Hi Burak!
Thank you for your response! I only have a few things to add...
The real reason I got rid of wsdlparse was because it did things the
java way -- it generated code. That's a fundamentally wrong approach
when you're working with Python because, by the virtue of being a
dynamic language, Python can parse wsdl and build required classes and
interfaces on the fly.
It makes sense (except for one thing explained below) to make this all
automatic. Actually I think it would actually even make more sense to
remove the @soapmethod modifier altogether to make things look even more
like regular method calls. This would require other programmatic
utilities to declare datatypes but in my opinion it makes sense too
since this part pertains more to the data representation than to the
implementation.
Regarding the fact that you removed the script, please note that except
of generating the code automatically, this has just the effect of
forcing the developer to do it manually. So as long as the wsdl "class
builder" is not in place, it's a bit useless since the effect is the same...
I also wanted to focus on making soaplib a proper soap server, because
that's what was lacking then in the Python world. (There even was a
time i'd entirely dumped client functionality in favor of suds.) Oh
and wsdlparse did not work most of the time anyway.
If suds wasn't already there, we'd probably have wsdl parsing in
rpclib by now. (I already have a javascript soap client which also
parses wsdl, here: http://github.com/arskom/qxsoap) But I did not
think duplicating suds' functionality in rpclib would benefit anybody,
so I didn't do it.
What do you mean by the "suds functionality" ? The fact that it's a
client interface library?
I'd nevertheless encourage you to work on a wsdl parser for rpclib,
because i feel it'd have a greater benefit outside of the SOAP world
-- I've stumbled upon a few people in the lxml mailing list who might
be interested working on an Xml Schema parser. They might want to give
you a hand if you ask around.
I would very like to, but I won't be allowed the time to do so.
There is another similar subproject (for the OpenChange project) that I
would ideally like to implement where the structure definitions are
encoded as an IDL and I don't think I will ever have time for this either.
Thank you!
Wolfgang
_______________________________________________
Soap mailing list
[email protected]
http://mail.python.org/mailman/listinfo/soap
