Re: XML Schema versioning in 4310bis

All,

The way I view this is considering what is done from an Object Oriented
perspective.  If I have a defined interface that is being used, to add
features but with maintaining backward compatibility I would create an
extended interface without touching the existing interface.  I¹m assuming
that some sort of factory is being used here to return the appropriate
object, where clients of the old interface would continue using the old
interface and clients of the new interface would type cast to use the new
interface.  As far as 4310bis goes, changing the URI is like changing the
package of the existing interface which will not allow the client to use the
existing interface.  I do see the issue of a client that wants to use the
new interface not being aware that it¹s supported.  In programming
languages, the client would use reflection to determine if the new interface
is supported.  

I have always been an advocate for backward compatibility, where a change to
the URI will break compatibility.  How about looking for a compromise on
this?  What if we reused urn:ietf:params:xml:ns:secDNS-1.0 for the protocol
messages but add the requirement that the server publish the new URI
urn:ietf:params:xml:ns:secDNS-1.1 in addition to
urn:ietf:params:xml:ns:secDNS-1.0 in the greeting if 4310bis is supported?
We could register the new URI namespace (not the schema) and use it strictly
for publishing the support for the later draft.  Clients are not forced to
deal with different versions across the servers, servers can support a
single version (either old or backward compatible new), and servers
explicitly publish support for the new specification in the greeting.

-- 


JG 

-------------------------------------------------------
James F. Gould
Principal Software Engineer
VeriSign Naming Services
[email protected]
Direct: 703.948.3271
Mobile: 703.628.7063

 
21345 Ridgetop Circle
LS2-2-1
Dulles, VA 20166

Notice to Recipient:  This e-mail contains confidential, proprietary and/or
Registry  Sensitive information intended solely for the recipient and, thus
may not be  retransmitted, reproduced or disclosed without the prior written
consent of  VeriSign Naming and Directory Services.  If you have received
this e-mail message in error, please notify the sender immediately by
telephone or reply e-mail and destroy the original message without making a
copy.  Thank you.



From: Andrew Sullivan <[email protected]>
Date: Tue, 26 Jan 2010 09:15:08 -0500
To: EPP Provreg <[email protected]>
Subject: Re: [ietf-provreg] XML Schema versioning in 4310bis

On Tue, Jan 26, 2010 at 09:01:47AM +0100, Ulrich Wisser wrote:
> Hi,
>
> in this case I have to agree with James. We discussed changing the URI
> and came to the conclusion that changing the URI will break old clients
> without any need to do so.

That argument is circular.  What is up for debate is whether there is
in fact a need to do this.

> But let's have a look on the alternatives:
>
> 1. Server supports only the "old" 1.0
>    - no changes on client or server
> 2. Server supports only the "new" 1.0
>    - only changes on the server side needed
> 3. Server supports both
>    - only changes on the server side needed

But in (2) and (3), the client _can't tell_ what support is in the
server.  The whole point of the handshake in the protocol is to make
it possible for people to tell.

> Which registry is going to support 1.0 after 1.1 has been released for
> any time longer then needed?

Who knows?  This is the Internet.  We have parts of the DNS
infrastructure (including several MSFT APIs) that don't support the
DNS Unknown type, even though that was specified over 10 years ago and
is terribly useful.

> Another question comes to mind: Why should 1.1 be backwards compatible?
> If we break the URI there is no need to keep the broken <rem/>, is it?

Right.

A

--
Andrew Sullivan
[email protected]
Shinkuro, Inc.
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
List run by majordomo software.  For (Un-)subscription and similar details
send "help" to [email protected]