- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Mon, 20 Sep 2010 10:11:09 -0700
- To: Paul Duffin <pduffin@volantis.com>
- Cc: Anne van Kesteren <annevk@opera.com>, Sylvain Galineau <sylvaing@microsoft.com>, www-style <www-style@w3.org>
On Mon, Sep 20, 2010 at 9:29 AM, Paul Duffin <pduffin@volantis.com> wrote: > Hopefully I am (enough of a grammar nerd ;-)) as I have checked and cannot find any issues. Some of the places are invalid according to CSS 2.1 tokenization (e.g. property name must be <IDENT>) but then we don't exactly follow that and if it was parsed by another CSS parser its error handling should just ignore them and carry on. My main concern was future enhancements that may have been discussed or planned. Right, error handling should be fine. Yay for the forward-compatible grammar! > I realise that you can't say 100% that it won't clash. If some feature does and we need to add support for the clashing part then I am sure that I will be able to work something out. You're pretty safe. '|' will *never* be used as a combinator (it's ambiguous), and it is basically certain to never be used anywhere else in selectors (because it would have to mean something different than what it does in types and attributes, which would be really bad). It's unlikely to ever be used in property names or values, or @-rules, simply because we already have a mechanism for qualified names (vendor prefixes) in the one situation in which they've proven necessary. I'll note that a few other suggested languages that lean on the CSS syntax (such as one I worked on with Eduard Pascal for extracting RDF triples from a page) *do* use | in property names (for the purpose of using XML Namespaces, because it drew property names from RDF syntaxes). It seemed to work okay there. ~TJ
Received on Monday, 20 September 2010 17:11:58 UTC