5 ms·

The Eight Fallacies were the impetus for the development of Jini and JavaSpaces. I wrote about this many moons ago here: http://www.softwarematters.org/jini-i

The Eight Fallacies were the impetus for the development of Jini and JavaSpaces. I wrote about this many moons ago here: http://www.softwarematters.org/jini-intro.html http://www.softwarematters.org/jini-intro.html Jini is one of those technologies that should have taken off but somehow didn't. At the time, the Jini proponents attributed the failure to the fact that J2EE sold more hardware. That might have been a factor, but I suspect the fact that Jini was just too much outside the mainstream was a bigger issue. Not unlike Lisp, really.

rdtsc 12y ago
I would imagine being tied to Java at least by name (JavaSpace, Jini) is an impetus to push others away from it. That is why a "Soup of REST-ful services" is becoming the lowest common denominator for distributed architectures (I think the buzzword I am supposed to use is "micro-services"). Because it is not tied to a language or platform. Speaking of Sun's distributed systems, anyone remember JXTA? Looked really fun on paper, peer-to-peer overlay networks, send messages to anyone anywhere, works just "like pipes". Anyone ever used that? I learned about it school at one point during a project.
- patrickmay 12y ago
  REST is great for decoupling but as generally implemented it adds too much latency for some types of distributed computing. It also makes distributed transactions difficult to manage. Jimi addressed a lot of those problems in a technically interesting way. The protocol itself is language agnostic, but the implementation took advantage of Java's mobile code capabilities.
  - rdtsc 12y ago
    > It also makes distributed transactions difficult to manage. Wonder if the typical REST-ful answer would be to make yet another service on top that manages distributed transactions and talks to a variety of other services. Sort of like APIs wrapping another APIs which wrap other APIs...
wstrange 12y ago
I have a feeling that Jini will be re-invented in the future. There are too many good ideas in there to ignore. "The End of Protocols" paper is another great read.

rdtsc 12y ago

I would imagine being tied to Java at least by name (JavaSpace, Jini) is an impetus to push others away from it. That is why a "Soup of REST-ful services" is becoming the lowest common denominator for distributed architectures (I think the buzzword I am supposed to use is "micro-services"). Because it is not tied to a language or platform. Speaking of Sun's distributed systems, anyone remember JXTA? Looked really fun on paper, peer-to-peer overlay networks, send messages to anyone anywhere, works just "like pipes". Anyone ever used that? I learned about it school at one point during a project.

patrickmay 12y ago
REST is great for decoupling but as generally implemented it adds too much latency for some types of distributed computing. It also makes distributed transactions difficult to manage. Jimi addressed a lot of those problems in a technically interesting way. The protocol itself is language agnostic, but the implementation took advantage of Java's mobile code capabilities.
- rdtsc 12y ago
  > It also makes distributed transactions difficult to manage. Wonder if the typical REST-ful answer would be to make yet another service on top that manages distributed transactions and talks to a variety of other services. Sort of like APIs wrapping another APIs which wrap other APIs...

wstrange 12y ago

I have a feeling that Jini will be re-invented in the future. There are too many good ideas in there to ignore. "The End of Protocols" paper is another great read.