controls missing if overflow: auto; and border-top-left-radius is set

All on trunk, Build identifier: Mozilla/5.0 (Windows NT 5.1; rv:2.0b8pre) Gecko/20101011 Firefox/4.0b8pre SeaMonkey/2.1b2pre

now passes w3c xhtml validation

http://validator.w3.org/check?uri=https%3A%2F%2Fbug603415.bugzilla.mozilla.org%2Fattachment.cgi%3Fid%3D482312&charset=(detect+automatically)&doctype=Inline&group=0 This document was successfully checked as XHTML 1.0 Transitional

See the same for firefox trunk Mozilla/5.0 (Windows NT 5.1; rv:2.0b8pre) Gecko/20101011 Firefox/4.0b8pre but firefox 20101011 doesn't

Bisect give me Due to skipped revisions, the first bad revision could be any of: changeset: 52297:c1cc7b565dc7 user: Zack Weinberg <zweinberg@mozilla.com>, L. David Baron <dbaron@dbaron.org> date: Thu Sep 09 08:21:47 2010 -0700 summary: Bug 451134 (1/2): switch to the official border-radius property names: purely mechanical changes. r=dbaron a2.0=blocking2.0:beta6 changeset: 52298:c954983caff1 user: Zack Weinberg <zweinberg@mozilla.com> date: Thu Sep 09 08:21:48 2010 -0700 summary: Bug 451134 (2/2): switch to the official border-radius property names: by-hand changes. r=dbaron a2.0=blocking2.0:beta6 changeset: 52299:8adb2f64c138 user: L. David Baron <dbaron@dbaron.org> date: Thu Sep 09 08:21:48 2010 -0700 summary: Add aliases for old names of border-radius properties. (Bug 451134) r=bzbarsky a2.0=blocking2.0:beta6 [command completed successfully Wed Oct 13 03:27:09 2010]

Before the changes you found with bisect, Firefox (and Seamonkey too, I would think) would have been ignoring your border-top-left-radius specification entirely. Could you change your test case so the style rule reads #content { overflow: auto; border-top-left-radius: 10px; -moz-border-radius-topleft: 10px; } and repeat the bisect? That will find the actual regression point.

Is that the same issue as bug 595842?

Possible. Bisect for a new testcase is running now. Will compare results with same from first post of bug 595842.

You'll get almost the same results for the bisect, since the patch for rounded clipping of overflow!=visible landed very close to the renaming. But I don't think it's worth the bother; this is a known bug.