Re: [css-parse] Getting this conversation rolling

On Tue, Nov 20, 2012 at 3:25 PM, Clint Hill <clint.hill@gmail.com> wrote:

> I agree that generically we will want some kind of CSSOM transform API.
> And yes inevitably that will require a parser.
>
> I have been working on a Parsing Expression Grammar generator (very slowly
> and very leisurely) and it is designed for this purpose - create a simple
> way to build parsers and extend grammars with object models. It's not new
> and in fact I took great ideas from a lot of other implementations (more
> info about PEGs here: http://bford.info/packrat/).
>
> Whether my parser is helpful here or not, I believe it is a core feature
> of the larger "engine" that we've discussed and will go a long ways to help
> define Prollyfills in an object model.
>
>
> Just so we can keep it on the list, Francois mentioned that Tab Atkins
(CSS WG) already has a parser in JS (https://github.com/tabatkins/css-parser)
and Daniel Glazman (co-chair of CSS WG) has one as well.  In truth - there
are more too - but the majority of them do something like what Tab's does
and basically creates a token stream - which still is pretty low level upon
which to be building most prollyfills IMO.  I have looked at Daniels a few
times - while there are no docs, it _sounds_ to me like maybe his does
something more like what I was suggesting in
http://lists.w3.org/Archives/Public/www-style/2011Jul/0501.html in that it
creates an object model that you can use as input to something else or even
reverse back out to CSS.

I know that from working with me on other projects, Clint knows what I am
looking for here - but I'm not sure whether anyone else does... Does it
make sense why you would want something like this or does it need more
explanation?