New GoAhead Performance

[Eric Valette]

Joel Sherrill wrote:
> 
> I have a suspicion that in the update/merger, we may have
> switched from one basic port of the socket wrapper to
> another.  What is the critical path through the main loop
> of the web server?  Is it using select() in a way it
> wasn't before?
> 
> Since we have eliminated hackery on the filesystem as the
> culprit, a bad choice socket porting layer seems likely.
> 
> Could someone see what significant changed in sock*.c
> between the two points?

There is a newsgroup dedicated to GoAhead Web SErver : news.goahead.com.
I found several possible fixes that may impact performance but have no
time to check them. One indeed is related to select system call "Select
problem found (I think) with GoAhead".

Just my 0.02$. NB : several bug have also been fixed in 2.1 release...

-- 
   __                 
  /  `                   	Eric Valette - Canon CRF
 /--   __  o _.          	Canon Development Europe Team Leader
(___, / (_(_(__         	Rue de la touche lambert
				35517 Cesson-Sevigne  Cedex
				FRANCE
Tel: +33 (0)2 99 87 68 91	Fax: +33 (0)2 99 84 11 30
E-mail: valette at crf.canon.fr	http://www.crf.canon.fr