[webkit-reviews] review granted: [Bug 90321] Convert HTML parser to handle 8-bit resources without converting to UChar* : [Attachment 152676] Updated Patch with another Speculative Fix for Chrome Linux

[bugzilla-daemon at webkit.org]

Oliver Hunt <oliver at apple.com> has granted Michael Saboff <msaboff at apple.com>'s
request for review:
Bug 90321: Convert HTML parser to handle 8-bit resources without converting to
UChar*
https://bugs.webkit.org/show_bug.cgi?id=90321

Attachment 152676: Updated Patch with another Speculative Fix for Chrome Linux
https://bugs.webkit.org/attachment.cgi?id=152676&action=review

------- Additional Comments from Oliver Hunt <oliver at apple.com>
View in context: https://bugs.webkit.org/attachment.cgi?id=152676&action=review


r=me, but make sure you checked that html parser perf is still okay.  Check
with eseidel, abarth, or antti as I believe that there is a html parser
benchmark in the repo somewhere.

> Source/WTF/wtf/text/WTFString.h:197
> +	   return m_impl->length() * (is8Bit() ? sizeof(LChar) :
sizeof(UChar));

If this turns out to be hot (i doubt it) we can make it branchless with length
+ length * !is8Bit().  But only if this is hot, as it would be hideously
non-obvious what was going on.

> Source/WebCore/platform/text/SegmentedString.h:81
> -    const UChar* m_current;
> +    int m_offset;

Hopefully this will actually result in a reduced struct size in 64bit, which is
nice.