> > Perl has a good SOAP module, but I've not recently tried 
> out the SRW 
> > WSDL in it. (It didn't work last time due to document/literal style)
> 
> I think the SOAP module that Rob's referring to is 
> SOAP::Lite, which is indeed well spoken of and works nicely 
> for many simple web services.  However, it struggles with 
> SRW.  I seem to recall that about a year ago, I did manage to 
> cudgel it into submission and persuade it to do an SRW 
> searchRetrieve, but it wasn't a pleasant process.  I don't 
> remember details, but I do remember a sense that I had to 
> force the module into a mould that was contrary to the 
> essence of what it's supposed to be.
> 
> More generally, I get the impression that the SRW WSDL is not 
> widely comprehensible to toolkits.  I don't know much -- 
> well, anything, really -- about WSDL so I can't comment 
> intelligently on this, but I get the impression that while 
> the SRW service definition is not intrinsically complex, the 
> current WSDL expresses it in a more complex way than 
> necessary, and it's this particular expression that trips up 
> the toolkits.  I'm sorry I can't be more specific, but if 
> anyone else has similar experiences, it would be good to 
> speak up.  I think it would be an enormous practical 
> incentive to the widespread adoption of SRW if the WSDL were 
> as undemanding as possible on the toolkits, even if that 
> simplicity has to be achieved at the expense of some elegance.

There were some errors in the WSDL - primarily because I was let down by
some of the validation tools - most of these are now fixed in the draft
WSDL.

The document/literal style thing is part of WS-I so if SOAP:lite can't
handle it then SOAP:list is not WS-I conformant.

Matthew