- From: L. David Baron <dbaron@dbaron.org>
- Date: Thu, 18 Aug 2011 12:52:05 -0700
- To: Brian Manthos <brianman@microsoft.com>
- Cc: "Tab Atkins Jr." <jackalmage@gmail.com>, Γyvind Stenhaug <oyvinds@opera.com>, Alan Gresley <alan@css-class.com>, "www-style@w3.org" <www-style@w3.org>
On Thursday 2011-08-18 18:29 +0000, Brian Manthos wrote: > 'Gecko follows the rule that if some subproperties of a shorthand are inherit or initial, then in order for the shorthand to serialize as other than "", all must be.' > > Actually, wait, that's incorrect. Sub-values at the "set at > initial" state are represented by the absence of them in the > shorthand. That's only true for a few shorthands; each shorthand is different in terms of whether parts can be omitted and what happens to those parts if it's not there. And in no cases can a part be the 'initial' or 'inherit' keyword. So, given: shorthand: some value; longhand-within-shorthand: initial; one can't make a general statement that it's still possible to serialize a value of 'shorthand' representing the pair of declarations. In Gecko, we use 'initial' as our internal representation (and what we serialize) only when the 'initial' keyword was used to specify the value. For omission within shorthands, I made an explicit decision *not* to use 'initial' but instead to use the actual initial value, for the serialization reasons I described. -David -- π L. David Baron http://dbaron.org/ π π’ Mozilla Corporation http://www.mozilla.com/ π
Received on Thursday, 18 August 2011 19:52:56 UTC