Le 2007-05-17 à 13:54, Adrienne Travis a écrit :
If we could NAMESPACE the predefined class names, that'd actually
remove ALL my objections to the idea of overloading /class/ instead of
creating a new attribute (/role/ or whatever). But unfortunately, that
totally breaks backwards-compatibility if not done very carefully. (If
i declare class="dc:author" , i CAN'T address that in my stylesheet.
Well, you *can*, but it's not a pretty selector:
.dc\:author { }
or
*[class~="dc:author"] { }
If the namespacing was handled with dashes, like class="dc-author",
that would work, though.)
So I suggest doing it like this:
class="-dc-author"
which can be selected this way:
.-dc-author { }
The idea of the leading dash is to avoid potential clashes with
existing values: it's common today to separate two words inside a
class name with a dash, it's much less common to start a class name
with a dash. So perhaps the spec should reserve class names starting
with a dash for namespacing purpose and define the meaning of those
as the XHTML Role Module would have defined them with the semicolon
syntax.
I don't think in any way that this is incompatible with the
predefined class names that were just removed (which I liked), it's
just a little addition so you're not limited to the few predefined
ones. The XHTML Role Module also has a couple of predefined global
roles it seems.
Michel Fortin
[EMAIL PROTECTED]
http://www.michelf.com/