[XML-SIG] WSDL library ?

Paul Prescod paul@prescod.net
Thu, 14 Feb 2002 13:33:12 -0800


Rich Salz wrote:
> 
> The discussion seems to be heating up a bit, so I'm gonna start to slow
> down in terms of posting. :)  

Understood. You don't have to reply if you don't have time or feel like
it.

>...
> The ability to transmit pointer-using data types (e.g, a balanced tree
> in C/C++), to make changes on the server, and to send the new tree back
> such that the client can reconstruct -- that can be important and
> useful. Sure, Corba proves that you can solve real-world problems
> without it, but that doesn't reduce its utility.  The SOAP id/href
> technique makes that possible.  Server or client-side state has nothing
> to do with it.

Okay, but this raises a whole host of questions for me:

 1. Where do you use this pointer feature in a business application?

 2. How did you decide that this should go in the protocol instead of at
the application layer? For instance, I would love it if SOAP supported
structs where the keys were themselves structured (e.g. a
string/float/Qname triple). Python has no problem with this. Java has no
problem with this. But I'm sure if I asked for it, SOAP's designers
would say that it should go at the application level.

 3. Internal pointers are nice. But to me, they are much, much, much
less useful and important than pointers from the message back into a
data store potentially on the same machine but potentially on another
machine. Unfortunately there is no standard way to refer to the results
of a SOAP method call so you can only make a standard pointer to
something exposed with HTTP, FTP or another URI-integrated protocol.
What do you see as the solution for this problem?

>...
> I don't believe it is totally fair to complain about the developing
> world of SOAP interop.  

If you're just trying to solve problems then it really doesn't matter
whether the complaint is fair or not!

The spec is fairly new, most suppliers are on
> their first generation of implementation, etc.  It is particularly not
> fair to compare it to Corba interop, which had was first defined nearly
> five years ago, after nearly five years of deliberately avoiding the
> issue. I am quite pleased by the state of SOAP interop; amazing progress
> has been made in a few months, so much so that at the end of the month
> there will be a two-day interop festival for early WSDL.

Okay

> In an earlier message I refered to XSD files at
> http://www.zolera.com/schemas/2001/11/ ; unforunately our ISP (good ole
> Win-chapter 11-Star) had problems, which seem to be fixed.  They are
> real, not contrived examples, used in our Tamarin product; a server that
> does XML signatures and encryption.

I don't doubt that you are solving real problems but it isn't clear what
advantage SOAP is conferring beyond traditional XML/HTTP ways to solve
the same problems. It seems like the schemas are doing the hard work of
data structure description and HTTP is doing the hard work of getting
the data from point A to B. So what benefit are you getting out of SOAP?
For example, I can't find occurrences of "href". You use arrays, but I
don't think they are SOAP arrays...

 Paul Prescod