Created attachment 263613 [details] foreignObject-01.svg The reftest foreignObject-01.svg fails with a red area where the scrollbar would be if the image were large enough to cause scrolling. It seems that the lime div in the foreignObject is not taking the full width of the viewport for some reason. This began failing between the 2007-05-01-04-trunk and 2007-05-02-04-trunk builds. http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=SeaMonkeyAll&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2007-05-01&maxdate=2007-05-02+04%3A00&cvsroot=%2Fcvsroot
Could be bug 378975. We're now guessing that we should have a scrollbar to start with, and expecting the kid to tell us if we shouldn't. Is the kid lying in this case? btw, I won't be able to debug this because that testcase hangs X...
So thanks to bz's speedy debugging skills it turns out this is fixed by the patch in bug 369827. Because the foreignObject has percentage width, and because the SVG now gets the two stage reflow as described in comment 1, it gets stuck on the initial "assuming scrollbars" reflow width. Marking dependent so I remember to re-enable this test once the patch in that bug gets review.
(In reply to comment #2) > gets stuck on the initial "assuming scrollbars" reflow width. Marking dependent > so I remember to re-enable this test once the patch in that bug gets review. You should mark the reftest as "fails" in the manifest, so the tree will go orange if whoever checks in the fix forgets to remove the "fails".
> You should mark the reftest as "fails" in the manifest, so the tree will go > orange if whoever checks in the fix forgets to remove the "fails". Done, with the exception of the tests that are crashing the reftest script on linux. I'll have to leave them commented out I guess.
This was fixed by the checkin for bug 369827.