BTW, the NNTP queue runner has now been ported to Mailman 3. You will need to re-run bin/buildout though, to pick up the new dependency on the mock library.
On Apr 01, 2012, at 04:08 PM, Alexander Sulfrian wrote: >> MM3 will be a better platform to build something like the NNTP >> access on. The question in my mind is whether this should be done >> as part of the various independent (but related) archiver projects, >> or whether it should be done as a separate "archiver". > >there is a second question connected with that: Should the messages >be kept in an additional storage for NNTP access or should the default >archiver be responsible for storage and should be extended with methods >for accessing specific messages? This is a good, but larger question. I've always thought that Mailman will require a "message store" as defined in the IMessageStore interface. What might make sense is to have a single implementation that satisfies the IArchiver and IMessageStore (and possibly other interfaces), but with a single on-disk storage. This could in fact be the thing that backs the prototype archiver. >> In mm3, there's an API for feeding posted messages to an IArchiver, >> but this is quite flexible. I could imagine that something on the >> other end of this vended messages via NNTP instead of HTTP. > >This would be the scenario if implementing the NNTP access in a new >archiver, separated from the other. With the above, you probably wouldn't need this except as you say, if it is a separate archiver. >> The one key difference is that you'd like to be able to post to the >> mailing list through NNTP, with probably some additional posting >> rules (e.g. if you're not a member, but we "know" you, or you've >> been approved for posting a few times, your message wouldn't get >> held for moderator approval). > >If it should be possible to post messages over the NNTP transport, >that does not match the classic design of an archiver. I do not know, >whether there is an API to post messages, but eventually it would be >better to implement the NNTP archive as external module, that could >maybe even run on a separate server. Yes, now that the NNTPRunner is functional, it should be possible to set this up as posting to an NNTP service that a site could run, independent of Mailman. >> If I was doing this, I'd probably looks seriously at Twisted as the >> basis for implementing the NNTP side of things. I haven't looked in >> quite a while, but at the time, it had great support for NNTP >> server-side. > >Yes, twisted should be the right choice. There is a twisted module for >implementing a NNTPServer[1], but it is not very well documented. But >even if it is not working, it should not be hard to implement it. The >NNTP commands described in RFC3977[2] do not look very complicated. > >Additional to that, there is also the question, whether it should be >possible to sync a few mailman server over the NNTP protocol. That >would be a possibility to do clustering for load balancing or >something like that. That's a pretty cool idea, actually. Something fun to explore for 3.1 perhaps. -Barry
signature.asc
Description: PGP signature
_______________________________________________ Mailman-Developers mailing list Mailman-Developers@python.org http://mail.python.org/mailman/listinfo/mailman-developers Mailman FAQ: http://wiki.list.org/x/AgA3 Searchable Archives: http://www.mail-archive.com/mailman-developers%40python.org/ Unsubscribe: http://mail.python.org/mailman/options/mailman-developers/archive%40jab.org Security Policy: http://wiki.list.org/x/QIA9
