On Sun, Oct 14, 2001 at 01:38:12PM -0400, Gianni Johansson wrote: > Hi Oskar: <> > Grudgingly I might have to admit that this looks like a JVM bug.
Try making the id var in Message.java non-final to see if that helps. Since we have seen JVMs have problems with final vars in the past. If that doesn't work, try applying this workaround patch with undoes the only slightly out of the ordinary thing about the way the id is set (very slightly). > Drain sent me his logs, and I have kept mine as well. Let me know if you want > them. > > Has anyone else seen the "YOUR NODE CAUSED THE HEISENBUG." message in their > logs? > > -- gj > > P.S. > Does anyone know where I can get a rh6.2 rpm for kaffe that has the bigint > patch? You'll want to build Kaffe from CVS. They release way to seldom. -- Oskar Sandberg oskar at freenetproject.org -------------- next part -------------- Index: Message.java =================================================================== RCS file: /cvsroot/freenet/Freenet/Message.java,v retrieving revision 1.46 diff -r1.46 Message.java 26c26 < public final long id; --- > public /* final */ long id; Index: message/NodeMessage.java =================================================================== RCS file: /cvsroot/freenet/Freenet/message/NodeMessage.java,v retrieving revision 1.9 diff -r1.9 NodeMessage.java 44c44,45 < super(source, readId(raw.fs), raw); --- > super(source, 0, raw); > id = readId(raw.fs);
