Two things would be needed here to unskip this test: * Implement isExpanded in GTK's DRT * Modify the layout test so it just prints out rolenames and leave the responsibility of checking whether they're correct to the, platform-dependent, expected files. This would imply updating the mac expected file, as it's the only platform providing one.
Created attachment 91020 [details] Patch
Created attachment 91022 [details] Patch Sorry, stupid mistake in the ChangeLog in previous patch
Created attachment 91025 [details] Patch Argh! God dammed jet lag! (read as: "this is finally the good one")
Comment on attachment 91025 [details] Patch OK. Seems odd. that we would have these different expectations for gtk, but I Think I understand why.
Comment on attachment 91025 [details] Patch my only comment is that we prefix the role so that looking at a failed test it will be a little clearer ie) debug("Role: " + combobox.role);
(In reply to comment #5) > (From update of attachment 91025 [details]) > my only comment is that we prefix the role so that looking at a failed test it will be a little clearer > ie) > debug("Role: " + combobox.role); I agree with this. I'll add that prefix (and update expectations) right before committing then. Thanks
Committed r84975: <http://trac.webkit.org/changeset/84975>