John Reiser said: > William Hooper wrote: >> John Reiser said: >> [snip] >> >>>All of this depends on the bittorrent clients+servers transfering a >>> whole >>>directory file-by-file, instead of as random pieces from all the files. >> >> [snip] >> >> But this isn't how Bittorrent works. It does random pieces so that >> everyone won't be after the same piece at the same time ... [snip] >> >> http://bitconjurer.org/BitTorrent/protocol.html > > Using bittorrent-3.3-0.fdr.2.fc1 as client for FC2 test 1 from > http://torrent.dulug.duke.edu/FC2-test1-binary-i386.torrent > I observe that FC2-test1-i386-disc1.iso is the only file which > receives data for the first 4 hours, [snip] So the observed behavior > definitely is _not_ random across all files in the directory. Maybe behavior has changed in the new version, but I believe you will find that you may not have a complete disc1 when it goes to disc 2. http://www.redhat.com/archives/fedora-test-list/2003-November/msg00618.html > In such circumstances, it would be handy to have useful partial > results, such as -disk1 complete and checkable via MD5SUM. One could always get the MD5SUM from the web page... http://torrent.dulug.duke.edu/MD5SUM -- William Hooper
