Wed 28 Feb 2007 11:38:45 PM UTC, original submission:
THIS ITEM WAS REASSIGNED TO TASK #7653
Hi,
gnutella and g2 seem to be mostly useless these days, at least when using mldonkey. I get (practically) no relevant hits, just the same dose of auto-generated spam. Somehow this doesn't seem to happen when using Shareaza.
For example, searching for "asdfgadcv" yields 72 results, e.g. "Unique asdfgadcv Promo (XXX).rar", "01 asdfgadcv (256kbps) (2).zip" and so on.
I believe it would be possible to automatically filter these, at least until the spammers come up with more sophisticated tools. If you search for a random string say every hour and add the hashes of resulting hits to a filter list, then the same spam hits needn't show up as hits when the user does a real search.
I'm not sure how feasible it would be to automatically block the IPs of the spammers; some of these files are pretty widespread (60+ sources reported on G2, but on ed2k I have seen upwards of several thousand).
Even if automatic filtering is not implemented, the option to manually ignore some search results in all future searches would be welcome.
Also, there seems to be an anomaly in search handling where search results from previous searches creep into the list of results for a later search. This only seems to affect Gnutella and G2. So, you search for foo, then several hours later search for bar, and you get a result for "foo" in the "bar" listing.
Additionally, the list of bootstrap/discovery services shipped with mldonkey is so outdated that none of the URLs work. For this reason, a new mldonkey installation can't connect to Gnutella at all. Shareaza seems to have an up to date list, maybe you could borrow that?
András
|