From: <tho...@us...> - 2014-01-11 18:54:22
|
Revision: 7774 http://bigdata.svn.sourceforge.net/bigdata/?rev=7774&view=rev Author: thompsonbry Date: 2014-01-11 18:54:15 +0000 (Sat, 11 Jan 2014) Log Message: ----------- javadoc comment from apache river dev list on RMI and interrupts. Modified Paths: -------------- branches/BIGDATA_RELEASE_1_3_0/bigdata-jini/src/test/com/bigdata/journal/jini/ha/TestHAJournalServer.java Modified: branches/BIGDATA_RELEASE_1_3_0/bigdata-jini/src/test/com/bigdata/journal/jini/ha/TestHAJournalServer.java =================================================================== --- branches/BIGDATA_RELEASE_1_3_0/bigdata-jini/src/test/com/bigdata/journal/jini/ha/TestHAJournalServer.java 2014-01-11 12:49:05 UTC (rev 7773) +++ branches/BIGDATA_RELEASE_1_3_0/bigdata-jini/src/test/com/bigdata/journal/jini/ha/TestHAJournalServer.java 2014-01-11 18:54:15 UTC (rev 7774) @@ -413,6 +413,20 @@ * I hope this helps, * Brian * </pre> + * <pre> + * This is one of the places where a lease could help. An extension of the + * existing JERI details could add a lease into the dispatcher layer so that + * a constant \xD2I am here\xD3 message would come through to the service. If the + * client thread is interrupted it would no longer be pinging/notifying of + * it\xD5s interest in the results. That would allow the service end, to take + * appropriate actions. I think that I\xD5d want the export operation or + * exporter creation, to include the setup of a call back that would occur + * when an client wants something to stop. I would make the API include a + * \xD2correlation-ID\xD3, and I\xD5d have that passed into the call to do work, and + * passed into the call back for cancellation. + * + * Gregg + * </pre> */ public void test_interruptRMI() throws Exception { This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site. |