The solution to this problem is really quite simple and has been done by both Limewire and Gnucleus, and I cannot understand why it hasnt been implemented in more popular gnutella clients such as Bearshare and Gnotella. It has been done also in Kazaa/MusicCity and I believe Napster had done the same to their software before their demise.

The solution is simple. Compare filesizes and filenames of returned results, and create a download queue which, if the download is interrupted, the client will go through and resume the file from another location. This can be performed in a loop so that if all hosts in the queue are either offline or busy, the download can continue when a host gets a free download slot.

**For those of you who might claim that comparing filesizes and filenames is not feasible due to the amount of files on gnutella with similar names/sizes, the benefit of such a feature greatly outweighs the extremely rare inconveniences of an occasional bad search result.

This already has been implemented, and I urge all Gnutella developers to create such a feature for the overall reliability of gnutellaNet and to distribute bandwidth across the network evenly.

Omar