[PATCH] for kid with i18n

[faide]

Performances with i18n fix.

When activating i18n filter, even withtout actually using it the performances drop a lot. And what is really a problem: the performances keep drop as long as the server is serving more files, rendering the site completely unusable (10 sec / request and up)

The problem comes from a bug in Kid which adds a new filter to the chain each time a request comes.

I "rediscovered" the problem tonight when doing some apache ab testing on my app. There was an  old thread in the ML about a fix but the guys did not bother to put a ticket in the track (Mark politely asked them to do so but they completely ignored him)

Attached is a patch against the trunk kid.

I have not found a way to add a ticket to their trac so I put it here.

[faide]

corrects the filter addup in kid

[jorge.vargas]

if you could explain what exactly is happening on the ticket it will be better :)

from the ML

The problem was that the i18n filter was added to the _filters list
each time the page was called... So of course you ended up with a huge
list of filters by the time the page received some traffic. 

[jorge.vargas]

now this is not a blocker because the fix is not actually on our control, since it depends on kid's release, but it is indeed a very important issue.

[faide]

This has been fixed in the Kid trunk.

 http://www.kid-templating.org/trac/changeset/482

Let's hope we get a release soon to base TurboKid? on it :)

[jorge.vargas]

I don't think kid will be released before 1.0.2

[alberto]

Batch moved into unclassified from 1.1 to properly track progress on the later

[faide]

http://trac.turbogears.org/ticket/1301 as been applied to TurboKid? to fix this problem.

The Kid patch as also been accepted by Kid and is due to appear in 0.9.6