Re: svn commit: r8935 - branches/1.0.x

On Tue, 2004-03-09 at 19:32, Greg Hudson wrote:
> On Tue, 2004-03-09 at 12:20, kfogel@collab.net wrote:
> > > Aw come on :) ;). It's just allocating out of a different pool
> > > (connection instead of request), and retrieving an item from a table (or
> > > putting one there if it wasn't previously). Pretty trivial.
>
> > Safety first, scouts :-).
>
> Oh, right, I should have triggered on this before. A performance
> optimization should not go into 1.0.x unless performance is demonstrably
> unacceptable without it.

Whether performance is acceptable is ofcourse something subjective
and rather specific to each site.

I personally think that reducing the number of times a file is opened
and parsed from the number of requests + subrequests (which can be
quite a lot) to one* can make a substantial difference.

I guess I could go and produce numbers, but I don't really feel that's
worth the effort.

> (That means I agree with Karl.)

Yeah, I got that one ;)

Sander

*) the number of connections to be exact, but hey

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Mar 9 22:28:11 2004