This morning I restarted limewire and let it settle down, then attempted to do some searches.

I put in one search term, hit enter, and my leaf promptly dropped one of its ultrapeer connections and then locked up. After several minutes, it recovered -- UI responsive again, etc. -- and I aborted the search (it had found nothing, unsurprising since it got launched with less than all green bars). I went to run the search again once LW recovered its missing connection, and LW froze again, but only for ~20 seconds this time. When it recovered it still showed all green bars, so I launched a couple more searches. After the second there was another pause, closer to 30 seconds; it still had all green connection bars at this point, so I launched a third. This time it seized for well over a minute. When it recovered, still all green bars, but the first search was well over half done and the second was approaching half, and none of the three had found a single file. And this was with queries that normally return 100+ each.

I contemplated aborting the searches, convinced that something had gone badly wrong somewhere, but before I actually touched the Limewire UI again, it dropped every single connection -- all of them, simultaneously as near as I could tell.

Needless to say I aborted all three searches.

It took nearly 5 minutes for it to recover to a status of all green bars. When I then went to relaunch the searches, there was a brief pause after launching the first, maybe 20 seconds in duration, but everything worked normally. The searches generated results, but maybe a quarter as many as expected for each. A repeat of the searches after harvesting the files I wanted that showed up in the first batch of results produced completely normal sized result sets.

What the hell is going on under the hood here guys? What caused this ... crash, or whatever it was? Why did it apparently not forward my searches to any ultrapeers initially -- and then later, drop all of its connections without my say-so? It had to be a problem at my end, since otherwise there had to be independent, simultaneous failures at each of several remote machines. But it wasn't a network problem -- and here's the evidence.

• It didn't simply show "disconnected" the whole time. It would have if there'd been no functioning network connectiom.
• Other network-utilizing apps, such as my Web browser, were functioning normally.
• I had some old downloads waiting in line or for busy hosts, which wouldn't be possible with no functioning network connection. They'd have shown "need more sources".

So it wasn't the network, from the API between Limewire and the OS on down to my ISP's gateway. It wasn't the remote hosts. That only leaves the Limewire. And it is a beta. Obviously, what we have here is a bug. Unfortunately, it may prove elusive and difficult to reproduce.

Also, I'm experiencing a weird problem with this BB site and Firefox 1.0.4. If I post, or even preview or something and then decide not to, that Firefox window will no longer load anything from gnutellaforums.com -- it will just spin, or pop up a timeout error, or otherwise behave incorrectly instead. If I close the affected window, open a new one, and navigate to the site again, it works fine again -- until the next time I submit or preview a reply, and then it does the same thing again. This started this AM sometime -- everything worked normally yesterday. Clearing the cache doesn't help. Restarting the window or the entire browser process does. Other sites, including very similar looking forum sites, are not affected. I haven't changed anything at my end. I think this is at your end, and suspect a cookie handling problem.