[cairo] Xlib/Xcb and EXTEND_PAD

[Zack Weinberg]

Thomas Jaeger <thjaeger at gmail.com> wrote:

> Zack Weinberg wrote:
> > ... because whether or not Render's equivalent of
> > EXTEND_PAD/EXTEND_REFLECT actually *work* is dependent on the exact
> > version of your X server and video drivers.  Render protocol 0.10
> > isn't good enough.  There are some cairo-only test cases in [2].
> 
> Oh, that's not good.  Do we have any data on how widespread this
> problem is?  Using the test case from [2], None, Normal, and Pad are
> working fine on both my systems (intel i965 and super-old radeon
> mobility), Reflect is the only one that produces garbage, and that is
> fixed on my intel system with xserver >= 1.5 and intel driver >=
> 2.5.1.  Is there a better test case that we could use?

The test program at
http://lists.freedesktop.org/archives/xorg/2008-February/032973.html
fails for both Reflect and Pad for me: inspecting the images, it
appears to me that both modes are acting like they're Normal.  

X server      1.4.99.906
Video driver  radeon from git as of Aug 26, 2008
Video card    ATI Technologies Inc R580 [Radeon X1900 XT]

Normally I would be a lot closer to the bleeding edge than I am, but
this is what I get until Debian's done with their freeze.

> How about this: disable cairo's fallbacks on X servers >= 1.5.99 and

I assume you mean >= 1.4.99?

> then fix the drivers when problems become apparent (where fixing can
> mean resorting to the server's software fallback for now).

That might be okay for cairo in the abstract, but I'm not sure it works
for Mozilla, how long is it going to take for up-to-date drivers to get
into the hands of our userbase?

zw