- From: L. David Baron <dbaron@dbaron.org>
- Date: Thu, 9 May 2013 16:09:56 -0700
- To: "Tab Atkins Jr." <jackalmage@gmail.com>
- Cc: Simon Sapin <simon.sapin@exyr.org>, fantasai <fantasai.lists@inkedblade.net>, www-style list <www-style@w3.org>
On Thursday 2013-05-09 15:10 -0700, Tab Atkins Jr. wrote: > On Thu, May 9, 2013 at 3:03 PM, L. David Baron <dbaron@dbaron.org> wrote: > > On Thursday 2013-05-09 14:59 -0700, Tab Atkins Jr. wrote: > >> On Thu, May 9, 2013 at 2:56 PM, Simon Sapin <simon.sapin@exyr.org> wrote: > >> > Le 09/05/2013 23:51, fantasai a Γ©crit : > >> >> You need to cascade and compute writing-mode and direction first, > >> >> then cascade the logical properties. Antenna House Formatter has > >> >> an implementation of this, and helped us to specify it. > >> > > >> > Ah, I see. Interesting. Did implementers notice some runtime cost in doing > >> > the cascade in two passes? > >> > >> That's not necessary, under David's proposal. You can cascade them > >> together and then determine at computed-value time what property > >> affect which sides. > > > > No, it does require doing the cascade separately. But Gecko already > > does cascading separately for groups of properties, so it's fine for > > us. > > Hm, can you explain why? As far as I can tell, it should work fine to > cascade them together, so long as you're okay with the "real" values > only showing up at computed-value time. I should probably say this more clearly: * for LTR, margin-start and margin-left must be cascaded together * this requires that the direction be known before cascading margin properties * this means the cascade has two passes (which is what Simon was asking about), with direction being done in the first pass and margins (cascaded together) in the second pass -David -- π L. David Baron http://dbaron.org/ π π’ Mozilla http://www.mozilla.org/ π
Received on Thursday, 9 May 2013 23:10:25 UTC