Tim posted his response to my response to his... on the pros and cons of WS-I profiling Schema.
He asserts a few good points:
1) Schema is more than just "implemenetations of OO serializers", despite the fact that all of us vendors market our tools that way,
2) Not using 1 tool from a vendor doesn't mean throwing away all the vendors tools, just maybe 1 of them,
3) There are dangers in doing the profile based upon today's tools. There are gadjillions of different things that might go wrong,
4) And that just because I might think that xhtml isn't useful for Web services, he has interestingly asserted his rights as a Customer/User to exchange xhtml in his application - MSDN.
These are some interesting counter-arguments which take a bit of time to go into. My cohort Mark Nottingham has done a good job of rebutting most of these in his response. But I tend to think the counter arguments are missing my real point.
From the customers perspective, they can't get interop of schema implementations. The vendors have successfully prevented the W3C from being a true standards body as there is no certification and branding revocation capabilities. There is no other "body" that can publish the conformance of the schema toolkits. So they don't know what works or doesn't in any given implementation.
In terms of counter-argument #3, all the scenarios that Tim lists might be encounted by the customer. So they have to do all the testing work on every toolkit to find out the conformance. This hurts them. A lot. For every different toolkit they *might* buy, they have to evaluate the toolkit against the schemas they use or might want to use. This will tend to mean that they will be fewer different toolkits, because there simply isn't interoperability/portability across the toolkits. And we all know who wins when the market starts picking based upon complexity - the biggest vendors.
As a customer, and they know that lack of interoperability costs money that they'd rather actually use to deploy some of those darned Web services, and they also know that WS-I won't help solve this problem, what would I think about WS-I?
And that's the nut. If schema profiling is going to fail, I think it should fail because the users in the room can't agree upon the subset, and not because the vendors said "You'll never be able to agree upon the subset so we'll try to stop forming a profiling group". I really worry about the customers becoming jaded and leaving WS-I if they are prevented, discouraged, or even not helped, in working on this problem.
Again, I do have serious reservations that customers will be able to agree on the subset of Schema. And I have heaps of complaints about Schema, like I said in my xml.com versioning article on what they should do about extensibility and versioning and why these are hard in Schema 1.0. At least Schema 1.1 looks like they will do some of the things I suggested, though not the complete set they should do. Another "it will get better some time in the future argument". sigh. Tim and I agree on a number of other things: XML can and often should be exposed to the developer and not hidden behind a serializer, that RPC is bad bad bad, that Action is really really important, and lots of other stuff.
But I really feel the customer pain. The only suggestion from the vendors so far on how to solve their interop problem for Schema is basically "wait, the tools will get better". Sigh. I remember the standards wars days of browsers too. How is that full CSS standards support in IE and Netscape going anyways?
Maybe if there were a concrete counter-proposal that was different than a profile, then I'd support that. How about "The vendors will agree to provide results of running all the tests that the WS-I Schema Exhaustive and Exhausting test suite Working Group come up with"? Then all the customers don't have to run the tests themselves. But I just can't see that one happening.
In the absence of anything better than waiting for that wonderful day when their Schema interop problems are over, I say they should be given every opportunity to try a profile to solve their problem. Maybe they won't succeed. But they will know that WS-I is a forum for users to strongly contribute to solving their problems, rather than a ... different ... kind of forum. Kind of the same way that I believe that part of democracy is accepting when the electorate makes decisions I don't like.