[css-parse] Getting this conversation rolling

It is inevitable, so I'd like to begin the discussion about parsing CSS or
CSS-like languages as one thing that would be necessary to whole families
of prollyfills.  I made a proposal for CSSOM a while back about how you
could add to standard CSSOM, but that didn't go anywhere because of unknown
rules handling and - I think because then Anne left...  CSSOM would be no
good for a preprocessor which is probably how many would want to push stuff
ultimately... Because of all of this I worked up a different idea[1] which
I actually used in another project (not hitch) and proposed you could
merely list under a different interface in the CSSOM API - I have the
parser, it is open-sourceable but while it is fast, it's really limited in
what it can/can't parse -  - after lots of threads back then I believe that
maybe at least a few FF and Webkit folks might think it is _plausible_ that
we could do at least one of those.

In any case, all of these CSS-related projects always wind up having to
start by parsing the CSS and I think that step 1 would be "define a
transform API" _something like_ (in vague terms) the one I lay out in that
email which could be describe CSS in easily parsed terms which could be
used for easy registration in all kinds of ways and for all kinds of
purposes in prollyfills.  Of course, you could lay a proposed CSSOM on top
of that if you wanted to...

http://lists.w3.org/Archives/Public/www-style/2011Jul/0501.html

This seems like actually a pretty interesting and fun place to start
conversations as I think it would help loop in some new/valuable people and
get the idea juices flowing...

Thoughts?