[cairo] build now works

[Uli Schlachter]

Hi,

Am 20.06.2011 21:17, James Cloos wrote:
> I uploaded two traces to:
>
>   http://people.freedesktop.org/~cloos/Cairo/traces/firefox-startstop-xlib.trace.xz
>   http://people.freedesktop.org/~cloos/Cairo/traces/firefox-startstop-xcb.trace.xz
>
> (They should be publically readable on annarchy, as well.)
>
> Both are of FF40, showing a startup and immediate quit.
>
> The xlib one uses cairo ced061124d0b compiled with --disable-xlib-xcb.
>
> The xcb one uses cairo 8fb9a6727e2b compiled with --enable-xlib-xcb.
>
> (The build process does a git pull.)
>
> They show the transparency lossage bug.

I can't replay the xlib trace, it complains about:

Error during replay, line 5134: invalid value (typically too big) for the size 
of the input (surface, pattern, etc.)


I tried to get an image out of the xcb trace. No idea if I got it right, but 
what I saw looked like https://bugs.freedesktop.org/show_bug.cgi?id=31931 (plus 
some clipped images). Could you check if you might be seeing the same bug?

> I'm currently uploading a seamonkey trace which ended with the
>
> ###!!! ABORT: RenderChangePicture: BadValue (integer parameter out of
>         range for operation); 317 requests ago: file
>         nsX11ErrorHandler.cpp, line 203
>
> ###!!! ABORT: RenderChangePicture: BadValue (integer parameter out of
>         range for operation); 317 requests ago: file
>         nsX11ErrorHandler.cpp, line 203
>
> warnings.  It will be:
>
>   http://people.freedesktop.org/~cloos/Cairo/traces/seamonkey-crash-xcb.trace.xz
>
> once it finishes.

Sorry, replays fine here on all enabled backends. Could you set the environment 
variable CAIRO_TRACE_FLUSH to 1 and grab another trace?

Cheers,
Uli