Oliver Plohmann
2008-11-20 18:27:21 UTC
Hello,
lately I asked myself how to implement a tuple space. Couldn't resist
and had to sit down and do some prototyping. The idea is that nodes are
added to java.util.CuncurrentHashMaps to which locking is delegated. All
other synchronization takes place inside special tree iterators and not
in the nodes. As it turned out the problem was then how to remove empty
nodes to reclaim memory while allowing for a high degree of concurrent
accesses to the tree and carrying out the node removal process with
little lock contention. And I learned that writeSingleton is hard to
implement in a thread-safe way with little lock contention...
I put a sketch with a short explanation onto my homepage:
www.objectscape.org/tstree.pdf. Maybe somebody would have a little time
to peruse it somewhat and tell me whether the approach IHHO makes
somewhat sense or not. I have some suspicion that while what I came up
with is not bad, there might be something I missed. If my little tuple
space prototype basically works, I have some idea what to do next... ;-).
Regards, Oliver Plohmann
===========================================================================
To unsubscribe, send email to ***@java.sun.com and include in the body
of the message "signoff JAVASPACES-USERS". For general help, send email to
***@java.sun.com and include in the body of the message "help".
To view past JAVASPACES-USERS postings, please see:
http://archives.java.sun.com/archives/javaspaces-users.html
lately I asked myself how to implement a tuple space. Couldn't resist
and had to sit down and do some prototyping. The idea is that nodes are
added to java.util.CuncurrentHashMaps to which locking is delegated. All
other synchronization takes place inside special tree iterators and not
in the nodes. As it turned out the problem was then how to remove empty
nodes to reclaim memory while allowing for a high degree of concurrent
accesses to the tree and carrying out the node removal process with
little lock contention. And I learned that writeSingleton is hard to
implement in a thread-safe way with little lock contention...
I put a sketch with a short explanation onto my homepage:
www.objectscape.org/tstree.pdf. Maybe somebody would have a little time
to peruse it somewhat and tell me whether the approach IHHO makes
somewhat sense or not. I have some suspicion that while what I came up
with is not bad, there might be something I missed. If my little tuple
space prototype basically works, I have some idea what to do next... ;-).
Regards, Oliver Plohmann
===========================================================================
To unsubscribe, send email to ***@java.sun.com and include in the body
of the message "signoff JAVASPACES-USERS". For general help, send email to
***@java.sun.com and include in the body of the message "help".
To view past JAVASPACES-USERS postings, please see:
http://archives.java.sun.com/archives/javaspaces-users.html