This issue manifesst on the removal of nodes as well as when changing size properties of nodes, this examples uses YUI for the complex animation but can be easily demonstrated with removeNode(), which it does in the first demo of the problem. Steps to reproduce: 1) It seems that size does indeed matter. You need to first resize your window to a comfortable size for IMing. For best viewing results, you'll want to make sure that both actions are viewable in the same window. 2) Scroll to bottom 3) Do Nothing. The timers will fire at 4 and 10 seconds removing the first and then the second action div. You can also trigger the animations by clicking on the action divs, however it seems that doing it this way does not cause the layout error for the first action. Notes: The images aren't necessary, but it makes it easier to see the graphical glitch for me. The use of YUI also isn't necessary as demonstrated by the first example, but it's included because it's one of the larger uses that this glitch is blocking me on. Additional Note: Firefox messes up the text layout in the last message bubble after I animate the removal of the second action. WebKit does not.
One issue visible in the test case is bug 13262, but without further reduction it is impossible to say if there are additional bugs here.
I expect this is related to the caching of absoluteOverflowClippedRects and layout() not being called often enough during the animation to update them. Thus the rect that the divs are invalidating during layout would be wrong. But that's just a guess.
I am unable to find the test URL on Wayback Archive. Plus we don't have any reduction to reproduce this bug. Plus there is no concrete reason, where this bug occur and it is outdated by several years in Comment 02, should we close it and focus on other bugs or we continue to keep it open?
(In reply to Ahmad Saleem from comment #3) > I am unable to find the test URL on Wayback Archive. Plus we don't have any > reduction to reproduce this bug. > > Plus there is no concrete reason, where this bug occur and it is outdated by > several years in Comment 02, should we close it and focus on other bugs or > we continue to keep it open? Yeah I agree, let's just close this.