W3C home > Mailing lists > Public > xml-dist-app@w3.org > March 2003

Re: WSDL 1.2 drops use="encoded"

From: Robert van Engelen <engelen@cs.fsu.edu>
Date: Fri, 7 Mar 2003 15:48:18 -0500
Cc: Jacek Kopecky <jacek@systinet.com>, XMLP Dist App <xml-dist-app@w3.org>
To: Rich Salz <rsalz@datapower.com>
Message-Id: <201C2074-50DE-11D7-AB34-000393C3075E@cs.fsu.edu>


Jacek, Rich, & others,

I am grateful to see a discussion on this topic emerge. From what I've 
read I believe the move to eliminate "use=encoded" is not a bad 
decision. The problems seem to outweigh the benefits and SOAP RPC 
encoding can still be supported with encodingStyle. Having said that 
though, I do agree with Rich that this move will likely diminish the 
role of SOAP RPC further and I share his concerns on this matter. The 
burden will become to rest on the shoulders of the developers and users 
of SOAP/WSDL tools. As a result of this, I see some problems that lie 
ahead to establish cross-platform SOAP interoperability. First of all, 
I recall that SOAP RPC has significant merit with respect to object 
graph serialization and the mapping to an application's data view. I 
have read that WS is working on an object graph serialization 
algorithm, but the spec has not appeared yet. Secondly, I believe SOAP 
RPC has demonstrated its success in the field, despite that literal 
encoding appears to be more popular (with popularity based on many 
non-technicalities). There are certainly many advantages to literal 
encoding. But no one so far has been able to convince me that the 
removal of "use=encoded" and the move to literal encoding is the right 
move. Hoping for widespread support for a new schema language to 
support something that resembles SOAP RPC is a gamble that I won't be 
happy to wager, not at this early stage with so little evidence. Or am 
I too pessimistic?

- Robert van Engelen, Genivia inc., gSOAP development.

>> If you and other people come to one of the groups and say you need 
>> SOAP
>> 1.2 Encoding supported directly in WSDL 1.2 because you actually need
>> SOAP Encoding, the groups would probably work on the schema language. 
>> I
>> don't think this is an issue of resources because I'd happily 
>> volunteer
>> to do an initial proposal, as I believe I've said before.
>
> Defining a new schema language, to say nothing of hoping for 
> widespread support for it within existing WSDL tools, is a fool's 
> errand.  No thanks.
>
>> The alternative to inventing a SOAP Data Model Schema Language was to
>> specify precisely how XML Schema works with use="encoded", which I
>> believe would have been much more work to get right and sort out all 
>> the
>> corner cases.
>
> I disagree.  An alternative would be to leave "use=encoded" and 
> require that the encdoingStyle URI be defined by the XMLP WG.  This 
> makes sense:  if you are using an encoding style to "modify" the 
> schema, then the obligation is on the modifier, not the WSDL group, to 
> define that modification.
>
> (Note that there is no requirement that the "modification" be 
> described in XML, although the idea of using an XSLT script to 
> transform an XML Schema so that the SOAP RPC attributes are natively 
> supported is kind of interesting...)
>
>> So, voice your needs!
>
> This has two answers.  The first is that I am fortunate to have a 
> public forum in which to express some views,
> and have done so a couple of times:
>     1.  http://www.xml.com/pub/a/ws/2002/11/20/ends.html
>     2.  http://www.xml.com/pub/a/ws/2003/03/04/endpoints.html
>
> Perhaps I flatter myself too much to think that WSDL and XMLP folks 
> read them.
>
> The second can be found just below.
>
> > Many (including myself) seem to share the
>> understanding that people are shying away from SOAP Encoding. This may
>> be because XML is just sufficient, or it may be because SOAP Encoding
>> doesn't play nicely with XML Schema and there are no alternatives as
>> yet.
>
> Developers are being forced away from SOAP RPC because vendors are 
> killing it off.  (I just looked at the XMLP and WSDL WG membereships 
> to confirm, and they are clearly dominated by ISV's who develop 
> SOAP/WSDL products, rather than developers who use such products.)
>
> I won't comment on motivations, but I suspect that looking through the 
> mailing lists at the time that XMLP decided to split SOAP encoding off 
> into a separate document might be informative.  At any rate, it was 
> clear to me that once that was done, the writing was on the wall.  The 
> only surprise is how quickly it's happened.
>
>> Hope we don't end up with any hard feelings,
>
> On a personal level?  Certainly not.
> 	/r$
>
Received on Friday, 7 March 2003 15:44:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:13 GMT