Re: Proposed changes to address issue-dbooth-3 (ambiguity)

Booth, David (HP Software - Boston) wrote:
> In the example below, the DTD clearly shows that this document was
> designed with GRDDL in mind and hence was *designed* to produce this
> variability.


Well yes, this was a specially produced test designed to reflect the 
variability, but no it doesn't follow that this is the only case.

In particular, the DTD allows g:transformation, but needn't for the 
general point to be made, which could have been made with namespace 
transformations alone.

Much closer to my heart is to do with HTML processing. I decided against 
producing an HTML test example mainly because the Jena GRDDL Reader is, 
by design, very liberal in its HTML processing, and doesn't have, but 
probably should have, a strict mode, where it does precisely what the 
spec says and no more.

So an example doc would be, a valid HTML doc, either referencing the 
GRDDL profile, or a GRDDL enabled profile, but omitting, say the xmlns 
declaration on the root element. This declaration is #FIXED (I believe, 
in the DTD), so that a validating parser would trigger GRDDL transforms.

Typically such GRDDL transforms, particularly a profile transform, are 
not expecting <html> input, but <xhtml:html> input. I see no reason why 
we should expect such transform authors to duplicate their code, or 
write code that is otherwise robust against this variability.

FWIW, I have now run the test with the Jena GRDDL Reader, and by default 
it appears to have validation on ... I need to dig down in the 
documentation to find the option to switch validation off. More on that 
later.

Jeremy

-- 
Hewlett-Packard Limited
registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Tuesday, 19 June 2007 14:02:17 UTC