W3C home > Mailing lists > Public > www-ws-arch@w3.org > September 2003

RE: Draft TAG finding on extensibility/versioning

From: Olivier Fehr <Olivier.Fehr@ofehr.com>
Date: Sat, 27 Sep 2003 23:22:07 +0200
Message-ID: <F92F63FC00C0AD4D88BECB3BCF90734B19E3@coyote.ofehr.com>
To: "David Orchard" <dorchard@bea.com>, <www-ws-arch@w3.org>

OK, I admit I am new to this list, so I may miss the point completely
and make a fool of myself, but please bear with me until I am up-to-date
with all relevant documents and developments. Here I go with some
comments:

Section 1.4 point 3. seems to use a different concept of version.
Whereas you are generally assuming - rightly - that languages evolve
thus giving rise to different versions, this is a case of different
versions (maybe flavors is a better word?) existing from the beginning
to ease transition.

Section 1.5
From my very basic point of view, I would expect elements and attributes
not to be removed, I think they should be marked as deprecated (or
something similar) and their further use discouraged. New elements and
attributes can be added at any time. A web service should just not take
those into account if they are not specified in its WSD.
Semantics is probably the toughest one to handle.

It seems to me that language changes basically occur when a namespace
that this language uses changes. Maybe we need to treat a namespace as
an interface, meaning once published it can never change, and you must
provide a new interface (namespace).
So if a web service was built for accepting v1.0 of a language, it 'must
ignore all' of V1.x additions to be backward compatible, no?
I guess that's your point in section 5, point 2.

Too bad a newer namespace cannot inherit and older one...

Best Regards
Olivier


-----Original Message-----
From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org] On
Behalf Of David Orchard
Sent: samedi, 27. septembre 2003 03:02
To: www-ws-arch@w3.org


Hi all,

Thought I'd forward this to show what my thoughts are on extensibility
and
versioning.  The TAG has not approved this at all, so it may change
dramatically, though there has been 1 round of light tag review.  I
encourage you to read and comment (to www-tag or privately..)

http://www.w3.org/2001/tag/doc/versioning.html

I think it should be obvious that I think that any discussion on web
services architecture should be about web services specific aspects of
extensibility and versioning.  Such as:
- Web services is a data format, therefore web services should use the
must
ignore all rule
- Web services is an open system, therefore web services should use
wildcards for extensibility given the no-touch possibilities.
- Web services specifications can be evaluated against whether they
support
extensibility and/or compatible changes.
- Web services use XML Schema, therefore the section on XML schema
applies.

I still really like uml models btw.

Cheers,
Dave
Received on Saturday, 27 September 2003 17:22:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:22 GMT