On Sun, Jan 11, 2004 at 06:28:49PM -0500, Daniel Macks wrote:
> On Sun, Jan 11, 2004 at 11:53:41PM +0100, Max Horn wrote:
> > Am 11.01.2004 um 23:31 schrieb Ben Hines:
> > >On Jan 11, 2004, at 2:22 PM, Daniel Macks wrote:
> > >
> > >>Switch to an object-oriented and consistent way of handling source
> > >>tarballs. Seems to slow down indexing quite a bit and make fink.db up
> > >
> > >If there is anything we don't need it is to slow down indexing. I
> > >don't care how much cleaner it makes the code, we need to index
> > >constantly. It needs to be fast.
> >
> > I agree with Ben. Indexing is already now terribly slow. We spent a lot
> > of effort to get it as fast as possible (using Storable, at least) - a
> > noticeable slow down is not acceptable, IMNSHO.
>
> I also agree, which is why I put it in a branch:)
>
> It started out as some changes that didn't have a timing effect, but
> the more I worked on it...
Okay, I just checked the things I did in HEAD. For 'fink index':
Benchmark: timing 10 iterations of now, was...
now: 169 wallclock secs ( 0.00 usr 0.04 sys + 146.45 cusr 13.56 csys = 160.05
CPU) @ 250.00/s (n=10)
was: 154 wallclock secs ( 0.00 usr 0.02 sys + 136.48 cusr 13.73 csys = 150.23
CPU) @ 500.00/s (n=10)
Depending on how you want to spin those number, it's either only 1.5
seconds slower on my not terribly recent G4, or it's 10% slower:) I
could probably bring that down a little but not much. Undo it all?
dan
--
Daniel Macks
[EMAIL PROTECTED]
http://www.netspace.org/~dmacks
-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html
_______________________________________________
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-devel
