Lists Home |
Date Index |
"...newer versions of XML Schema should endeavour to remain backwards
compatible with XML Schema 1.0."
I take "endeavour to" to be droll understatement.
Paul Downey wrote:
> On 12 Jul 2005, at 06:05, Michael Champion wrote:
>> Sure, jAXB has their mapping, other Java vendors have their mapping,
>> Indigo has their mapping ... getting them to interop is the problem
>> AFAIK.I don't know how much of this is everyone wanting to standardize
>> on what they do, and and how much of it is real conceptual
>> differences between the platforms. There are a lot of smart people
>> working on this and I don't get a sense that the problems are just NIH
>> / "can't we all just get along by doing it MY way". Premature
>> standardization got us into this mess, so I think that there is a lot
>> of skepticism that ad hoc standardization will get us out.
> The Chairs' report, published last night, attempts to summarise the
> discussion at the workshop around this very topic, see 'Profiles':
> I personally think standardisation of 'object mapping', even within
> a set of today's best of breed technologies such as Java/C#/Python is
> a little dangerous given XML is about exchanging documents, or at least
> interoperating with those who want to work with XML directly. What
> goes on behind the XML curtain is very much a per-implementation concern.
> Having said that, I believe there is real value in knowing which aspects
> of schema are most likely to give 'a good user experience' when using
> today's data binding tools. I tried to explain in BT's experience report
> how such an 'implicit profile' already exists - in particular what works
> with .NET code generation - that's who most people seem to test against.
> Unfortunately it's left as an exercise to each publisher to ascertain
> what actually works well through a process of trial and error.
> I've also heard many people asking how to express common data structures
> such as collections, arrays, indexed tables, etc to 'round-trip' to and
> from XML on the same platform or so they /might/ surface in similar
> another programming model. I think that's a related, though subtlety
> requirement to a 'profile' in that it is much more wide-ranging than
> and is currently being discussed as a possible topic of a WSDL WG note.