Jim Waldo wrote:
Well, as someone who participated far more actively than Jonathan in that action, let me assure you that his account was not a precise description of what happened. The terms of the settlement was, for the most part, that Sun would license the Kodak patents for a considerable sum. It didn't have any impact on the further development of RMI-- what did seem to limit that was the Java community process and the pressure to not put anything into RMI over JRMP (which I still think of as RMI) that couldn't be mirrored in RMI over IIOP.
Luckily were not constrained by Management or client demands, only
resources ;)
SPI mechanisms have been talked about for some time. I've never been convinced
that these add much, but if that is the way the community wants to go, there
are certainly designs that have been proposed that could be used as a starting
point. JERI actually allows a lot of plug-compatibility for new forms of
implementation, but if that is not sufficient, other things could be discussed.
Someone would have to be interested enough to drive it's development, or
contribute patches, I don't know the history behind it I'm afraid.
Perhaps someone might be kind enough to highlight a current issue that
an SPI implementation would solve?
If there is anything that is an issue for me, it is that Java
Serialization publicly publishes class private state or implementation,
JERI RMI is very good, I want to look into Serialization at some stage.
Cheers,
Peter.
Jim Waldo
On Apr 16, 2010, at 8:37 PM, Peter Firmstone wrote:
I'm guessing from Jonathan's recent comments on his blog, about Kodak suing
successfully for damages over RMI, that this impacted RMI negatively, nothing
has been done with RMI for a long time, I wonder what the terms of the
settlement were?
Cheers,
Peter.
Gregg Wonderly wrote:
What I would like to see is a mechanism for completely extracting the use of
the java.rmi package use into an SPI like mechanism. The ability to then plug
in some other details or more limited environments like those would be possible.
I don't know any details, but I suspect that some of the use of RMI stiff was
left over from the JSRs and maybe some was there as preparation for another
eventual JSR attempt.
In the end, RMI ended up not getting as much support, from the top brass and in
the Sun Developer Advisory Council meetings, I detected some personality
conflicts that were probably also a factor. Jonathan eliminated some of
those...
Gregg Wonderly
Sent from my iPad
