Incomplete files can be shared if done with some intelligence :

If a host is in the middle of downloading a file, it can report the original filename (without the "T-filesize-") and finished filesize to queries by clients.

If a client asks for a resume, and the resume location is available, the host will begin sending.

If a client should download all the way to the end of the incomplete file on the host, the host will keep sending it if the host can download it at close to the same rate it has been sending it at. If it is downloading it at a significantly slower speed, it will disconnect the client so that the client can look elsewhere.

Unfortunately the devil hides in the detail. It's not really possible to share incomplete files without HTTP1.1 downloading, since HTTP1.0 doesn't let you specify which part of a file you want in the request.

So, why not use http 1.1? He who leads the pack, leads the pack. Let's go.

This is an excellent idea.

I was wrong: It is possible with HTTP/1.0 - and it would be possible for LimeWire, as it is now, to do that.

If a host would allow the uploading incomplete files, it would have to keep track of which parts of the files were written (which it does anyway) and if a part of a file is requested, it would have to return any sub-part of it. The downloading host would than see, that it is not the part it specified and could adopt to it quite easily.

I played around with the source code a bit (no I'm not a programmer and all my attempts to modify limewire are somewhat amateurish, but I'm trying to learn java), and I don't think it's worth the trouble right now. Once HUGE with hashes et al is implemented it would be far easier to implement, when you could get a file by its SHA1-hash instead of its name.

One major problem is the following: Sharing incomplete files would make corruptions happen far more often. Right now, LimeWire will detect, if a file was corrupted through swarming in the process of download. With uploading incomplete files, however, it could upload corrupted file parts it didn't detect yet, which could spread pretty fast through the network.

Question. Is this mythical hash-enabled version going to happen any time soon, do you think?

Question two: can we save our incomplete files somehow so that when the new version does come out, we can save ourselves from having to download the same stuff all over again?

1) I can't tell you, when it's going to be released. But I believe (if LimeWire doesn't make another bugfix-release) the hashing version will be the next version to be released.

2) I don't know whether the next version will be compatibel but I don't think so - so you might want to complete your old downloads first with the old version, before installing the new version.