On Dec 5, 2007, at 2:22 AM, Rick McGuire wrote:
Kevan Miller wrote:
On Dec 3, 2007, at 1:45 PM, Rick McGuire wrote:
Below is a proposal that Matt Hogstrom, one of the mentors of the
Yoko project, has put forward for moving on with the Yoko
project. In a nutshell, the Yoko community has basically decided
there is not a lot of continuing interesting in moving this
project forward. This decision does have a major impact on
Geronimo, as Geronimo uses the Yoko ORB was a key element to
allow Geronimo 1.2 to support both the 1.4 and 1.5 JVMs, and also
was a necessary element for achieving j2ee5 certification. The
Yoko ORB gives Geronimo cross JVM portability which was not
available before. At the current time, there's probably no
suitable replacement that has all of the advantages that the Yoko
ORB provides.
In a nutshell, Matt's proposal is for the core ORB elements of
the Yoko project become a subproject of the Geronimo project.
These are the pieces of Yoko that Geronimo has a dependency
upon. These are essentially the org.omg.* clases, the
javax.rmi.* classes, plus the implementation classes backing
those spec interfaces. Along with the subproject, there are 6
committers who have expressed interest in continuing to work on
the core ORB code. 3 of the interested commiters are already
Geronimo committers. Matt's proposal would grant the remaining 3
Geronimo committer status as well.
There's one important caveat in assuming owership of this
package. The core ORB is also used by the Harmony project to add
CORBA and RMI support to the Harmony JVM. Included with assuming
ownership of the package would be a commitment to keep the core
ORB a "standalone" component. This means not adding direct
dependencies on Geronimo and keeping dependencies on other
packages to a minimum.
This code is fairly stable now, and has already passed
certification on multiple JVM instances, so I don't expect there
will be a lot of overhead in supporting this. The bulk of the
recent work to get this to pass certification have been done by
Geronimo committers, so Geronimo is probably the most appropriate
new home for this code.
Anyway, this needs to have some discussion and be put to a vote.
Below is the proposal that Matt posted to the Yoko dev mailing
list about this move. The Yoko community seems very much in
agreement that project does not have sufficient momentum to
graduate from the incubator.
Thanks for the summary, Rick.
I'm certainly interested in seeing support for Yoko move forward.
This seems like a positive move. It would have my support.
After a brief review of the Yoko dev list archives and based on
Matt's, and Alan's recommendations, I would support adding Lars,
Alexey, and Darren to as Geronimo committers.
Keeping Yoko as a standalone component is an easy decision, IMO.
Hard to see it any other way...
Actually, I have a whole laundry list of things that could be done
to Yoko to make it work better in the Geronimo environment that
could mess it's ability to function as a standalone server if not
done "correctly". For example, it would be nice if Yoko could hook
into the Geronimo thread pooling APIs. It's easier to ensure
things like this are done in the correct fashion if the constraint
of needing to remain standalone is stated right up front.
You make a good point. We should be very explicit about the
requirement that Yoko be standalone.
Regards,
Alan
