See attached .mov Anyone know what's causing that?

The tab has a width of 1px there; what you see is the CSS borders.

(In reply to comment #2) > The tab has a width of 1px there; what you see is the CSS borders. We should optimize the tab-close so that it vanishes after the tab is 3px wide, I think.

This hides the tab for the last 20ms of the closing transition and for the first 20ms of the opening transition. It also changes the transition target resp. start width from 1px to 0 (using 0.1px because min-width:0 would disable the min-width altogether). I tried 3px, but that wouldn't fix the bug, as that's still to narrow to draw the border properly, at least on Windows an Linux. It also caused a visible jump at the end of the transition.

Comment on attachment 483670 [details] [diff] [review] patch Can you add some comments? It took me a while to figure out why the different values were what they are. I'm assuming the idea is that (180ms delay)+(50ms "fade out" transition)+(20ms of opacity=0) is equal to the 250ms max-width transition length (for the closing case).