W3C home > Mailing lists > Public > www-tag@w3.org > August 2004

Re: XML and Versioning

From: <noah_mendelsohn@us.ibm.com>
Date: Sat, 28 Aug 2004 21:07:22 -0400
To: Rick Jelliffe <ricko@allette.com.au>
Cc: Dare Obasanjo <dareo@microsoft.com>, www-tag@w3.org
Message-ID: <OF843012DB.42A64B35-ON85256EFE.006EFB98@lotus.com>

Rick Jeliffe writes:

> At the moment, when you derive a new schema, you 
> cannot really create a new namespace for the 
> pragmatic reason that stylesheets and namespace-aware 
> software breaks.

Ignoring for the moment the question of whether leveraging a namespace 
hierarchy is an inappropriate break with the desire to treat URIs as 
opaque where practical, I agree that hierarchies would help with 
compatible consumption of evolving versions by namespace-aware tools. 
That's potentially important.

The problem I see is that it doesn't help at all with another show 
stopper:  I believe that any versioning architecture must scale for use 
over tens, perhaps hundreds of revisions.  In part because the instance 
syntax for namespace prefixes is unaware of any hierarchy, an instance 
document would presumably have to define and track prefixes for each level 
of the hiearchy actually used.  Perhaps worse, the user would have to 
remember which constructs were introduced by which version.  It seems to 
me that the proposal scales in practice only if we are willing to have 
prefix bindings in instances that are themselves aware of the hierarchy, 
Maybe you had this in mind, but your note referred only to awareness of 
the hierarchy in the schema language and other namespace-aware tools.  Am 
I missing something?  Thanks.

I do agree that an opportunity was missed to consider the admittedly very 
thorny problem of versioning when namespaces were being developed.  I 
understand that was a painful enough process in any case, but I agree with 
your intuition that tying versioning to namespaces would have been a 
fruitful angle to explorer.  FWIW, and as Matt Fuchs has often pointed 
out, what XML Schema calls local scoping might well have been easier to do 
clearly with, e.g., some facility for signaling that each element implied 
or was associated with its own default namespace.  We never worked out the 
details, but there too the intuition seems right to me.

Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
Received on Sunday, 29 August 2004 01:08:58 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:56:04 UTC