RPC 80/20 split (was: Re: RPC Style Issues (3))

"Tom Jordahl" <tomj@macromedia.com> writes:
> 
> Well, I guess I would draw the line right at the point where we 
> support the C/C++/C#/Java languages nicely.  This is certainly my
> own bias talking, but this would certainly capture the 80/20 cases.  

Fair enough, but even Perl does have multi-return stuff, as do
many other scripting languages I believe.

> I don't see this as an absolute requirement as Umit does, but I 
> like the idea that the next time (!!) I am writing a WSDL2Java, 
> I wont have to work as hard to create operation methods the same
> as GLUE or WASP or wsdl.exe, as the 'how should I do this [right]?'
> question in this case has been answered for me.

Again, fair enough. If the old parameterOrder solution seems to
work then maybe we should dust that off and take a look at it.

If you're wondering why I'm pushing so hard against this .. ;-):
its primarily because it breaks the overall approach for operation
styles IMO. We introduced a style notion, but now introducing 
additional (extension) attributes requires conflicing processing
rules. I think Roberto pointed this out to Umit too. Basically
its not just an XSD style any more but rather a whole set of rules.

If its critically needed, well, so be it. I'm not convinced that
it is, but obviously Umit is.

Sanjiva.

Received on Tuesday, 28 October 2003 15:09:16 UTC