Home   Info   DevZone   Wiki   UsersWeb  |  MainWeb  |  InfoWeb  |  DevZoneWeb  |  SupportWeb
[ RheiseOSQuotes ] [ not logged in ] [ Web: Imported ] goto:  options [ get info on or edit ] [ login or new user ] [ list of topics, hubs & nodes, or recent changes ]

ProcessPages

Release of rheise.os - multiprocess Java layer for JOS

• After a fascinating discussion on the arch@jos.org list about how to support multiple processes and IPC in JOS, I have finally put together what could become a foundation for the Java layer of JOS: [A469]

rheise.os - update

• I have worked out a way to load a separate version of java.lang.System for each process in a host JVM environment. [A470]
• Bravo! Now I /must/ take a look at this as soon as I can. [A471]
• Correct. There is one thing I forgot to mention in my last email, though. [A472]
• I suppose I should mention what the system requirements are: [A473]

rheise.os questions

• Clarification: the SystemObjects argument to createProcess is the set of statics in java.lang.System that must be replaced from process to process, right? [A474]
• Correct. Sort of like the optional envp argument to exec() on UNIX. [A475]
• On this point, the nativ package in the latest release is slightly out of date. [A476]
• We want decaf to know as little as practical about processes to ensure that its behaviour is not distinguishable from a conventional/single-process JVM. [A478]
• Okay, makes sense. [A479]
• I'd like to make it generic, but I suspected that things might get environment specific when we start to consider how we do object sharing is done. [A486]
• The idea was that setIn() is for user programs..., and a special init() method would only be executed by JOS and it would just set the attribute values directly. [A487]
• I disagree. There must be two mechanisms. [A488]
• Maybe that way's better. [A489]
• I wasn't particularly expecting you to handle the native-side of the nativ implementation :) [A492]
• I'll second this. [A493]
• Agreed. [A494]
• OK, I understand your point. [A495]
• In theory, we can run all AWT-dependent applications on JOS /because/ the implementation details are completely up to us. [A496]
• We even have proof of that, with SwingToolkit :-) [A497]

rheise.os-0.1.2 released

• This will be the last combined rheise.os/SwingToolkit release. [A499]
• How'd you manage this, BTW? [A501]
• I'm just leaving for work, but I can give you some quick directions. [A502]
• I didn't really "move" to doing that. [A504]
• I didn't look at enough of the code, obviously :) [A503]
• The System wrapper class is required to prevent one process from writing over the System.in that is being used by another process. [A505]
• Now I'm certain I wasn't thinking clearly. [A506]

rheise.os-0.1.3 released.

• One of the main changes in this release is that the manager classes are now purely static, meaning you can call ProcessManager.createProcess() instead of ProcessManager.get().createProcess(). [A522]
• At the right time, implementation should be separated from its interface. [A530]
• That was actually the goal of this release. [A538]

Content of these pages are owned and copyrighted by the poster.
Hosted by: