[ZPT] Parsing speed?

richard@bizarsoftware.com.au richard@bizarsoftware.com.au
Tue, 08 May 2001 12:00:59 +1000


Guido van Rossum wrote:
> 
> > Are there any moves to making the parsing of pagetemplates faster?
> 
> Not until we've got an indication that it's currently too slow.  The
> expat-based XML parser is about twice as fast as the re-based HTML
> parser, by the way.  The theory is that parsing occurs rarely -- the
> parse output is cached.  (We could cache more aggressively by making
> the cache persistent; it currently isn't.)

Yeah, parsing occurs pretty infrequently - I suspect it'll be easier on
everyone if we just provide some feedback to the user that the system is
actually doing something while parsing all the templates. I will also give
the expat parser a whirl and see how that affects our performance.


> > How about rendering speed? It seems pretty darned snappy at the moment, but
> > then I'm running on an 800 MHz processor, and most things are pretty snappy
> > on it ;)
> 
> The rendering engine is pretty optimized!

Definitely. I've been very impressed with it thus far :) I just realised
that one of the big slow-downs we get is the debug-mode stuff we've got in
there (a biggie is the call to pprint for the __traceback_info__ that we've
got in there). I was completely blaming the wrong culprit, sorry!


    Richard

-- 
Richard Jones
richard@bizarsoftware.com.au
Senior Software Developer, Bizar Software (www.bizarsoftware.com.au)