The exact same test is contained in WPT:
Created attachment 340582 [details]
Comment on attachment 340582 [details]
Clearing flags on attachment: 340582
Committed r231898: <https://trac.webkit.org/changeset/231898>
All reviewed patches have been landed. Closing bug.
It's not great to remove tests, because that loses all the history. In this case, how would one know that there was discussion in https://bugs.webkit.org/show_bug.cgi?id=68551 ?
I guess at least there's a link from WPT source. But it's so much harder to find.
(In reply to Alexey Proskuryakov from comment #5)
> It's not great to remove tests, because that loses all the history. In this
> case, how would one know that there was discussion in
> https://bugs.webkit.org/show_bug.cgi?id=68551 ?
> I guess at least there's a link from WPT source. But it's so much harder to
Yes, sorry about that. This came in the context of bug 177684, probably it would have make more sense for us to not apply the rule of splitting work into smaller patches here, so that we preserve the history link between the new file and old file (at least that's not worse than when people rename/move directories or files in WebKit). I'm adding a Bugzilla dependency to improve that a bit for future searches.
Anyway, I believe it makes sense to migrate to a WPT test and remove duplicate tests for this kind of bug reports, which could have been noticed earlier if the people who fixed the behavior in other browsers or clarified a spec had written new WPT tests and if one had sync CSSOM tests between WPT and WebKit.
For Web Platform development, it seems more helpful to me to link to interoperability tests, to the current specs or to other browsers' behaviors than to discussions that happened several years ago (not saying these discussions were not important of course).