Ideas that won’t die
by havoc
gnomedesktop.org linked to
a
VentureCake article on ways to improve GNOME. While the article
covers a lot of ground, as a kind of aside it brings up the age-old “X
would be faster if it knew about widgets and not just drawing”
theory.
I wonder why this theory comes up repeatedly – it is one hundred
percent bogus. In fact the direction over the last few years is the
opposite. With Cairo, the X server is responsible for hardware
acceleration and the client side implements operations like drawing a
rectangle. Moving more logic client-side was one of the reasons for
the switch to Cairo.
Must be a lesson here somewhere about the value of applying intuition
to performance.
(And yes, some things may be slower and others faster in recent
Cairo-using apps vs. older apps, but I assure you it has nothing to do
with whatever you think it has to do with – unless your thinking is
based on profile results from solid benchmarks.)
(This post was originally found at http://log.ometer.com/2007-07.html#5.2)