RE: Spec suggestion: spell out serialization parsing context

On Thu, 2006-11-30 at 15:51 +0000, McBride, Brian wrote:
> > > > Should the spec spell out that context so that transform authors 
> > > > know what they can rely on?
> > > 
> > > I believe the reason why we moved away from language about 
> > parsing or 
> > > concrete syntaxes (someone plz correct me if I'm wrong) was to move 
> > > away from having a closed list of supported serialization syntaxes.
> > 
> > Yes, so this sort of detail is out of scope for the main body 
> > of the spec.
> 
> Which I think implies:
> 
> 1. the main body of the spec is not sufficient to guarantee
> interoperability transforms expressed in XSLT (because the
> interpretation of same doc references is not specified).
>
> 2. the test cases specify more than the main body of the spec.

Hmm... I'm not sure about that... we don't directly specify
how HTTP redirects work either, but via a chain of references
(via the URI spec and the IANA URI scheme registry...) it's
all in there somewhere.

If you can think of a test case where somebody could reasonably
use same doc references in XSLT in a different way from the the way
the tests say they work, I'd be interested.

It could come down to "reasonably"; i.e. indeed, the tests might
say more than the main body of the spec guarantees, but the
combination is enough to get GRDDL deployed in the Web. This
perhaps argues that we should make the test cases a normative
part of the GRDDL REC.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Thursday, 30 November 2006 16:39:23 UTC