If we are going to do it this way - I think I would like to drop the
version number from the namespaces. This is going to be messy but we are
going to do it sometime (e.g. when we hit 2.0).
Most webservices to date, do the Rob approach of creating a new
namespace for the new server. A server which supports multiple versions
will do so at different URL endpoints (e.g. http://myserver.org/srw11
for SRW 1.1, http://myserver.org/srw10 for SRW 1.0.
At the moment there aren't any "good" or "well-adopted" strategies for
backwards compatible versions - although I know that some other
webservices are looking at the mechanism I was suggesting.
> -----Original Message-----
> From: Z39.50 Next-Generation Initiative [mailto:[log in to unmask]]
> On Behalf Of Robert Sanderson
> Sent: Wednesday, July 30, 2003 2:46 PM
> To: [log in to unmask]
> Subject: Re: 1.1 Versioning
> On Wed, 30 Jul 2003, Matthew J. Dovey wrote:
> > > How did OAI handle the 1.0, 1.1, 2.0 transitions?
> > "Is version 2.0 of OAI-PMH compatible with earlier versions?
> > OAI-PMH version 2.0 is not backward compatible. Results from the
> > Maybe. But, as the service gets more mature, the changes should be
> > smaller and maybe the choice will become easier. My hope
> is for rapid
> > stability.
> I don't think we'd be doing anyone a great disservice by simply
> abandonning 1.0 so long as 1.1 clients can still talk to 1.0
> servers with
> simple requests. (By simple, I mean that they don't include
> any of the
> 1.1 changes)
> So I think we don't increment the namespace at all and all of the new
> fields go in to the existing namespace. We can increment it at major
> versions if we feel the need.
> That said, I think a version parameter might still be useful,
> for SRU, but I'm in two minds about what it should default to. If we
> abandon 1.0, then we can say that version defaults to the
> latest release,
> then for the majority of clients and servers it's unneeded. If it
> defaults to 1.0 then it's required on every transaction just
> to benefit
> the people who aren't upgrading, which might be no one.
> A 1.0 client can be pretty quickly retrofitted with the
> version parameter
> to say 'I'm only 1.0 please be gentle' even though it wasn't
> a valid param
> in 1.0 ... SRU 1.0 servers will simply put it in the unknown
> fields and
> return a 1.0, but 1.0 SRW servers will fail the request with
> a SOAP fault.
> The client can always catch that and take off the version parameter I
> suppose. *ponder*
> ,'/:. Dr Robert Sanderson ([log in to unmask])
> ,'-/::::. http://www.o-r-g.org/~azaroth/
> ,'--/::(@)::. Special Collections and Archives, extension 3142
> ,'---/::::::::::. Nebmedes: telnet: nebmedes.o-r-g.org 7777
> ____/:::::::::::::. WWW:
> I L L U M I N A T I