users@jax-rpc.java.net

Re: Using unbounded sequences

From: Sunil Kunisetty <Sunil.Kunisetty_at_oracle.com>
Date: Wed, 24 Mar 2004 09:50:42 -0800

 Anne-

 Couldn't resist chiming in.

Anne Thomas Manes wrote:

> Roberto,
>
> I agree that it would be very useful if there were a formal specification
> of the wrapped convention. You might consider doing that in the next
> JAX-RPC maintenance release.
>
> I don't understand your reference to NIH, and I don't see how
> "childishness" comes into play in this discussion. The RPC convention was a
> mistake. It's a *bad* practice. It is not a convention that we should
> promote or propagate.
>
> There are lots of people at WS-I that really wanted to exclude the RPC

 I remember otherwise or may be you are referring to "SOAP Encoding".
 Yes, "SOAP Encoding" was (almost) unanimously discarded. Where as
 rpc/lit was voted very much in favor of it. This was voted at the first F2F
 at Oracle and we have 3 options on table:
 i) only rpc/lit
 ii) only doc/lit
 iii) rpc/lit and doc/lit

 If my recollection is correct, 17 or so companies have chosen (iii) and just
 2 or 3 opted for (ii).

 I also humbly disagree with that a rpc/lit is a *bad* practice. Infact, I feel that
 doc-wrapped/lit is somewhat bad as there is no formal specification for it
 as you also seem to agree with that fact. The biggest problem I've with doc-wrapped
 is that the "dispatching" (i.e. dispatch to which operation???) is not formal
 and by-default the message is dispatched to the tag name of the "wrapper".
 And also do note that doc-wrapped requires "additional" wrappers (one each
 for each operation).

 <snip>

 -Sunil