Re: Verbosity of XHTML 2

Chris Mannall <chris.mannall@hecubagames.com> wrote:

>  > Personally I'd like to remove mandatory support for the fixed set
>  > of character entities in XHTML 2.
> 
> Consider my hand raised as another person who wouldn't be at all 
> affected by this. I've only ever used the character entities predefined 
> in XML anyway.

Good.

> Masayasu Ishikawa wrote:
>  > It's very dangerous to simply copy and paste them without
>  > understanding what those are dealing with.  Namespace prefixes
>  > are variables, if you copy some portion of XLinks from other
>  > document, it might be using a different prefix and you'd have to
>  > redeclare it or change prefixes, and if you redeclare it with a
>  > different prefix, unfortunately your document cannot be DTD-valid
>  > anymore.  I still have concern if ordinary people can manage this
>  > level of complexity.
> 
> I don't think this has to be a problem - any difficulties that could 
> arise from "ordinary people" not fully understanding the namespace 
> mechanisms are easily countered.
> 
> The problems you have mentioned so far basically boil down to the 
> problems of "incorrect" (i.e. undeclared) prefixes.

That's only the tip of an iceberg.  Note that redclaring a namespace
with different prefix in the middle of a document is perfectly OK
from Namespaces REC, it's the DTD which is incapable of managing it
correctly and putting an unreasonable burden on authors.  DTD could
describe certain subset of XHTML 2 but cannot describe all of them.
Unless Part 9 of ISO/IEC 19757, a.k.a. DSDL, could produce something
magical to deal with datatype- and namespace-aware DTDs, I can see no
good way to manage it reasonably.

I certainly see the value of validation but I remain unconvinced
whether DTD-validity should be imposed in XHTML 2, that's part of
the reasons why I'm playing with other alternatives to see which is
reasonable and least burdensome to both authors and implementors.

> If you wanted 
> to safeguard against the unlikely event that some authoring tool would 
> decide to aim solely at the Schema and use different prefixes, you could 
> even go so far as to make the DTD-specified prefixes part of the 
> specification itself; I see no reason why you couldn't require XHTML2.0 
> documents to bind the XLink namespace to a specific prefix, and so on.

We'll be battered to death about totally inappropriate use of
namespaces if we do so.  While developing Modulariazation of XHTML,
we were asked to introduce a mechanism to NOT hard-code namespace
prefixes.  Thanks to Murray Altheim's heroic effort, we somehow
managed to add that, but that has made modularization significantly
harder to understand, unfortunately.

Anyway, so far people seem to say that my worries are more or less
needless.  I sincerely hope that's true.

Regards,
-- 
Masayasu Ishikawa / mimasa@w3.org
W3C - World Wide Web Consortium

Received on Tuesday, 13 August 2002 04:05:26 UTC