- From: Vasil Rangelov <boen.robot@gmail.com>
- Date: Fri, 18 Dec 2009 21:36:45 +0200
- To: public-xml-processing-model-comments@w3.org
2009/12/18 Florent Georges <fgeorges@fgeorges.org>:
> 2009/12/18 Vasil Rangelov wrote:
>
>> Though to be honest, if all enumerated option values were QNames, it would
>> be nice indeed. The more extensibility hooks, the better ^_^ .
>
> I think XSLT had a clever approach to that, for instance for
> xsl:output's @method. It can be any QName except an NCName, which are
> reserved for the recommendation itself. So standard values are
> simple, they look like simple strings, without any namespace stuff
> overhead ("text", "xml", etc.) while an implementation can provide
> more values in a clean way: by using a namespace-based mechanism.
>
> --
> Florent Georges
> http://www.fgeorges.org/
>
Yeah. This is exactly what I had in mind. And again, not just the
auth-method, but any enumerated option value or c:* stuff.
Some examples for that, which I collected after a quick glance on the spec:
1. c:request's method attribute.
2. p:insert's position option.
3. p:namespace-rename's apply-to option.
4. p:validate-with-xml-schema's mode option.
Regards,
Vasil Rangelov
Received on Friday, 18 December 2009 19:43:42 UTC