[webkit-reviews] review granted: [Bug 107764] HTMLDocumentParser::insert should be aware of threaded parsing : [Attachment 185570] Patch

[bugzilla-daemon at webkit.org]

Eric Seidel <eric at webkit.org> has granted Adam Barth <abarth at webkit.org>'s
request for review:
Bug 107764: HTMLDocumentParser::insert should be aware of threaded parsing
https://bugs.webkit.org/show_bug.cgi?id=107764

Attachment 185570: Patch
https://bugs.webkit.org/attachment.cgi?id=185570&action=review

------- Additional Comments from Eric Seidel <eric at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=185570&action=review


I assume this causes a change in behavior?  Could you explain better in your
ChangeLog how this affects tests (because it doesn't seem they can pass yet w/o
implementing the speculation abort.)

> Source/WebCore/html/parser/HTMLDocumentParser.cpp:706
> +	       if (isWaitingForScripts() || isStopped())

Could you explain this change?	It looks right.  I assume you were pumping when
you didnt' expec to?

> Source/WebCore/html/parser/HTMLTreeBuilder.cpp:856
> +	   if (m_parser->tokenizer())
> +	      
m_parser->tokenizer()->setState(HTMLTokenizerState::RCDATAState);

Should this just be setTokenizerState() on the parser, and the parser should
handle the check?  That would make it possibly easier for you to compare the
state after insert() with the expected state.