[XML-SIG] SOAP, Web services, hype, and Python

[Uche Ogbuji]

Interesting thread.  I have but a moment, so scattershot...

Michael McLay <mmclay@comcast.net>:

> Why is there so little discussion of "web services" on xml-sig? I've read Paul 
> Prescod's comments on web services verses REST. I generally agree with what 
> he has written. I've also seen Dave Winer be critical of UDDI/WSDL and the 
> "moras" that has been created. Unfortunately, my group seems to think this is 
> the next big thing because Microsoft and IBM are pushing this technology. Am 
> I being stubborn and they are correct, or is this stuff really broken and 
> difficult to work with?

Many of my recent entries on oreilly.net are apropos, I think:

http://www.oreillynet.com/weblogs/author/1054

I have come to the conclusion that until the SOAP and WSDL committees get 
their acts together and undo some of the damage that was done by the early 
iteratiosn of the technologies, that this stuff is indeed broken and difficult 
to work with.  And even if they band-aid some of the problems, there is still 
a fundamental nastiness in SOAP that makes it an obstacle to using the power 
of XML rather than an asset.

Web services folks, many of whom I respect greatly, continue to go through 
gymnastics to prove the pretend interoperability, as I was reminded at the 
recent XML Web Services One.  But on the other hand, some of the leading 
lights to the technology themselves have begin to realize that they may have 
taken an essentially good idea, and botched it.  I can tell you that my 
experience in the field gives the lie to that interoperability, even if I 
twist my head to the view of RPC that is supposedly the interoperable core.

I'll always say that the day WS folks decided to take on CORBA directly is the 
day they got themselves into the most trouble.


> I am working on a project that is attempting to use SOAP to send RosettaNet 
> messages between web portals. The group is using Java as the primary 
> implementation of the portal and I'm venturing off to use Python, as a test 
> of technology independance. I have found the exercise to be frustrating. I 
> have identified and worked with four partially completed implementations of 
> SOAP in Python. The test process has been useful, because I have identified 
> some minor flaws with the Java implementation, but I am somewhat concerned 
> that the Python community seems so uninterested in SOAP and the associated 
> web services tools, such as WSDL and UDDI. Are web services considered a dead 
> end in the Python community? If so, then what should I be investigating as an 
> alternative? Or, is the Python community just ignoring the industry's attempt 
> to standardized on XML based exchange and interface definitions? Or, was  it 
> simply the economic downturn that caused everyone to abandon SOAP in Python?

I can't speak for others, but I know that  have several times had to abandon 
the partial SOAP implementation in 4Suite, because implementing WS 
technologies brings me far too close to the ugliness of these technologies, 
and the itch has never exceeded the repulsion.  When I've been forced to use 
WS in consulting which has been, mercifully, not very often), I have tended to 
stick to the toolkits thya will be used on the other side: (JAX-RPC or .NET) 
because I don't believe in WSDL/SOAP interoperability, and my non-belief has 
been blstered often enough.

I had concluded that one day we could simply use ZSI or SOAPpy as a Web 
services implementation for 4Suite, but I haven't had enough success with 
either to take that step, yet.


> The WebServices implementation at Zope.org looks fairly complete, but I found 
> problems when trying to use it to read and use WSDL definitions from various 
> websites listed in xmethods.net.

I didn't even know there was one.  URL, please?  I'd like to mention it on 
xmlhack, and elsewhere.  Surprised it's never been mentioned on zope-xml.


"M.-A. Lemburg" <mal@lemburg.com>:


> I'll have to object to that. Web services are in wide spread
> use and the base is getting larger.

Well, same with Java, but I don't think that means we're all about to jump 
ship.

> I agree that getting everything to work isn't easy, which is one of the
> reasons why Tim Couper and I have developed the Siena Web Services
> Architecture (http://www.siena-tech.com/) -- it basically does away
> with all the complicated stuff and lets programmers think about the
> services themselves rather than all the logistics that go between the
> using application and the server.

Very interesting.  Another stealthy project I would have hoped would have come 
onto the scene a bit more loudly.  Seems the only download is for Windows.  
I'll be watching this one and will give it a try when I can more readily do so 
(time and technology-wise).

Rich Salz:

> Part of the reason is that folks who are using the various Python
> SOAP implementations generally find that they just work.  I don't
> know of any UDDI implementations, but the lower layers are fine.

Wow.  This really surpises me.  I've had trouble using ZSI and SOAPpy even on the XMethods toys (e.g. when writing an article), and they have positively colapsed when trying to work with a real-in-the-field Web service implemented with JAX-RPC, .NET or IBM WS Toolkit.  My typical experience with ZSI is an inscrutable fault from the other end, or unintelligible results.  My typical experience with SOAPpy as been an inscrutable traceback in SOAPpy itself.  I'd always assumed that SOAP/WSDL interop were understood works-in-progress with these packaes, so I've always really just shrugged about it.  Now that I know that interop i claimed, I guess 'll try to remember to file bug reports if I run into such problems again (which I hope won't be too soon, as I prefer not to work with WS at all).


-- 
Uche Ogbuji                                    Fourthought, Inc.
http://uche.ogbuji.net    http://4Suite.org    http://fourthought.com
Universal Business Language (UBL) - http://www-106.ibm.com/developerworks/xml/library/x-think16.html
EXSLT by example - http://www-106.ibm.com/developerworks/library/x-exslt.html
The worry about program wizards - http://www.adtmag.com/article.asp?id=7238
Use rdf:about and rdf:ID effectively in RDF/XML - http://www-106.ibm.com/developerworks/xml/library/x-tiprdfai.html
Keep context straight in XSLT - http://www-106.ibm.com/developerworks/xml/library/x-tipcurrent.html
Python Generators + DOM - http://www.xml.com/pub/a/2003/01/08/py-xml.html
Using SAX for Proper XML Output - http://www.xml.com/pub/a/2003/03/12/py-xml.html
SAX filters for flexible processing - http://www-106.ibm.com/developerworks/xml/library/x-tipsaxflex.html