Hi Tatsuhiro, hi list, Le 05.08.2010 à 17:05, Tatsuhiro Tsujikawa a écrit : > Hi, > > (08/05/10 11:55), Peter Pöml wrote: >> Hi Tatsuhiro, hi *, >> >> Am 21.07.2010 um 17:43 schrieb Peter Pöml: >> >> >> I believe I have fixed the issue, and download.services.openoffice.org >> delivers the correct BitTorrent Information Hash now: >> >> % curl -s >> http://download.services.openoffice.org/files/localized/ja/3.2.1/OOo_3.2.1_Win_x86_langpack_ja.exe.meta4 >> | grep magnet >> <metaurl >> mediatype="torrent">magnet:?xt=urn:btih:78ebf160c1924d2ff098535bb0362172f467708c& >> [...] >> >> This is not the same hash as the one that torrent.projects.ooodev.org gives >> -- due to a difference in block size (262144 vs. 131072 bytes). Otherwise, >> it should fit. I tested with the smaller block size and verified that the >> hash would be identical in that case. >> > > Thanks Peter. > > I think this is a temporary issue until next mirror brain release, but I > think btih info hash in magnet URI and info hash in tracker must be > matched to make magnet URI work. > Because magnet URI does not contain Metadata, BitTorrent client must > retrieve it from other peers. It first contacts tracker with btih info > hash. If btih info hash and infohash in tracker are different, tracker > complains that it does not know that torrent and no peers are returned. > When using current .meta4 file, all tracker returns failure message, > which means magnet URI is not usable.
I see. Indeed, this is the case now. I have made the chunk size in MirrorBrain configurable now, and I matched it to the chunk size that the torrent.projects.ooodev.org size uses. So now, the btih is the same. On a related note, since none of the used trackers currently allows unregistered access, I have set up an experimental tracker on my own (bt.mirrorbrain.org:8091) which allows it. For now, that's the only tracker sent along with torrents delivered by download.services.openoffice.org. Hence, the torrents should actually be usable now. I could successfully download various files a number of times today, when experimenting. Thanks a lot for your feedback, Peter --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
