[cairo] Where's the bottleneck in using glitz/cairo/librsvg?

[Mike Shaver]

On 6/15/06, Keith Packard <keithp at keithp.com> wrote:
> If they're closed-source X terminals, things will suck. If they're LTSP
> boxes, things will be fine. I know of several huge LTSP deployments
> running Gnome, I haven't heard of large scale deployments with
> proprietary X terminals though.

Do you mean here "if they're running open-source X, they'll already
have what they need to get good performance", or do you mean that
they'll be able to upgrade to a newer open-source X server later and
get performance fixes?

We might be unique in our concern for "legacy" open source X servers,
but our concern there is indeed pretty significant.  Even with an
early-2007 ship date, we'd be looking to support, f.e., FC4 and RHEL3
-- and their other-vendor contemporaries; no offense intended to my
friends at Novell and elsewhere!

Now, maybe we just end up detecting these retrograde X servers and
using some SIMD-optimized image backend.  That doesn't help the remote
case so much, but remote X performance is much lower on our list of
concerns than local X performance relative to our "pre-cairo" days.
(And remote X performance of our browsers has never been the sort of
thing that people held parades about either, to be frank.)

Figuring out the best practices here for people looking to deploy
tomorrow's graphics API (cairo) on yesterday's X server would probably
be of value to a number of cairo's users or potential users.  I'll
encourage people in our world to share their experience here, in case
that helps.

Mike