscribe anyone without their assent. Can you submit a bug report on this? We'll have to decide what to do long term. SB> Is there any way to avoid this problem, other than not using SB> the feature? This is yet another reason to move to URLs with SB> unique keys instead of using passwords. Such a URL could be SB> embedded in the invitation message. Take away the From header cookie and that's essentially what you've got in the invitation message. I'm up for any other ideas. -Barry

4 4

Sourceforge notifications considered harmful
by Tom Neff 19 Jan '03

19 Jan '03

It completely sucks, and has completely sucked ever since the decision was made a few months ago, that Mailman-Developers gets Cc'd on every Sourceforge update. I'm sure Barry (or whoever) felt they were doing a service, but it has practically killed discussion here and cost us many, many kilobytes of repetitive cruft, since everytime anyone adds a "yeah me too" comment to the end of a bug, the ENTIRE thing has to be reposted here. If you don't actually want to have a Developers discussion list, then shut it down and let people submit stuff via Sourceforge only. If you do want one, let us have one. Sure, it seems legit to let the list know what bugs have been added/updated. So run a nightly script that prepares a one-line-per-bug summary of adds & updates, and let that be a single (very useful) message on the list.

10 14