Rolling Skia DEPS from r2633 to r2666

Rolling Skia DEPS from r2633 to r2666

Control group (whitespace-only change) is http://codereview.chromium.org/8538012
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=109639

[epoger, 2011-11-11 13:49:37]

Here's an attempt at the Skia DEPS roll.  If/when the trybot results look good,
I will request a reviewer.

[epoger, 2011-11-11 15:14:33]

On 2011/11/11 13:49:37, epoger wrote:
> Here's an attempt at the Skia DEPS roll.  If/when the trybot results look
good,
> I will request a reviewer.

Results look reasonable so far, but more are still coming in...

[reed1, 2011-11-11 15:20:42]

early bird lgtm

[Stephen White, 2011-11-11 15:35:36]

FYI:  I landed a suppression for the canvas test failure in upstream WebKit
r99899, which looks like it has (just) rolled into Chrome, so the downstream
suppression may no longer be needed.  I suggest we land the suppression anyway,
and remove it later if we don't need it.

[epoger, 2011-11-11 15:57:07]

I think this is OK to commit.  Mike and/or Stephen, could you please evaluate my
logic below?  (comparing this review's trybots to the control group)

- linux: red, but no worse than control group
- linux_chromeos: green
- linux_layout: Canvas test failure, but that's covered in the test_expectations
file (which the trybots ignore…)  Some apparently flaky tests in webkit_tests,
don't seem to be graphics-related.
- linux_layout_rel: only the canvas test fails, and that's expected
- linux_shared: green
- mac: still running, but Skia turned off there anyway
- win: some failures, but not graphics-related
- win_layout: only new failure is the canvas test
- win_layout_rel: only new failure (other than canvas) is
2d.text.draw.fontface.notinpage.html .  I am willing to assume that that's just
flaky.
- win_shared: green

[Stephen White, 2011-11-11 16:00:12]

LGTM