Arrays etc. in WSDL (was: proposing closing issues 26, 3, 14, 23, 32, 65, 69)

Paul,

while there may be little interoperability on non-XML-Schema structures
like arrays, associative arrays and matrices, I believe the support for
actual XML Schema constructs is growing so it becomes only a one-party
issue on how they map high-level constructs into XML Schema.

We attempted to tackle the whole issue in XMLP WG, where the
requirements were structures, arrays and references, and we ended up
with the SOAP Data Model and SOAP Encoding. I don't think the WS-Desc WG
should go there because we would be hard pressed to agree even on the
requirements.

Anyway, I believe such material, if it should be produced, should be
worked on by XML Schema or related groups, not by the WS-Description WG.

Best regards,

                   Jacek Kopecky

                   Systinet Corporation
                   http://www.systinet.com/




On Thu, 2003-12-18 at 13:00, paul.downey@bt.com wrote:
> Jacek wrote:
> 
> > Issue 3 basically asks us how arrays (a SOAP Data Model term) are
> > declared in XML Schema. We don't deal with SOAP Encoding (nor the 
> > Data Model) at the moment so I suggest we close this issue. I think 
> > until we tackle SOAP Data Model fully (if ever), we shouldn't try to 
> > do bits of it.
> 
> WSDL 1.1 document included a set of rules for naming and representing
> ArrayOfBlah in an /encoded/ binding which greatly aided interoperability 
> of for rpc/encoded exchanges.
> 
> I'd be happy to close this issue for WSDL 2.0, but poor schema support 
> in many current implementations means we have a lot of interoperability
> issues surrounding how data structures such as arrays may be exchanged
> across a /literal/ binding. 
> 
> Also many systems need to exchange keyed data: associative arrays in 
> Perl, PHP, etc, hash tables in Java, C#, Ruby and indexed table from a
> database. 
> 
> Whilst I acknowledge the trend is a move away from data structures being 
> directly /encoded/ in WSDL towards XML 'documents', I'm concerned that 
> poor support for the whole gamut of schema in current code 
> implementations means that users in a code model are left to find a lowest
> common denominator of schema for themselves just to encode the most simple 
> data collections.
> 
> I therefore propose we provide suggested schema extracts for representing 
> a vector, a matrix and an associative array. These would not be normative, 
> but would provide a well supported pattern to follow when generating code 
> from WSDL and WSDL from code. 
> 
> Paul

Received on Thursday, 18 December 2003 07:48:24 UTC