If box-shadow is applied to an element inside a CSS multi-column, the shadow is cut off at the column's boundary. Testcase to follow demonstrating box-shadow working correctly for floated elements but not elements in columns. Tested on the Windows Safari 3 Beta (522.11.3). Probably not platform specific but I don't have a Mac to test against, sorry.
Created attachment 15018 [details] Testcase for box-shadow boundary on elements in multi-column
I believe this is part of the "layers ignore columns" problem.
No, this is just a clipping problem. At one point the spec said to treat overflow like it was hidden on columns. I don't think it says that now though.
<rdar://problem/26613471>
Yes, I find this quite frustrating. Here's an example: http://labs.jensimmons.com/examples/image-gallery-multicolumn.html See how all the shadows from the top of the boxes are placed in the bottom of the previous column? And how the shadow from the bottom if the box of the longest column is placed at the top of the previous column? Bug. Here's a photo to be more obvious: https://monosnap.com/file/wCjt9v38xTPPTmGQJ8t42jvEgqioEM.png It's one of the many little multicolumn bugs that prevent people from trusting or using multicolumn. This happens in Mac Safari, iOS Safari, and Chrome (old webkit).
It appears this now fixed in Chrome and Firefox, probably due to recent updates, as it was still broken in all browsers (IIRC) just a couple of years ago. https://labs.jensimmons.com/2016/examples/image-gallery-multicolumn.html Screenshots for all three can be found here, and each is labeled with the name of the browser from which the screenshot was taken: https://imgur.com/a/HJBUiZW This is using Safari Technology Preview Release 161 (Safari 16.4, WebKit 18615.1.17.6)
Created attachment 468816 [details] rendering in safari, firefox, chrome
Created attachment 468817 [details] rendering in safari, firefox, chrome Removing the webkit prefixes for columns We get more or less the same rendering. There are slight differences, but there are more related to things inside the columns and not the shadows.
Karl, would you mind providing screenshots for all three browsers of the original example where the border was cut off? It was a very specific edge case.
(In reply to Brandon McConnell from comment #9) > Karl, would you mind providing screenshots for all three browsers of the > original example where the border was cut off? It was a very specific edge > case. I did (?) my last two comments: Using https://bug-14137-attachments.webkit.org/attachment.cgi?id=15018 with -webkit- https://bugs.webkit.org/attachment.cgi?id=468816 without -webkit- (so we can see the behavior with firefox) https://bugs.webkit.org/attachment.cgi?id=468817
@Karl, I think the issue here is that the shadow is cut off at the column's boundary and is still in the new images you added. Will this be addressed?
@Brandon, I don't see the difference in between browsers in the screenshot. Could you upload https://bugs.webkit.org/attachment.cgi?id=468817 but with a circle on the area you think is different in between browsers in that screenshot? Or are you saying the rendering is bad in the 3 browsers? In that case probably this also should be opened on the CSS Working Group. That will help me to understand as I do not see it now. :) Thanks.
Created attachment 470708 [details] comparison of box-shadows across CSS columns @Karl See the differences as rendered here in Jen's example: https://labs.jensimmons.com/2016/examples/image-gallery-multicolumn.html
Created attachment 470709 [details] comparison of box-shadows across CSS columns - outlined @Karl here is that same attachment, but with the problematic area outlined per your request
@Karl I'm not sure when/why this got marked "RESOLVED CONFIGURATION CHANGED". It's still broken. Same as always. In fact, I've been making new multicolumn demos over the last few weeks, and they all have this problem. It's not just the drop shadow, but also margins that end up in the wrong column.
Ha thanks! This is a lot clearer. Now I understand the issue. We need a different test case to make it simple and obvious. I'll see if I can come up with something later. @Jen I closed based on the initial test case which was added to the bug. I was not seeing what was wrong. I didn't see it either in your demo until https://bugs.webkit.org/attachment.cgi?id=470709 Cool! Thanks both of you.