I committed the initial version of a testbed that I wrote the weekend. It
doesn't actually do anything yet, but I implemented the protocol layer
necessary for it communicate over local PipedInput and OutputStreams, and a
DataStore that saves data in byte arrays rather then files to make it faster.
You can try it out by running "java Freenet.testbed.TestBed n" where n is the
number of nodes to run. Currently it will just start that many nodes, give each
of them refs to the 5 last, and run one Request with HTL 20 on them. Not very
useful as it always fails, but at least it works that far.
With 100 nodes running, a 20 HTL request takes about a second to finish on my
computer - and that is with Sun's dog slow JVM. There is still a lot to take
in performance on top of that as well (it currently uses the text protocol, I
will make RawMessage pluggable so the testbed can use a faster binary
protocol). In other words, it should be possible to make valuable tests of the
networks development with time, given a good computer and a few hours to run a
batch job.
The tallest order in the whole thing is probably being able to get some
visualization and statistics collecting in there. Varying settings between the
nodes is also a little difficult because we set them as static variables, but
something like writing a cancer node and adding it to see what happens should
not be too difficult.
--
Oskar Sandberg
md98-osa at nada.kth.se
#!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
$/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)
_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
