Bug 39755 – Make HTML5 lexer not ASSERT when resuming partial parses

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Adam Barth]

Make HTML5 lexer not ASSERT when resuming partial parses

[Adam Barth]

Created attachment 57122 [details]
Patch

[Eric Seidel (no email)]

Comment on attachment 57122 [details]
Patch

This is an excellent test to have.  However, I'm confused by the output.  Are we running each script more than once?

 You should add a comment next to m_token as to why we need to hold it as a member.

isEndTagBufferingState is almost big enough to justify using a hash.  Unclear.

Ok, so I love the patch.  My big concern is the test.  Why is this forked from runner/webkit_runner.  Can we just add this functionality to one of those, or does this need to be a copy of the javascript forever?

r- for the above nits.

[Adam Barth]

> This is an excellent test to have.  However, I'm confused by the output.  Are we running each script more than once?

Yes.  We run each test multiple times: once for each position where we could interrupt the parser.

>  You should add a comment next to m_token as to why we need to hold it as a member.

Ok.

> isEndTagBufferingState is almost big enough to justify using a hash.  Unclear.

A hash?  That seems like overkill for eight branches.

> Ok, so I love the patch.  My big concern is the test.  Why is this forked from runner/webkit_runner.  Can we just add this functionality to one of those, or does this need to be a copy of the javascript forever?

Well, once we get external scripts working, we can pull out all the common bits into external JavaScript files.  ;)

[Eric Seidel (no email)]

Comment on attachment 57122 [details]
Patch

OK.  r+ with revision then. :)

[Adam Barth]

Committed r60253: <http://trac.webkit.org/changeset/60253>