RE: which layer for URI processing?

At 02:34 PM 5/24/00 -0400, David E. Cleary wrote:
>
>>I'd say it was incredibly poor design on the part of the chemical plant
>>handler, not on the part of the XML parser.  (It sounds like XPath/XSLT
>>absolutizes anyway, so I don't think it would get that far anyway.)
>
>>So no, it's not a problem.
>
>You do not think it is a problem that the qualified names of two nodes are
>considered equal at the XML + Namespace layer, but unequal at the XPath/XSLT
>layer? Whatever is decided, they need to be consistent in my book. Qualified
>names should not arbitrarily change between layers.

It's not an arbitrary change.

It's a change done at some layer in processing as decided by the program or
standards architects.

If you find it arbitrary, there's a very simple solution: don't use
relative URIs in namespaces, and tell your friends the same.  That's the
approach Common XML takes, for instance.  However, as there is decided
resistance to such a prohibition, taking the letter (and some say the
spirit) of the namespaces Rec seriously seems like a very good answer.

If you want consistency, don't try to drive semantic understandings into
syntactical layers.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
Building XML Applications
Inside XML DTDs: Scientific and Technical
Cookies / Sharing Bandwidth
http://www.simonstl.com

Received on Wednesday, 24 May 2000 19:32:37 UTC