Raphael Manfredi wrote: > Although GTKG correctly sends SHA1 searches, many servents do not know > how to handle those and will not route them correctly based on the > routing tables. > Therefore SHA1 queries are a waste of Gnutella banwidth space, which is > why GTKG currently does not automatically requery by SHA1 when there are > no more sources available.
However, it *does* work. Some content is likely to be present in many different versions. In this case searching by parts of the filename is probably better. For other contents especially those that are not supposed to be modified (by users), searching by SHA1 works much better. There's still a problem with the latter though. If anyone returns a fake result, GTKG will hardly be able to detect that the data doesn't match (while downloading) and the file will then turn out to be corrupt once the download is finished. If you search by filename it is less likely that a hostile peer can figure out the corresponding SHA1 and corrupt your download. Well, I usually leave searches open until the downloads are finished, so searching by SHA1 would hardly improve anything. IIRC, GTKG stops a search when a certain number of results has been returned, so this might not help finding alternates after awhile either. -- Christian
pgpGENHz5hOEB.pgp
Description: PGP signature
