When will the virtual memory consumption problem be fixed?
The longer lw runs on osx, the more virtual memory it sucks, and then it ends up crashing.
Even left alone by itself, lw still grabs memory.
The only way I've found to avoid this, is to turn off all the extra tabs (monitor, library..), turning off ultrapeer and hide lw as soon as it's started, and leave it there.
Basically, to avoid the problem is like not using it.
This issue has been know and was "supposed" to be fixed by 4.4 or 4.5, so will 4.6 fix it????

I wish.

But we still can't reliably replicate the VM growth to figure out the cause. For example, LW ran all this weekend without the growth here. Last I heard the VM might depends somehow on who is connected to you!?! (the "cloud").

Hmmm--I keep trying to forget about this until Apple releases a java update, but I wonder if type/number of incoming calls might be a factor . . . .

For me, it's very reproducible, it's every single time I try doing anything, the slightest little thing can trigger it. When the growth is triggered, it continues on by itself and doesn't stop until it either crash or it gets restarted to avoid the crash.
The only way I found to prevent it, is to hide it as soon as it's started, and basically leave it there without touching it much. I've occasionally brought it up and hid it again, and as long as I only bring it up without touching anything and hide it, it usually stays fine.
There are many other programs using java out there, I don't know of any that cause this.
It's very unstable that way and not really usable for any extended periods of time.

Out of curiousity how much RAM & how much spare HDD space do you have? May I also ask your mac model?

Whilst I'm at it can you tell me what type of connection you have? Is it a USB or ethernet modem?

Which OSX version do you have?

Do you resume all your files using the Resume button?

This is on a G4 dual 1ghz with 1.25gigs of ram right now, although I only had 768 until somewhat recently (dec-04). I have over 45gigs on free space currently on the boot drive (I have 4 80gig drives in that system, 2 in a mirror).
I'm on cable modem via ethernet.
I'm on the latest osx panther (currently x.3.8), whch I leave on 24hours a day (not for lw).
If I start lw and hide it, not doing any resume, it's usually fine, I can check things out once in a while and hide it again and it's mostly fine that way.
But I have transfers from modems in eastern europe, of huge files, sometimes the connection gets lost and I get the "need new sources" situation, on which lw is too dumb to even try actively to re-establish a connection, so I have to manually resume them. If I only rarely do a resume and one at a time on few transfers, it has a somewhat good chance of not triggering the vm growth, as long as it goes right back into hiding quickly. If I do a little bit too much, such as resume several files, or resume a big selection of files at once, then the growth is triggered, and it continues creeping up until the crash or I restart it before that happens, usually before it gets too close to 2gigs of vm.

With so much spare space I'm a bit surprised that a couple of gig of vm would cause so many problems. Plus the LW versions since about 4.2 I think use less physical ram. Mine's using 138 mb & 1.7 GB vm at present. But I have a lot of incompletes, about 3k shared. Likewise I only use Resume in circumstances where it's been downlding earlier. Though I often try other techniques 1st such as stopping & reselecting from S. results if there was any search.

Do you run as an ultrapeer. Under those circumstances it would be best not to. Disable UP cap. under Speed in LW's prefs.

A couple of other things to reduce VM, set your downlds to Auto-clear & delete search results that are a few hours old.

Delete any old incomplete's that are 0% from both incomplete & downld window.
Perhaps you should reduce no. of shared files also if you have many k's of them.

There was a time recently when I didn't have quite as much free space on the boot drive, there might've been some times when the free space was running a little low and caused extra crashes.
On some occasions, lw had grabbed well over 5gigs of vm. I keep it under 2gigs, I restart it to avoid crashes.
I disabled ultrapeer several weeks ago, which reduces the total vm consumption but doesn't prevent the inevitable. I also turned off all tabs, besides the search one of course.
I always have the auto-clear on and I don't keep on hand any search results.
I don't even run many searches anyway.
Which brings one point that's been bugging me too, there are so many times when it takes a new search to "hit" on a result that will cause an existing transfer to find a source and resume. Many times, the resuming on a transfer doesn't work and it can't find any sources, when the sources are actually there because a search that hits on them will force them found and restart the transfer. That's a pain, lw should more pro-actively seek out the sources instead of sitting there saying "need new sources" and doing nothing about it.
I can't be deleting old incompletes, even if they're several months old. I do keep them in small numbers though, but what I'm getting isn't popular stuff, it's much too rare and only one or sometimes two sources on modems have it, so if I delete them, I won't find them again easily.
I'm not sharing that many files anyway, I'll reduce that number a bit, but I doubt it'll make that much difference, the symptoms remain.