Bug 42495 – Canvas: Re-use existing CanvasStyle objects for color strings

• Status: RESOLVED
• Resolution: WONTFIX
• Component: WebCore Misc.

[Andreas Kling]

Setting fillStyle or strokeStyle on a CanvasRenderingContext2D is still quite slow.

We can optimize this further by re-using the current state().m_fillStyle or state().m_strokeStyle to house the new RGBA value.

[Andreas Kling]

Created attachment 61876 [details]
Proposed patch

Benchmarking with Hixie's skeleton test:
http://hixie.ch/tests/adhoc/perf/video/002.html

BEFORE:

Elapsed wall-clock time: 819ms (ideal: 640ms).
Elapsed non-idle time: 179ms (ideal: 0ms).
Speed: 18.32fps (ideal: 25.00fps).

AFTER:

Elapsed wall-clock time: 810ms (ideal: 640ms).
Elapsed non-idle time: 170ms (ideal: 0ms).
Speed: 18.52fps (ideal: 25.00fps).

[WebKit Review Bot]

Attachment 61876 [details] did not pass style-queue:

Failed to run "['WebKitTools/Scripts/check-webkit-style']" exit_code: 1
WebCore/html/canvas/CanvasRenderingContext2D.cpp:149:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 1 in 4 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Andreas Kling]

(In reply to comment #2)
> WebCore/html/canvas/CanvasRenderingContext2D.cpp:149:  More than one command on the same line  [whitespace/newline] [4]

Double semi-colon at end of line.

[Oliver Hunt]

Comment on attachment 61876 [details]
Proposed patch

Do you have perf numbers?

[Andreas Kling]

On second thought, I'm not sure this is such a great idea. It's adding a bit too much confusion to justify the barely measurable gain.

[Andreas Kling]

Comment on attachment 61876 [details]
Proposed patch

r-, not worth the added complexity.