From: Joaquín Cuenca Abela (cuenca@pacaterie.u-psud.fr)
Date: Sat Jul 20 2002 - 08:56:41 EDT
just to let you that I've found the problem. The layout redoes each run
n*(n + 1) / 2 times (n being the number of pages).
That gives us ~ 205^2 / 2 * the number of runs that we should really
have.
I don't have enough time to explain all that right now. I will try to
fix it tomorrow and send an explanation.
Cheers,
On Fri, 2002-07-19 at 20:00, Joaquín Cuenca Abela wrote:
> On Fri, 2002-07-19 at 10:23, Mike Nordell wrote:
> > Joaquín Cuenca Abela wrote:
> >
> > > Loading a 205 pages document takes ~2 minutes. That sucks.
> > > Here's what gprof has to say:
> > [...]
> >
> > Calling PP_evalProperty close to six-and-a-half-million times could have
> > something to do with it (understatement of the year? :-) ).
>
> Oh, this one is not the most called. I have some functions called more
> than 90000000 times
>
> > Also, I fail to see the reason of calling fp_Run::updateHighlightColor over
> > two and a half million times just to load such a document. It's not like
> > you'd be able to cram that many runs into a 205 page document. That would be
> > something like 13052 runs/page!
>
> Yup. I'm getting a mean of 3 runs by line (which looks reasonable), but
> fp_Line::updateBackgroundColor is called 662765 times (when it's
> impossible to have more than 10000 lines with Times 12pt in 205 pages).
>
> So it seems that we're having an inflate number of fp_Lines...
>
> I'll keep investigating
>
> Cheers,
>
> --
> Joaquín Cuenca Abela
> cuenca@pacaterie.u-psud.fr
>
-- Joaquín Cuenca Abela cuenca@pacaterie.u-psud.fr
This archive was generated by hypermail 2.1.4 : Sat Jul 20 2002 - 09:01:47 EDT