Here's the stack: get_scrollClientRect@chrome://global/content/bindings/scrollbox.xml:131:18 ensureElementIsVisible@chrome://global/content/bindings/scrollbox.xml:241:15 onxblunderflow@chrome://global/content/bindings/scrollbox.xml:765:13 This happens when increasing the width of a browser window. I don't understand how this action could cause the current tab to become invisible, because if there's no overflow anymore, all tabs are visible, right? The relevant code is at http://searchfox.org/mozilla-central/rev/a7334b2896ed720fba25800e11e24952e6037d77/toolkit/content/widgets/scrollbox.xml#762
Can we just remove these lines? http://searchfox.org/mozilla-central/rev/a7334b2896ed720fba25800e11e24952e6037d77/toolkit/content/widgets/scrollbox.xml#762-764
Not sure. This could be a workaround for weird layout behavior like http://searchfox.org/mozilla-central/rev/214345204f1e7d97abb571b7992b6deedb5ff98f/browser/base/content/tabbrowser.xml#5945-5953
Depends on: 1356705
So should we just try and ask QA to verify with some exploratory testing? Or is bug 1356705 going to fix this?
(In reply to Florian Quèze [:florian] [:flo] from comment #3) > So should we just try and ask QA to verify with some exploratory testing? Fine by me. > Or is bug 1356705 going to fix this? Given the current WIP patch, it would remove the layout flush, but leave behind this potentially useless code.
Assignee: nobody → florian
Status: NEW → ASSIGNED
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/mozilla-inbound/rev/3e59d06de614 remove potentially useless ensureElementIsVisible call (that currently causes a sync reflow) when an arrowscrollbox handles an underflow event, r=dao.
No longer blocks: photon-performance-triage
We requested QA here because we don't really know what the lines we removed here were doing. It's code related to one (or more?) tab(s) making us leave the overflow mode on the tab bar.
Whiteboard: [ohnoreflow][qf][photon-performance] → [ohnoreflow][qf][photon-performance][qa-commented]
Removing the qe+ flag from this issue since the change is from 55 iteration + the areas that might've presented regressions have been already covered in different bug verification or exploratory/regression testing.
You need to log in before you can comment on or make changes to this bug.