[Vrs-development] Re: [DotGNU]pnetlib and the VRS, SEE and other server concepts

[Chris Smith]

Okay. 'scuse my ignorance of this area of webservices, but I've got a bunch 
of questions!

Is the interface to a 'webservice' defined anywhere?
I suppose I'm asking if there is a defined entry point to webservices, ( 
main(a,b) in C springs to mind ) ??? Hmm, do I really mean entry point?  How 
about invocation method? (where method refers to both the entry point and 
mechanism by which the webservice is invoked).

I'm just thinking from a portability issue.  Joe Bloggs writes some ditty in 
C# on M$ .net and then decides to deploy within the dotGNU environment... If 
the invocation procedures differ, then there will be trouble.


The next question is that of request payloads.  Are all webservices in the 
dotGNU/.net world (I talk about both because I assume we're aiming to give 
people the *choice* to use something other than M$ kit!!!) supposed to use 
XML as the payload, or is is purely application dependent?

The transport mechanism isn't defined either is it?  M$ talk about SOAP 
(which incidental I'm quite keen on now - specifically SOAP with 
attachments!) but I don't think we should restrict the transport to anything 
in particular.  So long as dotGNU provides a network server 'plugin' for the 
transport and can decode and deliver the payload to the requisite webservice, 
then that'll be excellent.


If I can get these concepts/procedures sorted out I'll bung them in a FAQ 
type thing - it's essential 'To Know' info for people joining dotGNU from 
other disciplines, particularly if they are 'service' based too.... like me!

We also need to get these nailed down for the VRS project because it is 
highly modular and the abstraction is excellent - allowing us to put 
webservices 'here' and all the network handling 'there'......

Food for thought!

Chris