W3C home > Mailing lists > Public > www-tag@w3.org > June 2002

Re: PSVI architectural discussion

From: Mike Champion <mc@xegesis.org>
Date: Sat, 22 Jun 2002 22:29:12 -0400 (EDT)
To: www-tag@w3.org
Message-id: <XSIGLFURV1W98EB41IHVQ7FBPK970.3d153206@MChamp>




Tim Bray <tbray@textuality.com> said:

>8. Why not standardize on XML Schema's primitive data types?

>  W3C has invested a huge amount of effort in building a
> primitive-type system as a part of XML Schema.  I personally think it's
> too big and some gHorribleKludge types got in, but they're done and
> stable and I don't see any reason why they shouldn't serve as a basis
> for XQuery and anyone else who needs this kind of thing.

> Proposal: let's issue a TAG finding saying that if you need primitive
> data types, use XML Schema's, don't invent your own.

I would suggest thinking long and hard, and soliciting lots of input
from implementers and end users, before issuing such a finding.
TAG findings should, IMHO, reflect best architectural practice rather
than "it's done, it's stable."  
Why not investigate, strip out the gHorribleKludge types, and issue
a finding on what a reasonable core set of types is?  Or, better still,
consider a finding that there is no universally appropriate set of types
and direct the Schema WG to develop a pluggable type library architecture
in the next revision.  That would maximize consistency and the parallel
development between the W3C schema/query effort and the ISO DSDL effort.

Of course, the W3C base types may be suitable for Xquery, and the
XQuery folks seem to be comfortable with this idea, but I would prefer
that the TAG give them leeway to do whatever proves to be practical and
implementable in an interoperable manner ... they may choose to 
rethink this in light of actual experience with various gHorribleKludges.
Perhaps it would be best to accept that there isn't enough
real experience and understanding of these complex and controversial
issues upon which to base a TAG finding at the present time.

>The XQuery processor that's accessing this database will know from some
> sort of data dictionary implementation that a <detail> element has
> unitPrice= and quantity= attributes, and the primitive data types of
> each attribute.  While it uses primitive type names from XML Schema, it
> is possible in principle and plausible in practice that no schema has
> ever been written, let alone applied.

This may be overly pedantic, but I'd submit that for all practical purposes
a "data dictionary" that defines type information is essentially a 
"schema" for the purposes of this discussion.  As I understand it, it
was a requirement for the W3C schema language to map onto the various
database schema languages, and so in some sense DB schema languages
define the "schema" information items (or whatever they are to be
called) in another syntax.

I don't think this quibble has a substantive impact on the discussion.
But at least be very careful not to equate "schema" with "W3C XML
Schema definition language." 
Received on Sunday, 23 June 2002 21:36:51 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:32:32 UTC