why don't we use md5 checksums to identify the files that are equal (the same) instead of using the name of the file.

the client program could calculate the checksums a put it on a index table (file name -> md5 checksum) and when it sends a query the checksum its in the first characters of the file name.

this way we have a better hit-rate for file and an incresed probability of resuming downloads. all this without modifying the gnutella network servers and packages.