Re: so, is this claim about pango still true? or does nobody actually care?



On Thu, 2005-10-13 at 23:30 -0400, Paul Davis wrote:

> concerns on GTK+-2.x.  As a reference the perf issue I was having was
>  most evident when updating a textual label (reporting some ADC value)
>  at a rapid rate (10x per second or so).

This will cause the label to queue a resize on itself, so the whole
window will be size_requested.  That is an expensive operation for
numerous reasons.

If you'll have many labels that update quickly, you probably want your
own version of a Label widget that does not queue a resize each time the
text changes.  I know we do something like that in Evolution (probably
not for speed reasons, but to avoid windows resizing all the time when
some sort of status label changes frequently).

Also, do you really need to update labels 10 times per second?  The user
won't distinguish the changes.

This *needs* to be made faster, anyway; see below.

> is this still true? does anybody care? is there a way to avoid pango
> entirely and still get AA fonts inside GTK2? will this ever be fixed
> before everyone is using h/w acceleration to print button labels?

Many people care.  The problem is that up until now, we have not had
good/friendly tools to profile complex apps like GNOME's.  Now that we
have them, we can do something about our performance problems.

Hardware acceleration will not help you; for text, the largest time
consumer as of this date is just shaping and measuring in Pango.

I've been working on optimizing Pango recently; my progress is here:

http://primates.ximian.com/~federico/news.html

So far I have a 25% improvement on the time it takes to shape and
measure Latin text.  A lot is happening in this area; consult my
activity log frequently to get updates.

  Federico




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]