Re: [css-syntax] value grammar, <value> type and browser implementations

On Fri, Feb 8, 2013 at 3:27 AM, François REMY
<francois.remy.dev@outlook.com> wrote:
> ± This is correct - the actual grammar we use to parse CSS is a bastardized
> ± version of Appendix G.  David's written a bunch of tests which we fail and will
> ± continue to do so until we fix our parser.
>
> Seems like I made a good guess ^_^ I suppose the plan is to get rid of those issues before releasing Custom Properties, right? Hope this can be solved in all browsers rather quickly. Do FireFox/Opera also have some non-standard CSS parsers?

I do plan to try and rewrite our parser soon to match Syntax.

I think FF has a better parser, though I don't know the details.  Same
with Opera.

> ± Serialization is undefined. :/
>
> It would be good to get this defined before CSS Custom Properties goes on (if not in a spec at least discussed among implementers just to avoid a situation where serialization is completely different in all browsers, that would be a mess).

This is supposed to be something that happens in CSSOM.

> ± Yes, we don't keep around the original text, for good reason. But JSCSSP's
> ± serializer seems slightly wrong.  Unfortunately, it's in a way that doesn't
> ± matter for anything that isn't a custom property. ^_^
>
> What do you think JSCSSP is doing slightly wrong exactly?

Exactly what you said - inserting whitespace in the serialization that
didn't appear in the original.

~TJ

Received on Saturday, 9 February 2013 16:58:40 UTC