Network Fragmentation
 

It seems that qtella and limewire connect to completely different portions of the gnutella network. This never used to be the case, there was a time when almost every result I got searching in qtella was hosted on a limewire client. Now, virtually none. Doing the same search in both limewire and qtella shows dramatically which has the better connection, and I'm sorry to say it's limewire by a long ways. qtella's ui is leaps and bounds ahead, but the network connectivity is greatly lacking. I still don't seem to ever upload anything, and I haven't noticed the suposedly new swarm downloads in action.

I'm using the unstable 0.5.0 tar.gz under debian w/ KDE 2.2.2.

I LOVE some of the UI refinements, tho. Excluding files witht he same size as ones you already have is so simple yet sich a time saver. Seems lightning fast and responsive after limewire's java UI.

At this point I would suggest that development be strongly focused on improving connectivity and letting the GUI just completely sit for a while. Ransake the limewire core code, if nothing else. As much as I love qtella, it simply can't find rare files. For mp3s, its fine, there are so many people with every mp3 you might search for it makes no difference, but for rarer things, forget it.

Thanks,

-Dan

A possible explanation is that limewire implemented ultrapeers. This reduces the bandwidth for most users, and the searches will be better.

Somewhere I read that Etzi was thinking of adding ultrapeer support to qtella.

UltraPeer
 

That would be cool :)

I'm not sure its entirely nescessary, tho. It might be helpful just to better interact with the ultrapeer framework, for the moment.

I'd agree that the UI is fantastic, the best I've found in any P2P client... I can't see many other necessary refinements, mind you I don't think Etzi has been spending all that much time refining it because I've seen very few changes since I started using Qtella... just some extra fine tuning to the system. He seems to have been doing a lot of work on the guts of the system..

The most evidence I've found of multisegmented downloading (he never mentioned swarming I belive) is that I can group searches by sizee so it groups the same files in a tree-like structure, but I'm not sure if it's actually downloading from all the discovered sourcesat once.

It will be nice when ultrapeers are implemented. It'd also be really cool if a C programmer had a hack around with the Qtella code to see if he/she could find any improvements to its guts, because I'm sure staring at the code looking for improvements isn't the most entertaining or productive use of Etzi's time...