Bugzilla

(In reply to Kestrel from comment #10)
> 1 second of jank attempting to open list before failure

So fwiw, the profile shows us spending a long time processing `mousedown`, and then when the `click` registers, that actually causes the panel to hide because the target of the click doesn't appear to require keeping the panel open. This bit is kind of confusing because I can't reproduce, but it's not clear to me why this doesn't just break reliably. Even if I insert a 1 second busy-loop in the processing of the mousedown that opens the view, I can't reproduce.

It looks like the mousedown handling is something specific to the list-all-tabs panel, implemented in bug 1489124. :-\

I _think_ we can just stick `closemenu="none"` on the toolbarbutton, or `type=menu` or something (but that might break other things). Will create a try build and ask you to test if I can't get this to repro locally.

(In reply to Kestrel from comment #10)
> 1 second of jank attempting to open list before failure

So fwiw, the profile shows us spending a long time processing `mousedown`, and then when the `click` registers, that actually causes the panel to hide because the target of the click doesn't appear to require keeping the panel open. This bit is kind of confusing because I can't reproduce, but it's not clear to me why this doesn't just break reliably. Even if I insert a 1 second busy-loop in the processing of the mousedown that opens the view, I can't reproduce. It looks like for me the target of the click is the `<panel>` instead of the toolbarbutton (presumably because it moves out of the way by the time that the mouseup happens because my clicking is too slow, or something).

It looks like the mousedown handling is something specific to the list-all-tabs panel, implemented in bug 1489124. :-\

I _think_ we can just stick `closemenu="none"` on the toolbarbutton, or `type=menu` or something (but that might break other things). Will create a try build and ask you to test if I can't get this to repro locally.