Re: Found a solution for CSS -> XML mapping: JXCSS

On 5/11/07, Jo Rabin <jrabin@mtld.mobi> wrote:
> 3. I couldn't find a DTD or schema for either xcss or compact xcss, I
> couldn't see in the compact-xcss any hint of at-rule processing, so how
> complete do we think it is?

Yeah that may need to be tacked on.

> 4. Do we think that the Apache license is compatible with whatever
> license we will apply to our work (and what is that license?)

Good question. We are releasing under the W3C license. My impression
is that it's substantially like the Apache license -- you can link it,
change it, distribute it with the proper note. It's worth a question
to Dom and the legal team, though I see for example Jigsaw including
Xerces in its distribution and all that. I'd be surprised if there
were any problem.

> 5. Don't see any sign of widespread adoption, so it probably wouldn't
> break too much to refine it?
>
> I take it that as it is SAX based it would be relatively trivial to plug
> in an alternative renderer (replacement for XCSSSerializer?), in which
> case one could fix up the format to be a little more elegant and concise
> and complete, without too much pain?

Indeed, SAC itself provides a SAX-like parser directly with which you
can output what you like. This JXCSS package is mostly a bridge
between the two. You can also transform the XML with XSLT into another
form.

> Another thought is that perhaps it would be more convenient for our
> purposes to have the selectors as children of the properties, rather
> than the properties as children of the selectors. i.e. invert the
> structure.
>
> The bottom line is, how far do we want to go and why?

I think it was suggested since a DOM representation can be queried
with XPath and transformed with stylesheets. If the format suits that
purpose, stick with it; that's the only criteria.


> 7. There have been a lot of false starts in this area, and I am reminded
> of the saying "fools step in where angels fear to tread". I think we
> should check with the CSS folks before committing and/or proceeding.

Sounds fine to me. We're not publishing any canonical XML
representation here, just creating something for internal use, so I
mostly concerned with whether the representation is suitable for
everyone's purposes.

Our purposes are fairly limited: what properties are used, and, are
any of them using absolute measures? That seems to be queryable in
most any representation you can imagine.

Received on Friday, 11 May 2007 13:59:17 UTC