On 06/02/15 17:20, Louis-Dominique Dubeau wrote:
> On Fri, 2015-02-06 at 15:35 +0000, Lou Burnard wrote:
>> What's wrong with using Schematron for the additional constraint?
> There's nothing inherently wrong with that. I prefer to have the check
> in the schema as much as possible. Most of the tools I use work better
> this way. Most of the processing I do is better optimized if I can catch
> problems as early as possible, which means catching problems at the
> stage of checking that the document is valid against a schema.
Martin's original concern was whether he could count on xml:base
completion for URIs. Schematron wouldn't help in achieving that, because
it only looks at the surface, while Martin needs to inform the parser
about the relevant datatypes (and even that might not be enough for some
parsers to perform the URI completion).