>> While not having done any full integration yet, I am confident that
>> will boost overall performance by a few percents. And, with the then
>> more clear architecture and hopefully reduced recursiveness and thus
>> complexity, Genshi will become more accessible to other developers,
>> so that more people can begin optimizing it.
>>
>
> I don't think a few percent is going to cut it when Jinja is something
> like 2-3x faster and uses far less memory on similar pages. It is hard
> to overlook a speedup of that magnitude in favor of developer elegance.

For sure, 2-3 times faster means that Jinja lies somewhere around
50ms..75ms for a template Genshi now would take 150ms.

Given that stream collecting is being reduced with the new approach
to only a few places, like 2 or so (template parsing, and preprocessing),
I think that with the new approach we could achieve 80..100ms for the
same template. Being overly optimistic here...

> We tried offering a compat layer last time and I think we can all
> agree that it was under used and probably not worth repeating given
> the complexity.

Yes, this was transitional when dropping Clearsilver support in favor
of Genshi. Basically I believe that Genshi's XML based approach is a
much clearer concept and that clearly won over the Clearsilver based
solution.

So I reckon that once the transition to Jinja2 was made, there will
be no turning back, although Jinja2 templates are less capable compared
to Genshi templates, especially when it comes to overlaying templates
and in-template postprocessing of previous template output, e.g. completely
rewriting of the original output on the document level.

--Carsten


-- 
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