Add Protobuf support for cc/playback skia usage

Add Protobuf support for cc/playback skia usage

Add protobuf serialization to the following:
- SkRegion::Op
- SkXfermode::Mode
- SkRRect

These are used by cc/playback DisplayItem classes.

In the short term, will just use Skia's flatten for:
- SkColorFilter
- SkPath
- SkPicture

BUG=541321
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Committed: https://crrev.com/543858f59534642b57822e34786ef337ad32014e
Cr-Commit-Position: refs/heads/master@{#355865}

[David Trainor- moved to gerrit, 2015-10-12 22:14:03]

dtrainor@chromium.org changed reviewers:
+ danakj@chromium.org, vmpstr@chromium.org

[David Trainor- moved to gerrit, 2015-10-12 22:14:03]

ptal!  Here are the basic skia protos required for serializing DisplayItems. 
This patch relies on https://codereview.chromium.org/1394353002/ which is also
up for review.

Thanks!

Next CL will add the DisplayItem protobufs and basic serialization.

[David Trainor- moved to gerrit, 2015-10-19 23:18:22]

rebased this onto the gfx patch that just landed.  thanks!

[David Trainor- moved to gerrit, 2015-10-21 15:03:28]

friendly ping :).

[vmpstr, 2015-10-21 21:58:53]

Sorry for the delay.

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
File cc/proto/skia_conversions.cc (right):

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.cc:43: DCHECK(false);
If you handle all of the cases, then don't add the default case. That way the
compiler will complain if the number of enum changes.

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.cc:63: DCHECK(false);
Sam here

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.cc:70: case proto::SkXfermode::Mode_Clear:
/o\

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.cc:129: DCHECK(false);
Same here.

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.cc:195: DCHECK(false);
Same here.

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
File cc/proto/skia_conversions.h (right):

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions.h:31: CC_EXPORT void SkRRectToProto(const SkRRect&
rect, proto::SkRRect* proto);
Why the inconsistency with the rest? The rest return the value instead of taking
it my ptr.

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
File cc/proto/skia_conversions_unittest.cc (right):

https://chromiumcodereview.appspot.com/1399933004/diff/60001/cc/proto/skia_co...
cc/proto/skia_conversions_unittest.cc:21: EXPECT_EQ(op,
SkRegionOpFromProto(SkRegionOpToProto(op)));
Thanks for this test :) But I did manually check everything seems to match!

[David Trainor- moved to gerrit, 2015-10-23 15:40:51]

no worries. thanks for doing all of these reviews!

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
File cc/proto/skia_conversions.cc (right):

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.cc:43: DCHECK(false);
On 2015/10/21 21:58:53, vmpstr wrote:
> If you handle all of the cases, then don't add the default case. That way the
> compiler will complain if the number of enum changes.

Ah good point!

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.cc:63: DCHECK(false);
On 2015/10/21 21:58:53, vmpstr wrote:
> Sam here

Done.

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.cc:70: case proto::SkXfermode::Mode_Clear:
On 2015/10/21 21:58:53, vmpstr wrote:
> /o\

Right!?  Crazy method :(.  But I need this in case we add/remove fields later. 
Otherwise outdated clients will just pull the wrong value.  In the future the
protobuf and the enum constants might drift but that's okay as long as we have
this.

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.cc:129: DCHECK(false);
On 2015/10/21 21:58:53, vmpstr wrote:
> Same here.

Done.

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.cc:195: DCHECK(false);
On 2015/10/21 21:58:53, vmpstr wrote:
> Same here.

Done.

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
File cc/proto/skia_conversions.h (right):

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions.h:31: CC_EXPORT void SkRRectToProto(const SkRRect&
rect, proto::SkRRect* proto);
On 2015/10/21 21:58:53, vmpstr wrote:
> Why the inconsistency with the rest? The rest return the value instead of
taking
> it my ptr.

Ah because the general use case for building proto messages is something like
the following:

SkRRectToProto(rect, message->mutable_skrrect());

For the other types they're just basic data, so there's nothing to build into. 
This format is actually what most of the gfx_conversions.h functions look like.

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
File cc/proto/skia_conversions_unittest.cc (right):

https://codereview.chromium.org/1399933004/diff/60001/cc/proto/skia_conversio...
cc/proto/skia_conversions_unittest.cc:21: EXPECT_EQ(op,
SkRegionOpFromProto(SkRegionOpToProto(op)));
On 2015/10/21 21:58:53, vmpstr wrote:
> Thanks for this test :) But I did manually check everything seems to match!

Acknowledged.

[vmpstr, 2015-10-23 17:29:47]

lgtm

[David Trainor- moved to gerrit, 2015-10-23 19:30:51]

The CQ bit was checked by dtrainor@chromium.org

[commit-bot: I haz the power, 2015-10-23 19:32:03]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1399933004/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1399933004/80001

[commit-bot: I haz the power, 2015-10-23 20:14:04]

Committed patchset #5 (id:80001)

[commit-bot: I haz the power, 2015-10-23 20:15:02]

Patchset 5 (id:??) landed as
https://crrev.com/543858f59534642b57822e34786ef337ad32014e
Cr-Commit-Position: refs/heads/master@{#355865}