Restore optimization in LayoutBlockFlowLine width computation

Restore optimization in LayoutBlockFlowLine width computation

The LayoutText::computeCanUseSimpleFontCodePath method determine whether
the wordMeasurement cache is sufficient to satisfy width computations in
LayoutBlockFlowLine, in addition to the more obvious role of controlling
which text code path to use. By short-circuiting the text path selection
this inadvertently disabled the wordMeasurement text cache optimization.

This change removes the short circuit and restores the previous behavior
where this optimization is decoupled from the text path selection logic.

Given its dual function the canUseSimpleFontCodePath name is missleading
and I intend to rename it in a subsequent change.

BUG=545147
R=drott@chromium.org

Committed: https://chromium.googlesource.com/chromium/src/+/f86d2d35e903274237f7df803b805bdf8949c4e4

[eae, 2015-10-20 15:19:16]

Description was changed from

==========
perf complex

BUG=
==========

to

==========
Restore optimization in LayoutBlockFlowLine width computation

LayoutText::computeCanUseSimpleFontCodePath is used to control whether the
wordMeasurement cache can be used to satisfy logical width calculations in
LayoutBlockFlowLine in addition to controlling which text code path to use. By
short-circuiting the text path selection this inadvertently ended up disabling
the wordMeasurement cache optimization.

This change removes the short circuit and restores the previous behavior where
this optimization is decoupled from the actual text path.

Given it's dual use the canUseSimpleFontCodePath method is poorly named and I
intend to rename it in a subsequent change. I'd like to keep this CL as small as
possible though.

BUG=545147
R=drott@chromium.org
==========

[eae, 2015-10-20 15:19:17]

eae@chromium.org changed reviewers:
+ drott@chromium.org

[drott, 2015-10-20 16:11:59]

With your explanation re renaming, LGTM. I wonder though, where are these layout
test failures coming from?

[eae, 2015-10-20 16:15:12]

On 2015/10/20 16:11:59, drott wrote:
> With your explanation re renaming, LGTM. I wonder though, where are these
layout
> test failures coming from?

Thanks

Slight changes in metrics due to the use of the cache. In each case the new
result match the expectation better than before (i.e. with this patch the two
lines that are supposed to be identical in 12 of the tests are).

[eae, 2015-10-20 16:29:00]

Description was changed from

==========
Restore optimization in LayoutBlockFlowLine width computation

LayoutText::computeCanUseSimpleFontCodePath is used to control whether the
wordMeasurement cache can be used to satisfy logical width calculations in
LayoutBlockFlowLine in addition to controlling which text code path to use. By
short-circuiting the text path selection this inadvertently ended up disabling
the wordMeasurement cache optimization.

This change removes the short circuit and restores the previous behavior where
this optimization is decoupled from the actual text path.

Given it's dual use the canUseSimpleFontCodePath method is poorly named and I
intend to rename it in a subsequent change. I'd like to keep this CL as small as
possible though.

BUG=545147
R=drott@chromium.org
==========

to

==========
Restore optimization in LayoutBlockFlowLine width computation

The LayoutText::computeCanUseSimpleFontCodePath method determine whether
the wordMeasurement cache is sufficient to satisfy width computations in
LayoutBlockFlowLine, in addition to the more obvious role of controlling
which text code path to use. By short-circuiting the text path selection
this inadvertently disabled the wordMeasurement text cache optimization.

This change removes the short circuit and restores the previous behavior
where this optimization is decoupled from the text path selection logic.

Given its dual function the canUseSimpleFontCodePath name is missleading
and I intend to rename it in a subsequent change.

BUG=545147
R=drott@chromium.org
==========

[commit-bot: I haz the power, 2015-10-20 16:32:49]

Patchset 2 (id:??) landed as
https://crrev.com/f86d2d35e903274237f7df803b805bdf8949c4e4
Cr-Commit-Position: refs/heads/master@{#355074}

[eae, 2015-10-20 16:33:14]

Committed patchset #2 (id:20001) manually as
f86d2d35e903274237f7df803b805bdf8949c4e4 (presubmit successful).