> > > > Can Python interface with C? If yes, I have a solution for you:)
> > > Yes, but integrating with ZSI would likely be cauchemardesque.
> > And, you'd rather change the stable spec ?
> > We just begun the six-to-nine month
> implementation-experience period.
>
> We just changed it to 'correct' the WSDL:
> recordPosition vs recordNumber
> diagnostic/code as Integer, not String [Should be
> positiveInteger?] startRecord as positiveInteger
> maximumRecords as nonNegativeInteger resultSetIdleTime to
> nonNegativeInteger from duration numberOfRecords to nonNegativeInteger
>
> The service definition is -completely- wrong now if the WSDL
> is the official version. If the service definition is the
> correct version then the WSDL has -never- been correct and
> still isn't. So where is this stable spec?
>
> If Matthew was unhappy with the service definition, he also
> had weeks to speak up about it.
Errrm,
recordPosition vs recordNumber was changed (on Rob's request) a few
weeks ago before the spec was "stable"!
The type issues as regards startRecord, maximumRecords etc. are
negligible a non-negative integer is still an integer (and the
supporting text rules out negative values anyway) i.e. compliance to the
WSDL implies compliance to the service specification.
As regards the SOAP array question the spec says "array of record"
(etc.) and the normative text "sequence of record" - there's nothing
there about it being a SOAP-ENC:array of record!
So the only way in which the WSDL is really incorrect are diagnosticCode
and resultSetIdleTime, which are genuine errors (and easy to correct).
Matthew
|