LimeWire 4.11.1 Beta

[Sam]

We've released the 4.11.1 Beta. You can download it from the beta page at http://www.limewire.com/english/content/beta.shtml . PRO users can download it from their personal download page.

Changes include:

- Updated incoming connection handling to not require a new thread for each incoming connection. This will decrease memory and resource requirements for all Ultrapeers and anyone who shares a large number of files.

- Improved firewall-firewall transfer support. These kinds of transfers will use less memory and resources than prior versions.

- Vastly improved download handling. All downloads are now processed on a single thread, which will decrease memory and resource requirements for all users.

- Removed artificial bandwidth limitiation of ~150KB/s per host. LimeWire will now upload and download at the full available capacity.

- Improved temporary buffer storage. LimeWire will make use of preallocated buffers, using much less overall memory.

- Fixed download throttle to apply to all downloaded traffic (including HTTP header traffic).

- Updated DAAP (iTunes Sharing) to be snappier and use less memory.

- Updated DAAP (iTunes Sharing) to use the same I/O subsystem as the rest of LimeWire.

- Updated DAAP (iTunes Sharing) to allow Videos to be shared also.

- Added the ability to append (#) to duplicate downloaded files. Thanks to 'MaTZ' for contributing this!

- Fixed reading saved spam training data so that corrupted data will not crash LimeWire.

- Fixed 'Print All Stack Traces' in the Console to print multiple stack traces from threads with the same name.

- Improved the look & feel of the LimeWire Setup Wizard dialogs.

- Added the ability for LimeWire to automatically add itself to the Windows Firewall Exception list, removing one step from the LimeWire Setup Wizard for Windows XP users (and improving LimeWire's experience for those who set it up incorrectly).

- Removed bugs@limewire.com links and instead directed people to 'Support' on limewire.com.

- Fixed 'No Internet Connection' dialog so it only appears once instead of stacking up multiple times if you're away from the computer.

- Added a new 'Language Button' to the status bar, so users who see LimeWire in a strange language have an easy way to fix it.

- Added the ability to 'Get More Results', which will do another search and not erase results already displayed. Thanks to Philip Schalm for contributing this!

- Fixed 'bitrate' column to sort properly.

- Fixed 'length' column to display time in HH:MM:SS instead of a lot of seconds.

- Added the ability to edit the metadata of many files at once by selecting them all and choosing 'Describe' in the library.

Thanks,
The LimeWire Team

[stief]

Wow--that's a lot of significant changes, and much more in the offing. You guys really have been busy and productive. Thanks again for the changelog, and a special personal thanks to MaTZ--the automatic renaming has made LW much easier to use for my DL habits.

Bug noticed (I think). One LimeWire/4.0.5 leaf tallied up 20 duplicate incoming connections from the same unresolved IP (1 every 5 seconds). I don't want to post the IP and details copied from the Connections Pane, but don't see a quick way to PM it.

Glad too to see the browsing of 'red' hosts fixed in this release: Thanks

[Sam]

You were an Ultrapeer and had 20 leafs from the same IP?

[stief]

Yes. After I removed those connections, it happened again a few minutes later with a different IP (same vendor/version).

[Sam]

Ugh. Okay, That's a bug. We'll take a look.

[Sam]

Do you recall seeing this bug with earlier versions, stief? I can clearly see how this can happen by looking through the code, but for the life of me I can't see anything in earlier versions that prevented it. So it'd be very strange if it just suddenly is appearing now.

[stief]

Hi Sam

For a while I've been seeing repeated connections from MORPH or LW 4.0.5's, but this was the first time I noticed 20 connections from the same IP.

The MORPH's have been a problem since last summer at least, because I associate them with high spam search results and aggressive connection attempts, so I'm in the habit of clearing them before doing my usual "What's New" image searches. It can take up to 20 minutes to clear them long enough to get in a decent image search.

In case this is related, I've emailed you a screen shot that shows that the MORPH's are able somehow to exceed the 32 Peer connections (see the connected time column).

[stief]

Just a comparison: Captured multiple 4.0.5's connecting from identical IP's again with a build of a fresh checkout of the SVN code (May 26 23:11:13 2006 Checked out revision 20868).

13 connections of a 4.0.5 leaf from one IP; 8 from another.

[Lord of the Rings]

Sorry just remarking a remarkable no. of improvements to view. I'm just about to downld the beta so please excuse my pointless posting other to reiterate Stief's 1st post about a lot of work has obviously gone into these improvements. I was slow to catch onto the new beta's release. Can't wait to try it & check whether it solves my issues.

Edit: Nice to see you converted the splash from 4.10 to 4.11 :D I'm starting to wonder if running Classic in the background is not such a good idea whilst running LW on a mac. I tend to get temporary faded out areas as the interface is opening. A little similar to when some warning or choice window pops up & the options are not all visible. I'll need to see if that still happens also.

1354 shares. 2+ k incompletes but not all of these are loaded.
2 mins after running I'm using 33-34 threads ... 216+ MB ram, 620+ MB VM. Both the latter seem to be increasing quickly.
278 & 678 after about 9 mins ... still settling down .. no actions.

Seems to have levelled off at 281 & 679 after about 11 mins or so.

The Append # is a great idea!!! 30, 301, 691 after one search & downlding one file at 10 KB/s after 55 mins (I got distracted.) Interesting to see the threads dropped off even during the search.

[Sam]

>The Append # is a great idea!!!

stief posted one day in the gnutellaforums that he wanted to see a bounty for it, so I added a bounty and the next day MaTZ asked if he could do it. Funny how things work out. :)

[Lord of the Rings]

Great result for #. Good to hear.

Back to tracking ... early times ... perhaps too early since I haven't worked LW much over this time. The past 3 1/2 hrs now 37 threads, 340 MB ram & 722 VM. But only did some Find Sources for files which had some previous % progress, since earlier. Bit by bit of course. I'll see how it shapes up after leaving on for some time & then reinitiating searches. There was usually a breakdown point for me (& some others) on OSX so I'll report back later.

[Lord of the Rings]

It finished up with 40 threads, 438 MB ram, 876 MB VM. It did get up to 48 threads but that was whilst I was downlding a few files. It seemed to jump back when these finished. There still seems to be that ever-increasing VM but generally it was much more stable than previous versions including 4.11.0 In fact it was around 876 mb vm for quite some time.

One bug I did notice however is the one I noticed with 4.11.0 which was grouping of results. Some results would be same size but be separate listings. And some would be ticked as having downlded when I hadn't downlded those particular sizes. Including separate listings for those that I had so ... it looked as though I had downlded perhaps 15-25 files of the one title of file.

A couple of error messages that were blank. I think the 1st was a preferences setting, & the other I think was a file of the same name as sample below shows. BTW I have a handful of bug reports via the console if you're interested which were at times of downldling, browsing, static downlds at various periods. Let me know if you want to see them. (also snapshots) Oops I forgot to mention ... OSX 10.4.6 Java 1.5 latest update ... 1.5 GB ram.