RE: [ciq] xLink namespace declarations in CIQ v3.0/xBRL

-------- Original Message --------
Subject: [ciq] xLink namespace declarations in CIQ v3.0/xBRL
From: colin.wallis@ssc.govt.nz
Date: Sun, June 11, 2006 8:58 pm
To: ciq@lists.oasis-open.org

Thanks Ram

OK. I have not substantiated the claim, but on the assumption Hugh is correct then I think we do indeed have a problem.

I think we do need to keep interop at the top of our minds...and particularly with xBRL who are strong candidates to use our schemas.

I don't know what is involved in making this change, but before doing so, I think we should approach FIRST W3C Xlink 1.1 WG "hard and fast" to get them to give us a definitive view of how to fix it. They are the root cause of the problem becuase they have not provided normative schema so I think they should take some part of the responsibility in the solution. If they instruct us that going back to TR/xlink is correct then it binds them into an approach that they will think twice about changing later.

Cheers
Colin (Wallis)..no relation to Hugh Wallis below that I know of!:-)

...<snip>....

While investigating the consequences of requiring external schemas to
be included in the DTS for the Dimensions spec, Ignacio produced an
example which exposed the following issue which I think is potentially
rather serious (although the solution is not difficult from our point
of view). Before I go on, however, please note that this topic is NOT
a discussion about the Dimensions spec itself but about a deeper
issue.

In our implementation of the schemas for XLink we have created a
schema "xlink-2003-12-31.xsd" which purports to define the namespace
"http://www.w3.org/1999/xlink". This consists of a set of global Type
declarations and some attribute declarations that use them. In
addition we have created another schema "xl-2003-12-31.xsd " which
defines the namespace "http://www.xbrl.org/2003/XLink". This second
schema references the global type declarations that we created in the
namespace "http://www.w3.org/1999/xlink".

Now when we take a look at the schemas (attached) produced by the
OASIS CIQ TC, which also make use of the XLink spec, we can see that
they have done a similar thing, except that they have done it
differently. Of particular note is the fact that they have created a
schema "xLink.xsd" which purports to define the namespace
"http://www.w3.org/1999/xlink" and which consists of a set of
attribute group declarations which are then referenced by other
schemas in their system.

Now when you come to use all of these schemas in one cozy environment
you hit upon a namespace definition conflict since the definitions of
"http://www.w3.org/1999/xlink" are not the same. If you try to, for
example, import the CIQ schemas into an XBRL taxonomy schema (e.g. try
XBRL validating instance-B.xml), XBRL processors will fail to schema
validate the whole set since they have already decided that the XBRL
definition of the "http://www.w3.org/1999/xlink" namespace is what
they are going to use and they (justifiably) fail to follow the
schemaLocation hint provided by CIQ because of this.

I think that both the OASIS TC and XBRL have made the same mistake
here - and that is to define something in the
"http://www.w3.org/1999/xlink" namespace that does not belong there
since the XLink spec does not define either the types or the attribute
groups I mentioned. As a result we have prevented interoperability.

I believe that both XBRL and OASIS must change their schemas that
define the namespace "http://www.w3.org/1999/xlink" to bring them into
line with just the minimum that is defined by the XLink spec and then,
even if we take slightly different approaches (e.g. use of NMTOKEN vs.
token), we should achieve consistency and hence interoperability.

It is instructive that this issue has taken until now to surface.
Hitherto we have been living in our own world where no other XLink
users have managed to come close enough to surface this issue and so
there have been no consequences of our misbehaviour in respect of the
definition of "http://www.w3.org/1999/xlink". Similarly for the CIQ
folks.

Once we are agreed on this (changing our XLink schema to make it
"legal") I think that we should liaise with the OASIS TC group on the
issue (suggesting they change there XLink schema to make it "Legal")
as well as with the W3C XLink 1.1 WG (pointing out the consequences of
not providing a normative schema for their namespace or at least clear
guidance as to how to go about creating interoperable schemas for it)

Thanks

Hugh

Hugh Wallis
XBRL International Inc. - Standards Development
hughwallis@xbrl.org
+1 416-238-2553
Skype: hughwallis
MSN: hughwallis@hotmail.com
Yahoo IM: hughwallis

ciq message