- From: Christian Biesinger <cbiesinger@google.com>
- Date: Wed, 22 May 2013 11:49:48 -0700
- Cc: www-style list <www-style@w3.org>
On Tue, May 21, 2013 at 7:17 PM, Boris Zbarsky <bzbarsky@mit.edu> wrote: > On 5/21/13 9:45 PM, Christian Biesinger wrote: >> >> "Some values of ‘display’ trigger the generation of anonymous boxes. >> For example, a misparented ‘table-cell’ child is fixed up by >> generating anonymous ‘table’ and ‘table-row’ elements around it. >> [CSS21] This fixup must occur before a flex container's children are >> promoted to flex items." > > > This seems quite clear, for what it's worth. > > First you do the fixup for table stuff, then you figure out what your flex > items are, then you update the computed display values of flex items... > > The biggest problem here is that this makes computed display depend on the > box tree in some sense, but you can actually statically determine at > computed style time what the box tree will be for purposes of this stuff. > > I do agree it might be even clearer to say that no computed display coercion > happens on internal table display types, though. Hm... in that case I'd like to ask, why is this behavior desirable? It seems like lots of complexity for little gain compared to just promoting the display types to block. -christian
Received on Wednesday, 22 May 2013 18:50:15 UTC