[webkit-reviews] review granted: [Bug 50516] Avoid strlen() in AtomicString::fromUTF8 : [Attachment 79110] Patch

[bugzilla-daemon at webkit.org]

Darin Adler <darin at apple.com> has granted Patrick R. Gansterer
<paroga at paroga.com>'s request for review:
Bug 50516: Avoid strlen() in AtomicString::fromUTF8
https://bugs.webkit.org/show_bug.cgi?id=50516

Attachment 79110: Patch
https://bugs.webkit.org/attachment.cgi?id=79110&action=review

------- Additional Comments from Darin Adler <darin at apple.com>
View in context: https://bugs.webkit.org/attachment.cgi?id=79110&action=review

> Source/JavaScriptCore/wtf/unicode/UTF8.cpp:334
> +	       if (dataLength)
> +		   *dataLength += 1;

Seems better to not have if statements like this inside relatively hot loops.
We should instead have callers supply a place to put the length even if they
won’t use it.

> Source/JavaScriptCore/wtf/unicode/UTF8.cpp:369
> +	       if (utf16Length)
> +		   *utf16Length += 2;

No reason to have these branches in since utf16Length is never null.

> Source/JavaScriptCore/wtf/unicode/UTF8.h:75
>      unsigned calculateStringHashFromUTF8(const char* data, const char*
dataEnd, unsigned& utf16Length);
>  
> +    unsigned calculateStringHashAndLengthFromUTF8(const char* data,
unsigned& dataLength, unsigned& utf16Length);

I’d group these together rather than leaving a blank line.