Am 31.01.2007 um 00:08 schrieb Christoph Zwerschke:
> I just had a look on your benchmarks and yes, you have correctly  
> put the
> compilation/parsing in the setup and measure only the rendering  
> time. It
> seems the performance loss happened in 0.9.4 only; the 0.9.5 does not
> seem to be any slower, or did you notice a loss in 0.9.5 as well?

The two links I sent were supposed to show the differences for each  
version in comparison to the previous. You're right that the slowdown  
was a lot more dramatic for 0.9.4, but 0.9.5 also saw a slowdown,  
just much smaller. I guess that might be related to the added  
location info(?)

> The loss in 0.9.4 is easily explainable with the flexbile  
> formatting of
> the output which has been added in that version and is not  
> optimized for
> performance yet. I'll work on that for the next release.

I fully expected this would have to do with the added formatting  
options. I'm seeing a similar pattern in Genshi, where every tiny  
tweak to improving the correctness/flexibility of the serialization  
phase has a huge impact on overall performance. Serialization is  
really a performance bottleneck for the kind of processing model  
employed by both Genshi and Kid.

Cheers,
Chris
--
Christopher Lenz
   cmlenz at gmx.de
   http://www.cmlenz.net/


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
kid-template-discuss mailing list
kid-template-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kid-template-discuss

Reply via email to