Turn on Analytic AA in Chrome (Attempt 3)

Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2 https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001 (http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://chromium.googlesource.com/chromium/src/+/fae14ce1a1116a1d312a2a8639bf260d6fe50083

[liyuqian, 2016-10-31 15:32:34]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[liyuqian, 2016-10-31 15:32:57]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[liyuqian, 2016-11-04 15:49:42]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[liyuqian, 2016-11-04 15:49:42]

liyuqian@google.com changed reviewers:
+ fmalita@chromium.org

[liyuqian, 2016-11-04 15:56:11]

Hi Florin, actually I've changed the SkUserConfig.h a little because in the
future we will use the SK_NO_ANALYTIC_AA flag instead of SK_ANALYTIC_AA flag.

However, since I'm still fixing something inside Skia now so it's not ready yet.
Hence I'll leave the SK_ANALYTIC_AA flag and remove the SK_NO_ANALYTIC_AA flag.

[f(malita), 2016-11-06 12:43:20]

LGTM

The Chromium tree is now open, I'll go ahead and land manually with rebaselines.

[commit-bot: I haz the power, 2016-11-06 12:50:17]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/fae14ce1a1116a1d312a2a8639bf260d6fe50083
Cr-Commit-Position: refs/heads/master@{#430189}
==========

[commit-bot: I haz the power, 2016-11-06 12:50:17]

Patchset 2 (id:??) landed as
https://crrev.com/fae14ce1a1116a1d312a2a8639bf260d6fe50083
Cr-Commit-Position: refs/heads/master@{#430189}

[f(malita), 2016-11-06 12:51:42]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/fae14ce1a1116a1d312a2a8639bf260d6fe50083
Cr-Commit-Position: refs/heads/master@{#430189}
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed:
https://chromium.googlesource.com/chromium/src/+/fae14ce1a1116a1d312a2a8639bf...
==========

[f(malita), 2016-11-06 12:51:43]

Committed patchset #2 (id:20001) manually as
fae14ce1a1116a1d312a2a8639bf260d6fe50083 (presubmit successful).

[f(malita), 2016-11-06 13:09:39]

On 2016/11/06 12:51:43, f(malita) wrote:
> Committed patchset #2 (id:20001) manually as
> fae14ce1a1116a1d312a2a8639bf260d6fe50083 (presubmit successful).

  6293 files changed, 265 insertions(+), 241 deletions(-)

Note for sheriffs: the landed revision includes a massive rebaseline; if a
revert is needed, it'll also have to be manual (reverting this CL via Rietveld
won't help).  I'll check the tree and re-rebaseline any layout tests we missed. 
If you notice trivial anti-aliasing pixel diffs, feel free to rebaseline or
suppress and assign to me or liyuqian.

Thanks.

[qyearsley, 2016-11-06 18:18:09]

On 2016/11/06 at 13:09:39, fmalita wrote:
> On 2016/11/06 12:51:43, f(malita) wrote:
> > Committed patchset #2 (id:20001) manually as
> > fae14ce1a1116a1d312a2a8639bf260d6fe50083 (presubmit successful).
> 
>   6293 files changed, 265 insertions(+), 241 deletions(-)
> 
> Note for sheriffs: the landed revision includes a massive rebaseline; if a
revert is needed, it'll also have to be manual (reverting this CL via Rietveld
won't help).  I'll check the tree and re-rebaseline any layout tests we missed. 
If you notice trivial anti-aliasing pixel diffs, feel free to rebaseline or
suppress and assign to me or liyuqian.
> 
> Thanks.

Yep, as expected, some layout new baselines were missed, so there should be a
follow-up to rebaseline the remainder.

Example results:
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/168...

[f(malita), 2016-11-06 18:23:04]

On 2016/11/06 18:18:09, qyearsley wrote:
> On 2016/11/06 at 13:09:39, fmalita wrote:
> > On 2016/11/06 12:51:43, f(malita) wrote:
> > > Committed patchset #2 (id:20001) manually as
> > > fae14ce1a1116a1d312a2a8639bf260d6fe50083 (presubmit successful).
> > 
> >   6293 files changed, 265 insertions(+), 241 deletions(-)
> > 
> > Note for sheriffs: the landed revision includes a massive rebaseline; if a
> revert is needed, it'll also have to be manual (reverting this CL via Rietveld
> won't help).  I'll check the tree and re-rebaseline any layout tests we
missed. 
> If you notice trivial anti-aliasing pixel diffs, feel free to rebaseline or
> suppress and assign to me or liyuqian.
> > 
> > Thanks.
> 
> Yep, as expected, some layout new baselines were missed, so there should be a
> follow-up to rebaseline the remainder.
> 
> Example results:
>
https://storage.googleapis.com/chromium-layout-test-archives/WebKit_Win10/168...

Thanks, I've landed the expectations update, most of these should get
rebaselined soon: https://codereview.chromium.org/2479023002/

I'll check again after that auto-rebaseline lands.

[f(malita), 2016-11-07 14:40:51]

Just a note: this change (and its dependent rebaselines) got reverted in
https://crrev.com/568cf3e90941a42d1e85108417c385537c7d72ca due to
http://crbug.com/662914, http://crbug.com/662862.

[liyuqian, 2016-11-07 19:17:42]

Description was changed from

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed:
https://chromium.googlesource.com/chromium/src/+/fae14ce1a1116a1d312a2a8639bf...
==========

to

==========
Turn on Analytic AA in Chrome
(This is our third attempt, which has the same diff as our attemp 2
https://codereview.chromium.org/2459433003/)

Update cc_unittests png files to reflect our changes.

We also downgraded some ref tests to pixel tests because
there are some (dR <= 1, dG <= 1, dB <= 1) pixel difference on the
circle boundaries. This seems to be reasonable as the test html usually
has more layers than the ref html. In the future, we may set a
configurable threshold for ref tests so we don't have to downgrade ref
tests for this reason.

This is the clean version of the closed issue
https://codereview.chromium.org/2407393003/ (That issue has too many pngs
(7000+) that cause some troubles with the rietveld.)

We will review this CL, then manually patch all the remaining 7000+ pixel tests
locally, and land it manually.

BUG=

R=fmalita@chromium.org
TBR=ajuma

patch from issue 2407393003 at patchset 20001
(http://crrev.com/2407393003#ps20001)
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed:
https://chromium.googlesource.com/chromium/src/+/fae14ce1a1116a1d312a2a8639bf...
==========