See URL and its linked stylesheet for example. Hovering over the 42 causes the "detail" division to be displayed, but moving the cursor away causes only 42 to be repainted, leaving the rest of the division visible. Bug is only triggered if overflow on the containing block (class "cell" in the example) is switched from hidden to visible on :hover. The effect is present in at least google-chrome 11.0.696.57 beta (Linux SL6 gtk2-2.18.9-4), 11.0.696.48 beta (Linux Fedora 14 gtk2-22.0-1), and Safari 5.0.5 (6533.21.1) on Mac OS 10.6.7
Please attach a self-contained test case to this bug. Thanks.
Created attachment 92198 [details] Self-contained test-case
*** Bug 62641 has been marked as a duplicate of this bug. ***
Created attachment 147765 [details] simplified test-case Attaching simplified test-case (spotted this bug myself and made this test before finding this report). Swapping overflow of container AND swapping display of child node which is absolutely positioned apparently does this.
Created attachment 147766 [details] simplified test-cases' screenshot Screenshot of above test-case. Same appearance in Safari 5.1.5 @ W7 64; Chromium 21 @ Ubuntu, Google Chrome 19 ...
This is still an issue for me in Chrome Version 29.0.1547.65 in Sept' 2013. As reported in this stack issue, a workaround is to add -webkit-transform: scale3d(1,1,1); to force webkit to use the 3d rendering engine. http://stackoverflow.com/questions/11002195/chrome-does-not-redraw-div-after-it-is-hidden
Hi Team - I am unable to reproduce the following issue on Safari 15.4 and the behaviour on simplified test and self-contained is aligned with Firefox Nightly 102.a1 (20220528190153) behaviour. Just trying to help in reproducing old bugs to see, if they are valid any more. Although what seems to be bug is that on hover, if I select any text and then move cursor away, my text selection is now reset (for Safari) while in case of Firefox Nightly 102 and Chrome Canary 104. It is not the case, the text selection remain available upon re-hovering.
Thank you for checking! We'd need a separate bug to track the other issue, to prevent potential confusion.