IDs, Classes : could CSS usefully use a third (abstract) concept ?

Looking at a colleague's CSS recently, I couldn't help but
think that "good" CSS might be a lot easier to compose if
CSS were to support the concept of clustering.  By this
I mean a third type of entity, directly analogous to
IDs and Classes, but referenced solely by CSS, not by
HTML.  At the moment, if one wants to use the same
subset of properties in a number of environments,
one enumerates the environments (separated by commas)
and then enumerates the properties to be used by
those environments, as in :

	<tag 1>, <tag 2>, <tag 3>.<class>, tag 4#ID, <tag 5> <tag 6> ...
		<property 1>; <property 2>; ... <property n>

If those environments are also to have properties unique to
each, or to some subset of the original cluster, it is necessary
to enumerate them again, as in :

	<tag 1>, <tag 2> {<properties common to only <tag 1> and <tag 2>}

	<tag 1> {<properties unique to tag 1>}
	<tag 2> {<properties unique to tag 2>}


This is less than obvious to some CSS authors, who simply
duplicate the common properties over and over again in their
style sheets.

If, however, there were to be a third type of entity whose function
were solely to permit the clustering of common properties, and if the syntax
were to denote such entities by the use of an initial character such
as "&" (analogous to "." for classes and "#" for IDs), one could
then have rules such as :

	&<name 1> {<property 1>; <property 2>; ... <property n>}

The properties associated with <name 1> could then be used within
other CSS rules, as in :

	<tag 1>, <tag 2> {&<name 1>; <properties common to only <tag 1> and <tag 2>}

I would be interested to learn if this idea has been proposed previously,
and if so, what the consensus was.

Philip Taylor

Received on Thursday, 10 March 2005 18:01:50 UTC