ACTION-45: Follow up on the specification made in Java or IDL

Jo Rabin proposed an interesting idea on today's conference call.

It is probable that more people know how to write interfaces in Java
than how to write them in IDL; could Java be used as a means to create
IDL?

Some things we would like to know are:

1.	Have people used Java successfully to create IDL (presumably
with the intention of using the resulting IDL to define compatible
interfaces in other languages)?
2.	Are there tools to assist in the derivation of IDL from Java
sources?
3.	Are there particular problems we should be aware of?

For example, what is the experience of using the "rmic -idl" tool that
implement the OMG mapping [1]? Would it be better to just use IDL from
the beginning?

One problem I am aware of is that it is not recommended to derived Java
code from the IDL that was derived from Java code. This "round-tripping"
can product unpleasant results. The down-side of this is that the
original Java would become the normative interface specification, not
the IDL. This is not what we would like to see.

Alternatively, the Java-to-IDL approach might be helpful in the early
stages, but ultimately we could move to a normative IDL approach.

Thoughts and comments to this list, please.

---Rotan.


[1] http://www.omg.org/cgi-bin/apps/doc?formal/01-06-07.pdf