[ZPT] [Fwd: Re: ZPT Speed Tutorial Wanted]

kapil thangavelu kvthan@wm.edu
Mon, 16 Sep 2002 06:44:49 -0700


On Monday 16 September 2002 05:18 am, Chris Withers wrote:
> What do other people think?

pysco is not likely a good candiate. its ideally suited for algorithmic code, 
but more importantly for particular functions, as its memory use grows rather 
quickly the more lines of code it needs to cover. i don't think binding it to 
zpt would be feasible.

the other options are optimizing the alt-tal/zpt implementation, or writing 
zpt/tal in C (which i've heard rumors that shane has done in his spare time).

or maybe just using dtml :-)

-k

> cheers,
>
> Chris
>
> -------- Original Message --------
> Subject: Re: ZPT Speed Tutorial Wanted
> Date: Mon, 16 Sep 2002 13:38:47 +0200
> From: "Arnar Lundesgaard" <arnar.lundesgaard@creuna.no>
> To: <chrisw@nipltd.com>
>
> "Chris Withers" <chrisw@nipltd.com> wrote in message
> news:3D85BB36.30301@nipltd.com...
>
>  > Andy Meier wrote:
>  > > My final bottleneck seems to be using Page Templates.  Even a
>
> simple Page
>
>  > > Template costs me 400ms to render.  To compare, rendering a fairly
>
> simple
>
>  > > DTML document takes 9ms.
>  >
>  > This sucks :-( The difference, I'm hoping, is that DTML has been
>
> optimised over
>
>  > the years and has a C implementation.
>  >
>  > > Is there anything I can do to increase the speed of Page Templates
>
> (can I
>
>  > > "pre-compile" the templates)?  Should I convert my page template
>
> into a
>
>  > > python script or dtml document header and footer?
>  >
>  > Hmmm, we really need to find a way to optimise the ZPT process
>
> itself, anyone
>
>  > got any ideas?
>
> Just a wild guess that I wanted to try myself but have not found time
> for yet.
>
> I read on c.l.p about psyco, that gives some speedups for python.
> Perhaps someone could try 'psyco.bind' on the ZPT module. (Dunno if
> this is possible even.)
>
> It is supposed to work out of the box with ordinary python code, but
> I don't know if the PageTemplates code can be classified as such.
>
>
> Long shot indeed :-)
>    Arnar Lundesgaard
>
>
>
>
> _______________________________________________
> ZPT mailing list
> ZPT@zope.org
> http://lists.zope.org/mailman/listinfo/zpt