Per http://krijnhoetmer.nl/irc-logs/css/20130403#l-455 , the CSSWG has resolved that table cells should be painted as pseudo-stacking contexts. Filing this bug on implementing that change.

(I filed bug 858332 on the same change for flex items. Ideally, this bug and that bug should land around the same time, so that we're consistent on this behavior.)

Please note that dbaron later raised some issues: http://lists.w3.org/Archives/Public/www-style/2013Apr/0145.html I think we should wait until those are resolved.

Here's a simple testcase, for the record. Current Firefox has the word "First" printed on top of the yellow background here (intermixed with the word "Second"). So, we're painting all of the cells' background display items first, and then all of the cells' content display items (on top of all of the backgrounds). If we were treating them as pseudo-stacking-contexts, we'd expect the background & content to be painted atomically.

(In reply to Mats Palmgren [:mats] from comment #2) > Please note that dbaron later raised some issues: > http://lists.w3.org/Archives/Public/www-style/2013Apr/0145.html > I think we should wait until those are resolved. This is item #4 on the agenda for tomorrow's CSSWG meeting, BTW: http://lists.w3.org/Archives/Public/www-style/2013Apr/0381.html

Just kidding, it was deferred & will be revisited later: http://krijnhoetmer.nl/irc-logs/css/20130417#l-413

...though it sounds like Tab and fantasai have expressed support for "option b" from dbaron's email (making none of the backgrounds, not even the cell's, part of the cell's pseudo-stacking context): http://lists.w3.org/Archives/Public/www-style/2013Apr/0404.html http://lists.w3.org/Archives/Public/www-style/2013Apr/0405.html I'll proceed assuming that's what we're going with.

(though it's not yet clear what we should do w.r.t. the cell's background if the cell is styled with e.g. opacity to promote it to a an _actual_ stacking context: http://lists.w3.org/Archives/Public/www-style/2013Apr/0420.html ) So: I'm holding off on this at the moment, pending a more definite/complete resolution from the csswg.

Note that the thread continued in May, e.g.: http://lists.w3.org/Archives/Public/www-style/2013May/0660.html

Unassigning for the time being, since I'm not actively working on this, and since this isn't a priority at the moment (per the end of dbaron's www-style post in comment 8).

The group came to a resolution described in: http://lists.w3.org/Archives/Public/www-style/2013Jul/0081.html

And just to be clear, this is the sort of change that we should try making, but if it has bad Web compatibility implications we should just revert it and tell the working group to undo the change.