[cairo] Unbounded operators patch
cworth at redhat.com
Mon Aug 8 09:58:09 PDT 2005
On Mon, 08 Aug 2005 09:13:45 -0700, Carl Worth wrote:
> PS. Those are my comments on your notes. I'll look at the patch next,
> (though the results can speak for themselves quite a bit in a case
> like this).
I started by looking at results rather than code.
I've got an old copy of Xfake here that I've been using for testing,
(since it can PASS all of "make check"). Using that X server, and
what's in CVS, I have all of "make check" acting as expected for image
and xlib backends.
After applying this patch I see 8 failures. I'll break them down
These are new tests and the patch didn't have reference
images. So I'm assuming these are just correct results.
With xlib, only a small rectangle of the correct result
appears. Still passes with image backend.
With both image and xlib, the background border now appears
white instead of black as in the reference image.
With xlib, and the bottom three rows of the matrix, only the
right half of each result appears. With image, the result
still matches the reference image.
With image and xlib, only 2 squares are appearing instead of
3, and they are offset to the upper-left. This looks similar
to the mask-not-getting-transformed bug that inspired these
As in the case with the mask test, there are failures in xlib
with the last 3 rows. The first of these two rows are showing
the right-half only as in "mask". The final row is showing
just a small rectangle as in "clip-twice".
Owen, are you not seeing the same results? Am I missing any recent
required patches or something? Some of the xlib failures could be due
to the version of Xfake I'm using, but that wouldn't explain the
failures with the image backend.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/cairo/attachments/20050808/206855e0/attachment.pgp
More information about the cairo