On Jan 26, 2015, at 12:49 PM, Robert O'Callahan <[email protected]> wrote: > > We need a resolution here sooner rather than later, please. It'd be great if > someone from the IE or Chrome teams could contribute to the discussion... > > A bug related to this issue just cropped up in the wild: > https://bugzilla.mozilla.org/show_bug.cgi?id=1125767 > <https://bugzilla.mozilla.org/show_bug.cgi?id=1125767>, > http://jsfiddle.net/lastw/auv3x2y9/ <http://jsfiddle.net/lastw/auv3x2y9/>. In > that bug, a filtered element E has an abs-pos descendant D for which E is not > a containing block ancestor. In Gecko, D is cropped to E's bounds. In Chrome, > D is not filtered. I think both of those behaviors seem wrong.
Sad as it is to diverge from opacity, I slightly prefer this solution: > 1) Specify that 'filter' is a containing block for all positioned descendants > (like 'transform' already is). I would, however, like to hear from web devs about whether it’s surprising that applying the ‘filter’ property would suddenly affect the layout of positioned descendants. Simon
