Slow hgweb.cgi

Dominik Psenner dpsenner at gmail.com
Thu Apr 14 01:30:47 CDT 2011


> -----Original Message-----
> From: Isaac Jurado [mailto:diptongo at gmail.com]
> Sent: Wednesday, April 13, 2011 9:15 PM
> To: Dominik Psenner
> Cc: mercurial-devel at selenic.com
> Subject: Re: Slow hgweb.cgi
> 
> On Wed, Apr 13, 2011 at 9:05 PM, Dominik Psenner <dpsenner at gmail.com>
> wrote:
> > So, hg serve is debuggable? How?
> 
> hg --debugger serve

Good morning!

Thanks Isaac for your insights. Unfortunatly debugging hg serve won't help
me (see below).

Here the time of both:

dominik at hgserver:/var/hg/repos/XXX$ time hg in http://hgserver:8000/
comparing with http://hgserver:8000/
searching for changes
no changes found

real    0m0.228s
user    0m0.080s
sys     0m0.056s
dominik at hgserver:/var/hg/repos/XXX$ time hg in http://hgserver/hg/XXX/
comparing with http://hgserver/hg/XXX/
searching for changes
no changes found

real    0m14.385s
user    0m0.100s
sys     0m0.024s

As you can observe, hg serve works fast and well. I see no performance
penalty at all. So the problem sits somewhere between the cgi script and the
actual code behind sync. And what this test proofed that I cannot find the
problem by debugging hg serve. A good try would be to find out in what
pieces of code the cgi script does behave differently from hg serve?

@Python hackers: what's the best method to do that?

Greetings from Italy,
D.



More information about the Mercurial-devel mailing list