It turns out that some recent changes in the freenet code have touched
flaws in the Sun 1.3.1_02 jre. It worked for 0.5pre2 but doesn't for
nightly builds after that build was released. I had thought that I
upgraded the jre but appearently the rpm released by Sun does not change
the symbolic link in /usr/bin/ for the java app. It continued to point
to the old jre installed either with an old original installation. I
just remade the symbolic link and all was good.
Anyways ... I remain stuck at 0.5pre2 though. While I can now get java
to run all the snapshots, I can't retrieve anything through them. When I
switched to 2002-10-13 snapshot, it had datastore issues (I am running
the new native datastore)
Exception in thread "main" freenet.KeyException:
java.lang.reflect.InvocationTargetException
at freenet.Key.readKey(Key.java:84)
at freenet.node.rt.ReferenceSet.<init>(ReferenceSet.java:61)
at freenet.node.rt.ReferenceSet.getProperty(ReferenceSet.java:27)
at freenet.node.rt.TreeRoutingTable.<init>(TreeRoutingTable.java:42)
at freenet.node.rt.CPAlgoRoutingTable.<init>(CPAlgoRoutingTable.java:34)
at freenet.node.Main.main(Main.java:483)
I deleted the store directory and let the node rebuild a fresh one (I
suppose it was a instinctual reaction from the monolithic datastore
corruption days). The node started fine but as soon as I looked for any
keys it came back immediately with a DNF (not RNFs as I've experienced
long ago when Freenet was overloaded) no matter what htl I used. This
was the case with 2002-10-11 snap on up. I didn't try earlier ones. At
first I thought that I might have had bad seednodes so I tried some
older ones with no luck. However, As soon as a downgraded to 0.5pre2
then the node actually took the time to search for the keys and found
many successfully.
Is anyone else seeing this kind of behaviour with the recent snapshots?
Mike
_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl
