Add testing of optimizations to GM

First pass at adding persistent local testing of pictureRecord optimizations.

[Justin Novosad, 2013-03-25 18:34:55]

You should also have tests for the cases where the pattern matches, but the
optimization aborts in merge_savelayer_paint_into_drawbitmp because the paints
can't be merged.

https://codereview.chromium.org/12843028/diff/1/gm/optimizations.cpp
File gm/optimizations.cpp (right):

https://codereview.chromium.org/12843028/diff/1/gm/optimizations.cpp#newcode198
gm/optimizations.cpp:198: SkASSERT(check_pattern(*post, postPattern));
This really feels like it should be a unit test rather than a gm.   What is the
use of running this test through all the different gm configs and canvas
flavors?

https://codereview.chromium.org/12843028/diff/1/include/core/SkPicture.h
File include/core/SkPicture.h (right):

https://codereview.chromium.org/12843028/diff/1/include/core/SkPicture.h#newc...
include/core/SkPicture.h:120: This flag disables all the picture recording
optimizations. It is
This is too vague. I.e. What specific optimization does it disable. It does not
disable kOptimizeForClippedPlayback_RecordingFlag

[robertphillips, 2013-04-01 18:08:52]

I have altered the GM so that (for each optimization) it tests 3 flavors that
fire the optimization and one in which the optimization fails.

Note that if the tile rendering enabled in
(https://codereview.chromium.org/13165011/) is enabled and the fix to the state
tree is disabled, this GM will reproduce the problem seen in impl-side painting.

https://codereview.chromium.org/12843028/diff/1/include/core/SkPicture.h
File include/core/SkPicture.h (right):

https://codereview.chromium.org/12843028/diff/1/include/core/SkPicture.h#newc...
include/core/SkPicture.h:120: This flag disables all the picture recording
optimizations. It is
I have changed the name and updated the comment - don't know if that is
sufficient.

[robertphillips, 2013-04-02 21:56:42]

Any more comments on this Justin?

[Justin Novosad, 2013-04-04 15:35:11]

https://codereview.chromium.org/12843028/diff/14001/gm/optimizations.cpp
File gm/optimizations.cpp (right):

https://codereview.chromium.org/12843028/diff/14001/gm/optimizations.cpp#newc...
gm/optimizations.cpp:345: SkASSERT(check_pattern(*pre, prePattern));
SkASSERT? Any way we could report these failures in release builds, and without
interrupting test execution?

[robertphillips, 2013-04-08 17:57:51]

How about this?

https://codereview.chromium.org/12843028/diff/14001/gm/optimizations.cpp
File gm/optimizations.cpp (right):

https://codereview.chromium.org/12843028/diff/14001/gm/optimizations.cpp#newc...
gm/optimizations.cpp:345: SkASSERT(check_pattern(*pre, prePattern));
On 2013/04/04 15:35:11, junov wrote:
> SkASSERT? Any way we could report these failures in release builds, and
without
> interrupting test execution?

Done.

[robertphillips, 2013-04-10 14:01:25]

PTAL

[Justin Novosad, 2013-04-11 14:53:57]

On 2013/04/10 14:01:25, robertphillips wrote:
> PTAL

lgtm

[commit-bot: I haz the power, 2013-04-11 22:01:03]

No LGTM from a valid reviewer yet. Only full committers are accepted.
Even if an LGTM may have been provided, it was from a non-committer or
a lowly provisional committer, _not_ a full super star committer.
See http://www.chromium.org/getting-involved/become-a-committer
Note that this has nothing to do with OWNERS files.

[robertphillips, 2013-04-12 14:53:51]

committed as r8654

[robertphillips, 2013-04-12 15:14:12]

committed in 8658 (was reverted in 8655)