On Thu, 04 Mar 2010 06:49:54 -0800 Mark Sapiro <mark@msapiro.net> wrote:
On 3/4/2010 4:23 AM, Cedric Jeanneret wrote:
I think I found what's the problem is : the script works now, but as I write my own archiver, it doesn't do the pipermail part (i.e. update mails in archive)... I thought that this code :
mlist = MailList.MailList(maillist, lock=False) msg = email.message_from_file(sys.stdin, Message.Message) f = StringIO(str(sys.stdin)) h = HyperArch.HyperArchive(mlist) h.processUnixMailbox(f) f.close()
did all, but after reading a bit of code, it doesn't exactly. It saves to .mbox file, right ?
No. It doesn't save to the .mbox file. If you look at the ArchiveMail() method in Mailman/Archivers/Archiver.py. it first saves to the .mbox by doing
if mm_cfg.ARCHIVE_TO_MBOX in (1, 2): self.__archive_to_mbox(msg)Then it either calls the external archiver or executes essentially the above to archive the mail in the pipermail archive.
What you are missing is
h.close()
and that's why it doesn't work.
I tried to find where it does the pipermail stuff, but it's a bit complicated [I'm not so at ease with Python].
Yes, the archiver is very convoluted because classes are subclassed and methods overridden all over. Don't feel bad. I've been looking at it for years and still only barely understand it.
hmmm, I use the h.close() a bit after (I catche its latest ID so that I ca build the direct URL for my indexer). But for now, I guess I'm done. I've opened a bug (didn't figure where I could put my stuff) on launchpad: https://bugs.launchpad.net/mailman/+bug/531942 It contains my scripts, and some informations on how to use them.
Indeed, "arch" script uses locks. I copied it, removed the lock stuff, and used this version. All work fine now.
I'm happy I could understand a bit (well... very little bit) how mailman works.
Thanks again !
-- Cédric Jeanneret | System Administrator 021 619 10 32 | Camptocamp SA cedric.jeanneret@camptocamp.com | PSE-A / EPFL