Lists Home |
Date Index |
I interpret it to favor the W3C specifications
even if they are works in progress over other
standards organization works even if these are
complete. That eliminates competitors if there
is a W3C candidate.
So we aren't to compare the relative merits of
each but take the W3C uber alles. That's bad
policy. It does let the Project Management
have the last say, and that is pretty much
what policy has always been. It does enable
justifications such as the wall-to-wall use
of one vendor's technology predisposing some
parts to be proprietary. Lots of wiggle room
They do need to clarify and they do need to
understand that some technologies will have
better alternatives for some cases. RELAXNG
is probably the star case for that.
From: Rick Jelliffe [mailto:firstname.lastname@example.org]
> Take a look at the Draft Federal Standard for
> XML Developers. Note that it prefers W3C specs
> over works from other organizations even if the
> W3C specs are works in progress. So in a policy
> document, the sort of thing Gosling mentions is
> indeed happening.
How do you get that? In
it merely says that full standards must be used, and where there
is competition, favour a W3C Recommendation. If there is a
W3C Proposed Recommendation, you may decide to favour
that if you commit to upgrading when the Rec comes out.
"Ensure" is a funny word to use, and may have some American public
service meaning, I suppose, but on the face of it, if there is no WXS REC
in sight, you should fallback to another standard, and if there are
no standards, you are then (naturally) in the province of proprietary or
custom applications. They need to clarify.
It also specifically mentions and allows ISO, ebXML and OASIS standards.
Indeed, if there is an ISO or OASIS standard and a proprietary
technology, people must choose the standard.
Then it goes on to say "use XML Schemas" (unless you are continuing
with SGML-ish DTD projects). This specifically shuts the door
on using DTDs and RELAX NG for being the published structure definition
languages for public schemas.
It seems that Schematron (assuming it is accepted as an ISO standard)
would fit in OK if used to augment a primary WXS schema. The mention
of annotations suggests that embedded Schematron is OK too. In particular,
it would not be a "proprietary extension" and could not be used to "define
structures": instead it is providing (executable) documentation on things like
links to external vocabularies and co-occurrence constraints, which are
out-of-scope for WXS.