Re: Creating a Query for a SHA-1 Hash

"Aaron Walkhouse" <[email protected]> Sat, 27 Dec 2008 23:09:22 -0000
Newsgroups gmane.network.gnutella.devel
Message-ID <[email protected]>
It's still the old-fashioed broadcast method for that type of search.
 Query routing didn't come out until much later and hash queries were
left out because there would be no benefit in filling tables with
hashes of hashes alongside the hashes of keyword fragments. 

Full hashes are so specific to their corresponding files that no
bandwidth savings would result in trying to route hash searches on the
way out. Simply letting them go to the end of their TTL down all
possible paths is still the most efficient way to deal with them even
if you don't respond or don't have a matching file.

Hits on the way back to a searcher would be very few and 100% accurate
since the specific file being sought is the only possible result. 
That means there is no point in trying to direct queries to one peer
at a time, a strategy best suited to searches that may either generate
a flood of replies or need a lot of retries down different paths to
satisfy the searcher.

Exactly how to send a hash search is probably in the docs or wikis,
since it's been in use for several years.  Ask the Shareaza developers
for examples and sample code too.