[dsdl-discuss] Re: Namespace renaming

Ken

> I'm sorry I don't get more time to participate at this depth. And
> I'm sorry I'm making you type so much when your arm is not
> well! BTW, you will probably lose my participation next week because
> my hospital visits begin on Monday morning.

Fortunately my arm is on the mend, which is why the spate of emails in the
last few days.

Sorry to hear you have to start treatment before Christmas. Hope it doesn't
completely stop your enjoyment of the season.

> You don't say if your xia prefix is mapped to your own
> namespace. Let's consider the two answers to this question:
>
> (1) - source: <xia:div xmlns:xia="urn:x-martin:stuff">
> target: <xhtml:div xmlns:xhtml="http://www.w3.org/TR/REC-html40">
> - in this case you aren't just renaming the prefix you are
> renaming the entire element
>
> (2) - source: <xia:div xmlns:xia="http://www.w3.org/TR/REC-html40">
> target: <xhtml:div xmlns:xhtml="http://www.w3.org/TR/REC-html40">
> - in this case the online system shouldn't care what the prefix
> is and should just as well support <xia:div> as an HTML construct

The secanrio I have descibed is (1) - beacuse you cannot validate a subset
against the full schema so should give a it a different URI. [But see below
re a scenario that requires (2).]

> If, however, the online system isn't namespace URI aware and is keyed
> on the prefix, then that's another story. If the system isn't XML
> compliant and we need DSRL to accommodate non-XML systems, then my
> arguments don't apply.

Remember we also have to allow for DTD validation, and for SGML files which
have adopted names with a colon which have the effect of namespaces when the
are seen by XML aware recipients. Renaming must work in these scenarios as
well.

> >My real problem is that we have a scenario that has:
> >
> >Source Schema --> DSRL Map --> Target Schema
>
> Right ... and if the two schemas are in *different* namespaces, then
> you are renaming the constructs in their entirety (prefix and URI)
> and my issue does not apply. And if the two schemas are in the
> *same* namespace, then DSRL is irrelevant and the target system
> should accept the source instance without change.
>
> I fully support DSRL features that totally rename source constructs
> to target constructs when the target construct is in another
> namespace. Of course the prefix has to change. That isn't the
> situation that I was trying to flag. I understood you earlier to say
> that you wanted a DSRL instruction that changed the prefix but *not*
> the namespace URI, and my only warning was that if you do so then
> namespace-qualified names in attribute value contexts would lose
> their prefix declarations because DSRL doesn't know (and can't know)
> what attributes to change.

The renaming xs to xsd to harmonize things like the use of datatype
definitions is the use case I had in mind for renaming where the same URI
occurs as stated in (2). In the general case I can see a need for allowing
this as well.

> >I know the namespace of the two end points in this list, and that the
DSRL
> >Map is in the DSRL namespace, but what information does the DSRL map need
to
> >hold re the prefixes and associated URIs of the end-points, and should
this
> >information be held in a form that is "DOM viewable" or in the form of
> >attributes whose values could be slotted into XSLT statements at
appropriate
> >points?
>
> Sorry, not sure what you are asking here.

My concern is whether or not namespaces used in maps need to be declared in
maps. If they are not declared, how do we know which URI was meant to be
associated with a prefix in the transformed document?

Martin

--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)