Hashes and Swarming
 

The GDF is discussing the introduction of file hashes into the protocol for a very long time now. Meanwhile, programs like Xolox have already implemented hashing on their own, and it seems to work. My question is: When are we to expect a Lime version that offers file hashing, safe resuming and swarming ?

Have you ever thought about a resuming mechanism like phex or gnucleus? It is next to impossible to download a file from multiple sources using lime, even without swarming. I asked for a phex-like download candidate list many times now, not necessarily one that auto-researches for new mirrors, but one that can be refreshed with new mirrors if the old ones are out of date and that would resume a file with the same length from a different source.

Using candidates and hashes, the introduction of swarming should be a fairly easy task. It does not even need queries for hashes - one can just search again using the original search term, check the size of the results, ask the hosting peer for the file's hash and download if it is identical.

I clearly consider Limewire the gnutella reference implementation - it has always been one of the most advanced servents and now even open-source (besides i am a java fanatic)... but i must say that (and often did) that its download capabilities are not suitable for anything more than downloading in a one-source-per-file manner. on a connection breakdown one can as much as never complete the file without a complete reload.

You must work on this! It is the one task Limewire has to master, not the implementation of an internal mp3 player and not even metadata are as important (though i greatly like the current supernode/fastpeer/metanode i don't know what you decided beta).

(1) When to expect hashes ?
(2) When to expect stable candidate-based resuming ?
(3) When ro expect swarming ?

Any comment of a Limewire developer would be greatly appreciated.