Bug 43436 – Simple diffs for ~5% complex layout speedup

• Component: Text
• Status: RESOLVED FIXED

[Ned Holbrook]

Attached please find suggested changes that, using a simple test case of my own devising, improve by 4-5% the time to render a Roman string of 69 words using "text-rendering: optimizeLegibility". These changes have no effect on "run-webkit-tests --complex-text fast/text".

[Ned Holbrook]

Created attachment 63357 [details]
Proposed changes.

<rdar://problem/8237336>

[Darin Adler]

Comment on attachment 63357 [details]
Proposed changes.

This looks good.

> -        RetainPtr<CFMutableDataRef> m_coreTextIndicesData;
> +        Vector<CFIndex, 64> m_coreTextIndicesVector;

Maybe m_coreTextIndicesStorage would be a better name? Or m_coreTextIndicesBuffer? The name should communicate the fact that we should access this through the m_coreTextIndices pointer and not directly.

> +    m_coreTextIndicesVector.reserveCapacity(m_stringLength);

Since this is always used on a new vector, you could use reserveInitialCapacity for slightly better performance.

r=me without additional changes, but you might want to do one of the improvements above. If you decide not to do either improvement, please let a committer know so they can set commit-queue+ on the bug.

[Ned Holbrook]

(In reply to comment #3)
> Maybe m_coreTextIndicesStorage would be a better name? Or m_coreTextIndicesBuffer? The name should communicate the fact that we should access this through the m_coreTextIndices pointer and not directly.

Since the naming of the new member variable matches that of the existing m_glyphsVector and m_advancesVector, I think I'll leave this one as is.
 
> Since this is always used on a new vector, you could use reserveInitialCapacity for slightly better performance.

Sounds like a plan!

[Ned Holbrook]

Created attachment 63364 [details]
Changes per review.

[Eric Seidel (no email)]

Comment on attachment 63357 [details]
Proposed changes.

Cleared Darin Adler's review+ from obsolete attachment 63357 [details] so that this bug does not appear in http://webkit.org/pending-commit.

[WebKit Commit Bot]

Comment on attachment 63364 [details]
Changes per review.

Clearing flags on attachment: 63364

Committed r64734: <http://trac.webkit.org/changeset/64734>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.