- From: <bugzilla@wiggum.w3.org>
- Date: Wed, 04 Feb 2009 20:39:30 +0000
- To: www-xml-schema-comments@w3.org
http://www.w3.org/Bugs/Public/show_bug.cgi?id=6522
--- Comment #12 from John Cowan <cowan@ccil.org> 2009-02-04 20:39:30 ---
(In reply to comment #2)
> (Speaking only for myself, not for the WG.)
>
> John, thank you for calling this issue to our attention.
>
> For the record, the change in question was adopted by the XML Schema
> WG at a call on 13 January 2006 as a resolution of bug 2214 (q.v.),
> with the relatively laconic comment in the minutes that
>
> Point of information: QT not referencing it, semantic web best
> practices has no reference from their document, Googling returns
> references, but seem to be from tutorials.
>
> MSM found one live use in metadata for collection of (geospatial?)
> data. And more uses in drafts of other specs, but newer drafts
> removed.
>
> (http://lists.w3.org/Archives/Member/w3c-xml-schema-ig/2006Jan/0036.html
> [member-only link])
>
> From the wording in the minutes, it appears that the rationale for
> thinking the change would be harmless to the world might be
> effectively undercut by a reference to a normative document for Relax
> NG that does refer to the namespace in question. A quick Google
> search for the namespace name turns up a Relax NG tutorial at
> http://relaxng.org/tutorial-20011203.html but no normative document.
>
> John, can you possibly oblige with a reference? It would perhaps help
> persuade members of the WG that the facts are not now as we thought
> they were when the decision was made.
The bibliographical reference is ISO/IEC 19757-2:2003/Amd.1:2006(E), clause
C.4, Declarations. The relevant text reads:
A datatypes declaration declares a prefix used in a QName identifying a
datatype. For example,
datatypes xsd = "http://www.w3.org/2001/XMLSchema-datatypes"
element height { xsd:double }
In fact, in the above example, the datatypes declaration is not required: the
xsd prefix is predeclared to the above URI.
> I think the reasoning behind the change has been ably summarized in
> comment 1 by Michael Kay, at least as far as I have so far been able
> to remember it, under the influence of the report at bug 2214 and the
> minutes linked from there.
>
> If we do wish to reverse the deprecation, perhaps the right way to
> make the magic of the matter less mysterious is to call it out, front
> and center, and specify it very bluntly: certain classes of processors
> (who?) are required to recognize that the names in the namespace
>
> http://www.w3.org/2001/XMLSchema-datatypes
>
> denote the same datatypes as the corresponding names in the namespace
>
> http://www.w3.org/2001/XMLSchema
>
> We may be aided in this by the fact that in XSD 1.1 we have worked
> with at least partial success to make the term "datatype" have an
> extensional, not an intensional, sense, and we specify clearly that
> different simple type definitions can define the same datatype.
>
> It is not clear to me at the moment whether it matters whether names
> like
>
> {http://www.w3.org/2001/XMLSchema-datatypes}decimal
> {http://www.w3.org/2001/XMLSchema}decimal
>
> are defined as, or are in practice taken as, denoting (a) just the
> datatypes in question, not the simple type definitions, (b) both the
> datatypes and the simple type definition, which are after all closely
> related, or (c) huh? what are you talking about?
>
> I suspect that among editors of specs for the various language at
> issue here, opinion may be divided among (a) and (b), at least until
> someone goes to look up just what words are used in the spec, and that
> among implementors and users of the various technologies involved,
> answer (c) is likely to predominate.
>
> In any case, if we are going to undeprecate the datatypes namespace,
> we are going to need some story about what its names denote. I see
> three stories to choose from:
>
> (a) they denote the datatypes, but not the simple type definitions;
> that is, they have lexical spaces, value spaces, lexical
> mappings, and the like, but not necessarily unique names
> or {target namespace} properties.
>
> (b) they denote the simple type definitions in the XMLSchema
> namespace, and (by metonymy) the datatypes which are the
> extensional interpretations of those simple type definitions.
>
> That is, for example, the expanded name
> {http://www.w3.org/2001/XMLSchema-datatypes}decimal
> denotes the following simple type definition:
>
> {name} = decimal
> {target namespace} = 'http://www.w3.org/2001/XMLSchema'
> {base type definition} = anyAtomicType Definition
> {final} = The empty set
> {variety} = atomic
> {primitive type definition} = [this Simple Type Definition itself]
> {facets} = a whiteSpace facet with
> {value} = collapse
> {fixed} = true
> {fundamental facets} = {
> ordered = total
> bounded = false
> cardinality = countably infinite
> numeric = true
> }
> {context} = absent
> {item type definition} = absent
> {member type definitions} = absent
> {annotations} = The empty sequence
>
> Note that the namespace portion of the expanded name is not the
> same as the target namespace of the simple type definition.
> This will cause some readers (and possibly WG members)
> heartburn, to which the only possible response is: get over it.
>
> (c) They denote what they have always denoted, whatever that is,
> but we continue to deprecate their use on the grounds that no one
> is really comfortable saying just what it is.
>
> (d) They denote the simple type definitions, as described in (b), but
> not the datatypes, which are different kinds of things. Metonymy
> is to be frowned upon.
>
> John, since the use of the Datatypes namespace by Relax NG appears to
> be the major (or rather: only) argument thus far advanced in favor of
> undeprecating the namespace, it would be very helpful to know whether
> the differences among interpretations (a), (b), and (d) (or even (c))
> matter for RelaxNG's purposes, and which interpretation is most
> helpful to RelaxNG's use of the XSD datatypes. It would be helpful to
> have your view; it would be even more helpful to have the views of the
> editors of the RelaxNG spec, or the groups responsible for maintaining
> it,
I think, agreeing with most other posters, that (a) is the most useful choice
for RELAX NG purposes.
--
Configure bugmail: http://www.w3.org/Bugs/Public/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
Received on Wednesday, 4 February 2009 20:39:41 UTC