Looks like the test for fragment'ness when a synthesized <html> element for an innerHtml fragment no longer functions properly in HTMLHtmlElement::insertedByParser().
// When parsing a fragment, its dummy document has a null parser.
if (!document()->parser() || !document()->parser()->documentWasLoadedAsPartOfNavigation())
That test no longer weeds out these elements so the method continues when it shouldn't. This has bad consequences for the appcache system.
See http://code.google.com/p/chromium/issues/detail?id=72986 for some history.
Working on it. I actually think this will be fixed by the patch I'm uploading to https://bugs.webkit.org/show_bug.cgi?id=48719, which will not create an HTML element during fragment parsing.
Michael, can you verify that this assertion is no longer hit in a build with r80231?
Sorry, I had assertions on the mind. I meant to say bug.
Sure, when we roll this revision into view for chrome i'll take a look.
Looks good to me. I'm no longer seeing multiple invocations of selectCache() for html fragments.
(In reply to comment #6)
> Looks good to me. I'm no longer seeing multiple invocations of selectCache() for html fragments.
Thanks for checking Michael!