According Shadow DOM specification (https://dvcs.w3.org/hg/webcomponents/raw-file/tip/spec/shadow/index.html#shadow-root-attributes) "Accordingly, the nodeName attribute of a ShadowRoot instance must return "#document-fragment".". But in Chrome 23.0.1271.97 m ShadowRoot nodeName is "#shadow-root". Test: <html> <head> <script> function test() { var SR = window.ShadowRoot || window.WebKitShadowRoot; var host = document.createElement('div'); document.body.appendChild(host); var s = new SR(host); alert(s.nodeName); } </script> </head> <body onload='test()'> </body> </html>
Created attachment 180481 [details] Patch
If a bot became red, maybe I have missed some rebaselineing. Let's see it.
(In reply to comment #2) > If a bot became red, maybe I have missed some rebaselineing. Let's see it. Why don't you grep them instead of manually rebaselining them? We know that all instances of "#shadow-root" should simply be replaced by "#document-fragment", right?
Comment on attachment 180481 [details] Patch Clearing flags on attachment: 180481 Committed r138376: <http://trac.webkit.org/changeset/138376>
All reviewed patches have been landed. Closing bug.
This broke about 60 tests on Mac: http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r138378%20(5188)/results.html
(In reply to comment #6) > This broke about 60 tests on Mac: > http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r138378%20(5188)/results.html :-( rolling out...
Reverted r138376 for reason: Broke a bunch of tests on mac. Committed r138386: <http://trac.webkit.org/changeset/138386>
(In reply to comment #7) > (In reply to comment #6) > > This broke about 60 tests on Mac: > > http://build.webkit.org/results/Apple%20MountainLion%20Debug%20WK2%20(Tests)/r138378%20(5188)/results.html > > :-( rolling out... Don’t we need to just rebaseline them all?
This is why I said that you should just run grep in comment #3 so that you can rebaseline all expected results for all platforms.
This patch broke tests on Qt too - http://build.webkit.sed.hu/results/x86-64%20Linux%20Qt%20Release/r138379%20%2846257%29/results.html
Oops... Very sorry... Ryosuke, you are very right. Dimitry, sorry for bothering you. I'll check all the tests again. Also I have to have some method to check the Node is ShadowRoot in the port where SHADOW_DOM flag is not enabled. Maybe internals?
Created attachment 181866 [details] Patch
I wonder if having #document-fragment in all those test results will just be more confusing. The person looking at the test must make a mental jump to "oh, that's shadow root". Maybe we should present it differently (although that's also problematic, because we're special-casing shadow roots)?
Comment on attachment 181866 [details] Patch Clearing flags on attachment: 181866 Committed r139198: <http://trac.webkit.org/changeset/139198>