On 3/23/2010 12:43 AM, Greg Troxel wrote:
Christian Boos<[email protected]>  writes:
On 3/22/2010 2:45 PM, W. Martin Borgert wrote:
Quoting "Christian Boos"<[email protected]>:
We're getting quite close to being able to release a beta of
0.12. While there are still a few important tickets to be finished
first (#9060 for example), we're really getting there.
AFAIK, 0.12 needs Genshi>  0.5.1[1], but 0.6 is not yet released[2].
For Debian (and other distribution as well, I assume) it would be
nice to have a Genshi release before Trac RCs.
Yes, it would be nice.
Are there any plans to synchronise Genshi 0.6 and Trac 0.12
releases? Thanks in advance!
There's no plan, only hope ;-)

More seriously, I'm not really at ease with the fact that either
Genshi and Babel are more or less on life support, since  Christopher
has switched to other interests. Pedro Algarvio seems our best chance
here ;-) Pedro?

Related:
  - http://genshi.edgewall.org/ticket/369
  - http://babel.edgewall.org/ticket/212
Speaking as someone using trac seriously, it seems scary to be talking
about a 0.12 release that depends on unreleased versions of other tools.

I understand and share your concerns.

If that's really the case, then I'd say the other tools need to be
encouraged/helped to get to release or forked and released before 0.12
has an rc.

We have considered that. It looks like we're going to have a Babel 0.9.5 really soon (I see the packages on ftp.edgewall.org as I'm writing this), and perhaps a Genshi 0.6 release at some later point (keep finger crossed).

The fact remains that those libraries are currently strong dependencies for Trac (Genshi even more so than Babel), and neither has currently the level of maintenance we were used to have, every sign indicates this trend won't change. I've also stopped to hope there will be any drastic performance improvement, and while there has been some improvements last year (1), I rather feel that this means that all that could be realistically done has now been done on this topic. Of course, theoretically someone could step up anytime and bring in a radically new idea and dramatically increase performance, but that's precisely what I've been hoping for since the last 4 years. While there has been such attempts, made by bright minds, none really succeeded.

So don't get me wrong, I actually like Genshi, I was really pleased to ditch the ugly and hard to maintain Clearsilver templates and enthusiastically converted most of the Trac templates to it. I also helped to fix (easy) Genshi issues during the early stage of its development, but was always intimidated by its complexity. I still enjoy very much *using* Genshi from a developer perspective, but for the reasons exposed above, from the point of view of Trac maintenance and its future evolution, we really have a situation to address, lots of users are still hesitant to move away from Trac *0.10* because of performance concerns (Genshi is perhaps not the only factor at play there, but it's an important one).

It seems however that there could be a way forward, by considering once again a switch of the templating engine. This time it would be in favor of Jinja2 (2), which seems to have a great momentum and have the adequate capabilities and speed we need. In addition, its main developer is also a long time Trac supporter and has shown interest in supporting us for the switch (hello Armin!).

There has been some informal discussions on this topic on the #trac IRC channel, so the logical next step is to discuss it more formally here on Trac-dev. To me at least, the status quo is not really an option. I'm aware this will cause some inconvenience, especially for the plugins that currently depend on stream filters. However this will be mitigated by the fact that we can continue to support Genshi, and migrate to Jinja2 only gradually, starting with the time critical parts (like the changeset view, the browser view and the timeline).

-- Christian

(1) - http://genshi.edgewall.org/changeset/1038 and 1052
(2) - http://dev.pocoo.org/projects/jinja/

--
You received this message because you are subscribed to the Google Groups "Trac 
Development" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/trac-dev?hl=en.

Reply via email to