I was referring to the new Morpheus client, which is nothing more than Gnucleus with most occurrences of "Gnucleus" changed to "Morpheus". (If you look through options screens you'll see a few instances of "gnucleus" still in there.) I agree that the FastTrack-based client was pretty good; much better than anything the gnutella network has to offer at the moment.

I'm fairly sure the only difference between the various versions released is in the packaging and installation routines.

Someone else asked for an easier way to get a patched version. The two main reasons I didn't just provide a link to a windows installer are 1) that I don't run windows, and 2) that I don't have anywhere to host the file.

diefastr: Let me know how your patched version works out. In my experience resuming hasn't been completely foolproof after the patch, but it works most of the time. Unfortunately, it seems that the version of the source available at the link I gave above has another bug which causes instability after running for a few hours.

Finally, let me try to provide some insight into why resuming suddenly stopped working after version 1.8 or so. It's mainly because of the "swarm downloads", or whatever they're calling them. In order to support downloading the same file from multiple hosts, LimeWire splits up the file in a number of smaller parts, each assigned to one host.

To take a simple example: Let's say you're downloading a 10 meg file from two hosts. LimeWire will start downloading the first 5 megs from one host, and the second 5 megs from the other. If the download breaks off at 50%, you'll have an incomplete file containing the parts from 0-25% and 50-75% of the original data. (This is assuming the download speed has been the same from both hosts, otherwise you'd have different percentages.)

This, naturally, makes resuming a much more difficult process, because LimeWire can't just take a look at the size of the incomplete file and start off at the end. It has to keep a record of all the incomplete files and which parts of them are complete. There were mechanisms for this in the source I downloaded, but it wasn't working properly. My patch fixes some of it (probably not all), but the one thing it doesn't fix at all is all the incomplete files you had prior to installation of the patched version, because it simply has no way of knowing which parts of those incomplete files are missing.

I was planning on adding a mechanism for recovering "unknown" incomplete files by assuming they've been downloaded continuously from a single host, and not split up in any way. This should work in most cases, at least for my files; I just haven't gotten around to it yet.

[EDIT]By the way, this is the same Unregistered user who posted the patch above. I decided to register, to avoid confusion.[/EDIT]

I just bought the pro version but I am disappointed. Resume obviously does not work. I have also had it hang a couple of times. Probably a function of being based on Java. This is a 2ghz XP machine with 1 gig of memory. Everything else works great. I seriously doubt my problems are machine specific. I really like the idea of getting away from kazaa to a truly public network such a gnutella with what I thought was the best client "limewire." So far it's a no go.

So from what I can tell so far there is no solution to the "resume file download question" and it doesn't work on the paid pro version either.

Does that sum things up?

My patch still works, at least for me. I just did a checkout of the latest source from cvs, made the same modifications there, and compiled; resuming works fine.

New patch
 

Attached to this message is a new patch, applicable to the current cvs tree. It tries to deal with old incomplete files in addition to doing what the other patch did.

Instructions for CVS checkout can be found here: http://core.limewire.org/servlets/ProjectSource. (To compile a working client you'll need the gui project as well: http://gui.limewire.org/servlets/ProjectSource) Remember to modify the gui run script to use the patched local core project instead of the bundled jar.

Apply the patch by placing it in the root directory (below the core and gui directories) and typing
patch -p0 < &lt;patch-filename&gt;.

ok, this patch sounds great, but I have no idea how to use it...
can someone post step by step "patching for idiots" instructions? from what I understand, we have to modify the code and compile it, but I just don't know how to do that. it'd be nice if someone could do all the modifications and just post the file online somewhere for ppl to download, or, even better, contact the developers and get these modifications included in the next version of limewire so that we don't ever have to deal with this again...

As I recall somebody saying part of the resume problem is the swarm downloads, I think part of the solution would be to save each download as a separate file, then merge the files when they are all done. (ie save File A as File A (1 of 5, goal: 0 to 100 kb), File A (2 of 5, goal 100 to 200 kb) and on down the line). When these are merged, these fragments can be deleted, either by the program or manually.

Only problem I can see with above would be character limits on some operating systems.

would the patch also fix the problem with different file names?
the separate files idea sounds like it would