See http://codepen.io/abucur/pen/JtevB/
Created attachment 228503 [details] Patch
Created attachment 228511 [details] Patch v2
Comment on attachment 228511 [details] Patch v2 I do not agree with the approach taken in this bug and feel it's too intrusive into RenderBlock's layout (and FloatingObjects). It seems like you should be able to come up with an approach that is contained in the flow thread/region code.
Created attachment 228843 [details] Patch v3
Comment on attachment 228843 [details] Patch v3 Ok, so I finally understand why you need this code. The basic issue is that regions have been improved to handle monolithic elements differently from printing/columns. When a monolithic element overflows a region, you're actually rendering the element entirely within the region it overflows. This is great, and I wish printing and multicolumn behaved this way as well. I could use a reminder of where the code lives that does this for when printing/multicolumn change to behave like this in the future. Anyway, I think we need a comment explaining that the reason this function exists is because monolithic elements do not get split across regions, and note that this is a difference from printing/multicolumn in the comment. The function will also need to be tweaked to check for flow threads/regions, etc. to make sure you don't turn this on for printing or multi-column (since right now the gap makes sense for those systems).
Created attachment 228959 [details] Patch which implements reviewer feedback
Comment on attachment 228959 [details] Patch which implements reviewer feedback r=me
Comment on attachment 228959 [details] Patch which implements reviewer feedback Clearing flags on attachment: 228959 Committed r167018: <http://trac.webkit.org/changeset/167018>
All reviewed patches have been landed. Closing bug.
The changelog says you've modified FloatingObjects.cpp, but I don't see it in the patch.
Yes, it was in the first version of the patch and after changing it, it seems I forgot to update the CL. Sorry.