Re: [gtk-list] Speeding things up.

[Paul Barton-Davis <pbd op net>]

In message <Pine.GSO.4.00.9903211418290.8227-100000@orff-5>you write:
>As we know, loading pixmaps is more than slow...
>When using Xlib, you could allways speed things up by combine all pixmaps 
>to one big, load it and split it.
>
>Does gtk/gdk provide functions for this type ofa manipulation?

just use the regular pixmap functions, but when you draw it, use
gdk_pixmap_draw() with relevant offsets and sizes into the pixmap. You
get to specify where the src pixmap should be considered to begin, and
how big it is. There's a program called "gsyn" which does this, for
example: its "knobs" are pixmaps containing 32 different images of the
knob in various states of rotation, and it just calls
gdk_pixmap_draw() with the appropriate offsets (the size of each
section is known ahead of time).

this raises a question of my own: how much overhead is there in the
server when you load N pixmaps instead of 1 pixmap that contains as
many pixels as all N combined ?

i ask this because i have my own "knobs" (see
http://www.op.net/~pbd/quasimodo/) which I manage as a set of pixmaps,
instead of using the "section-within-a-pixmap" approach, and I'm
wondering how much server memory/processing overhead i incur doing
things this way.

--p