tables/mozilla/bugs/bug1188.html needlessly depends on HTML Parser text node handling
Created attachment 58492 [details] Patch
Comment on attachment 58492 [details] Patch I've validated with hyatt and by reading https://bugzilla.mozilla.org/show_bug.cgi?id=1188 that it's OK to modify this test and that the modification does not change what the test was testing. As far as I can understand, the </table> in the table was causing some sort of corruption of mozilla's column cache.
Comment on attachment 58492 [details] Patch Ok. Shouldn't the new test fail in test-html5-parser? If so, where's the updated expectation?
Comment on attachment 58492 [details] Patch Nope. It fails with the new parser, but not the old one. The old one gets this case "right". The one one doesn't correctly coalesce text nodes yet. But for now that's an acceptable "regression" in behavior. Once we add text node coalescing we'll likely break more tests, but in the other direction. :)
Comment on attachment 58492 [details] Patch Clearing flags on attachment: 58492 Committed r61033: <http://trac.webkit.org/changeset/61033>
All reviewed patches have been landed. Closing bug.
Right, but test-html5-parser tests the new parser. You need to update the -html5-expected.txt results or else you're not testing what you think you're testing.
Oh true.
Created attachment 58507 [details] Patch for landing
re-opening so the commit-queue will see this to land.
Comment on attachment 58507 [details] Patch for landing Clearing flags on attachment: 58507 Committed r61054: <http://trac.webkit.org/changeset/61054>
http://trac.webkit.org/changeset/61054 might have broken SnowLeopard Intel Release (Tests)