This is a problem I've had since 2.1.0. (I'm using 2.2.1 in Ubuntu Gutsy now.)

KTorrent, when seeding, will constantly be connecting to seeds, and then disconnecting from them after the Availability column shows that they are a seed. Shouldn't KTorrent be ignoring other seeders when seeding, and not connecting at all?

Kulag wrote:This is a problem I've had since 2.1.0. (I'm using 2.2.1 in Ubuntu Gutsy now.)

KTorrent, when seeding, will constantly be connecting to seeds, and then disconnecting from them after the Availability column shows that they are a seed. Shouldn't KTorrent be ignoring other seeders when seeding, and not connecting at all?

And how do you figure out if a peer is a seeder when you only have an IP address and a port number ?

The only way todo this if trackers leave the seeders out of the reply. Not many trackers are doing this for what i know.

And how do you figure out if a peer is a seeder when you only have an IP address and a port number ?

I thought trackers informed clients which peers were seeds, since the other clients I've used don't seems to do this. My bad.

Maybe you could have KTorrent keep a list of all the peers on a particular torrent that are known to be seeds, and check against the list when connecting when seeding. (Or does it already do this?)

Kulag wrote:

And how do you figure out if a peer is a seeder when you only have an IP address and a port number ?

I thought trackers informed clients which peers were seeds, since the other clients I've used don't seems to do this. My bad.

It would be nice if the trackers do this.

Maybe you could have KTorrent keep a list of all the peers on a particular torrent that are known to be seeds, and check against the list when connecting when seeding. (Or does it already do this?)

Why bother ?

Why bother?

To save constantly connecting to peers you're just going to disconnect from anyway. I guess the savings would be somewhat trivial, however, compared to the time needed to do this.

Kulag wrote:

Why bother?

To save constantly connecting to peers you're just going to disconnect from anyway. I guess the savings would be somewhat trivial, however, compared to the time needed to do this.

I have far more important things to do then preventing a couple of short lived connections from happening.