I wrote recently about an interesting JSR proposal JSR 323 – Strong Mobility for Java. This proposal concerned allowing the full running state of a Java program to be serialized and deserialized to continue running elsewhere.
In interesting news, the JCP just voted down JSR 323 with 8 no votes and 3 abstentions. It seems like the consensus was that both the technology is premature and that no such standard is needed. Some of the more interesting comments below.
Sun believes this could be an interesting area of research, but does not yet see the kind of community or vendor support that would motivate standardizing on a mechanism in the JCP. Currently we think that os virtualization techniques are providing more complete solutions for the kind of problems outlined in this proposal.
Generally, for standardization in JCP, the request should represent the interests of varied approaches to a problem or various alternative implementations where there is a community that needs standardization in order to progress. We don’t see evidence for that in this JSR. We also aren’t convinced that, if this were to be done in JCP, it isn’t better to do it in the VM rather than as in this proposal. Research in this area seemed to have tailed off a few years ago without any consensus, with interest shifting to hypervisor based virtualization. We aren’t convinced there is a need for the standardization suggested in this JSR.
This work is not ready for a JSR. I encourage the proposers to work on and gain experience with implementations before re-proposing as a JSR.