On 2008-02-05 09:19:23 PM, Christian Neumair wrote: > 1. Plone > > This mailing list's archive seems to suggest that a Plone migration is > planned, and basic code has been in SVN for a while. > > On the other hand, according to Murray Cumming (IRC), no proper Plone > l10n seems to be possible as of writing, and I couldn't find any plans > when it would be available. From what I've seen, one of the hugest problems with Plone has been the high barrier for starting to work with it. Plone setups always seem to have one or two experts constantly answering questions/performing complex tasks, and when these few people are unavailable, then development pretty much stalls. Repeatedly. This is a very serious (and frustrating) downside that I don't think we really considered in our original decision. > 2. webwmr (debian) > > Because of the excellent language selection facilities on debian.org, > today I asked on debian-www, and they offer their webwmr code on debian > CVS: The websites seem to be static [compiled with a template system > from CVS], with internationalized templates. The translator can receive > change notifications through email (including diffs), and update his own > static (translated) page template. I think we're currently using webwml as well, but without PO-based translations. From what I've seen, Debian's translations aren't 100% in PO files either, so I'd expect this to require a very specific/different translation workflow. As I mentioned in IRC, this would be similar to Plone in (to a far lesser degree) requiring one or two experts to understand the hacky makefiles/the full process by which each page is generated (or people that are willing to go through a lot of work to trace certain errors). > 3. MoinMoin > > It has also been proposed to migrate to live.gnome.org (i.e. MoinMoin). > Page-specific l10n doesn't seem to be planned at all, and the page > contents' layout doesn't seem to be arbitrarily complex HTML, just some > basic markup. I'm sure that Infrastructure will agree that MoinMoin does not scale very well, given it's file-based storage. I think the Fedora Project currently has (one of?) the largest known installs, and we've hit some issues with Moin (http://fedoraproject.org/wiki/MikeMcGrath/MoinIssues). > Conclusion > > Instead of migrating back and forth between various systems, maybe we > should look into keeping the current architecture, writing our own > translation system, and stick with static pages. The translation could > be po-based, or custom language-specific static pages [or templates] > could be created and committed (similar to webwmr). Personally, I would not mind completely abandoning the Plone attempt (it's not like we'd lose a lot of momentum or anything), but this is not my decision to make. I definitely support the idea of static pages, though. As a side note, python.org has a really slick system for generating its pages (no translations), and fedoraproject.org's system is based off of some code from there, with additions to use the Genshi templating engine (and handle PO-based translations), although we are still working through some issues with how we should handle partially translated pages, etc. Thanks, Ricky
Attachment:
pgpw5RzRpNmHE.pgp
Description: PGP signature