Re: Two bugs in 4.1.1

--- Tuomas_Härkönen <[email protected]> wrote:
> > <snip>
> > summary: 10k characters on a line is slow, much much slower than 10 chars
> > on a thousand lines.
> > </snip>
> > <...snip...>
> > Could you try this without any of your own regexps loaded (or confirm you 
> > didn't have any of your own regexps loaded) so we can be sure the bad regexp
> > is in the core code?
> 
> The only regexp I've changed is SEMI_REGEXP in utils.py, which is the 
> command separator. However, since I've made some other local hacks, I 
> also tested this on the 4.1.1-release as well as CVS version (checked 
> out yesterday), and the problem persists.

It was the line regexp in net.py, it was using a look ahead that consumes
an awful amount of time for long lines.  Worse, it would do it each time
it read from the socket so it would spin CPU scanning a 1k string, 2k string,
3k string .. to whowever big the string is.

I checked in a change to net.py that avoids this.  A 50k line prints almost
instantly. Sourceforge lags reality by a day or so, let me know if you want 
a copy by mail.

-Brazen



		
__________________________________
Do you Yahoo!?
Yahoo! Mail - Helps protect you from nasty viruses.
http://promotions.yahoo.com/new_mail


-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click