[I'm cross posting to tuscany-dev and tuscany-user again, to be sure all the
Tuscany community sees this.  Please use"reply" rather than "reply-all" so
that we keep responses to the user list,  to help make following the thread
easier and continue to reach the wider audience]

Some of you may be aware that I'm involved taking SDO Java through the Java
Community Process (JSR 235) to achieve standardisation for SDO Java. One of
the things that this standardisation requires is that a reference
implementation (RI) and technology compatibility kit (TCK) be produced
alongside the specification document.

To satisfy the RI and TCK requirement we've been constructing a proposal to
Apache to create a new incubator podling [1] to do RI and TCK development of
JSRs related to SDO. To date the proposal suggests that development of the
RI would be best done separately from Apache Tuscany,  primarily because
developing an RI requires an environment that can focus on the Java
interface as specified, and in Tuscany we do cross language SDO, SOA and we
encourage innovation outside the spec.

I did some digging and questioning to find out if developing a JSR RI in
Apache was feasible,  and it is,  there are examples.  However, from the
feedback were getting to the proposal it's clear we should be thinking more
creatively about how the Tuscany SDO Java community and its aims overlaps
with the aims of the new proposal.

This has got me thinking now that perhaps we could have a discussion about
whether we, the Tuscany community, might like to take the opportunity to be
part of establishing a common home for SDO Java development in Apache.

A significant thing to consider is that, if you take a look at the proposal
[1] you'll see that  as part of the JSR agreement,  the responsibility for
creation of the RI belongs to BEA, and the RI would be created using an
initial donation of code from BEA of an SDO implementation. The RI would not
be based on the current Tuscany SDO Java. So at this point Apache has two
implementations, Tuscany's EMF based implementation,  and the new code base
seeded by BEA.

I'd like to try to understand how the Tuscany community feels about all the
aspects of transferring the general responsibility for developing SDO in
Apache to a common SDO project.  What do you think we, as part of the wider
Apache community, would do with our two implementations?  How much do you
care about what technology underpins an implementation?  Would you come and
be a community member in the new project?

[1] http://wiki.apache.org/incubator/NoNameYetProposal

Regards, Kelvin.

Reply via email to