3333.33 mg Taurine Per Dollar, And, Milk, The Bittorrent Client: Part IV

You can't read the label, because my hands are shaking like I've just consumed 2000 mg of Taurine. An entire pot of coffee doesn't get me like these damn things do.

In other news,

IT'S FUCKING DONE!

That's right folks, Milk works. Download it here. A single purpose free-leeching, sequentially downloading, super lightweight torrent client. You can get the source now, or I'll probably try to make binaries (because I know SOOO many of you care). OMG DON'T U GUISE LOVE JAVA TOO!?

what it does do

• downloads at over 1MBps (that's 8Mbps, folks, just to rub it in your face) while using < 10% of processor time on my Macbook Pro 2.2 GHz
• It now has a (more) properly abstracted structure. file and piece management is no longer tied to the peer. There are convenience classes for Blocks, BlockMaps, Pieces, and such things to make the code easier to understand and more portable.
• uses 16 MB RAM and 11 threads (on my MacBook Pro) of JVM overhead
• when given the name of a torrent file on the local system, talks to the tracker, gets a peers list, connects to as many peers as possible (won't connect to peers that force encryption .... yet)
• for better or worse, it doesn't contain any GPL code
• allows for as many parallel downloads as you want, probably limited by machine memory
• downloads multifile or singlefile torrents (joy)
• enables you to download legal content sequentially
• enables you to download illegal content sequentially (don't do this, it's WRONG)
• stores the tracker response to a file, and won't pester the tracker until 30 minutes have passed for more peers (unless you want to)

What it doesn't do (and possibly should):

• pester the client for lots of peers automatgically
• announce to the tracker after it's done download that, well, it's done downloading
• it doesn't have a GUI
• use more than 20 MB RAM, typically (memory usage should be close to (piece size * number of peers) or about 25 MB plus JVM overhead worst case
• keep any state beyond the last tracker update between runs

And for those who say that sequential downloading is bad for the swarm, yes, it IS bad for the swarm. : D Do not use this with private trackers, they will kick you the fuck off.

In traditional blog style, I'm going to post some links. I wish I was in Chicago. Wait, I've never even played bike polo.

I just recently found the world of bicycling blogs. Each time I add a new one to google reader, another 5 come up that I have to read. It's painful, really. There's Bike Hugga. I don't really know who they are or what they do (I just look at the pretty pictures), but they have an incredible amount of money, judging by the bikes they ride, and the fact that some of their writers seem to star in cycling photo shoots. Or maybe the money comes from the three square miles of ads on their site. Anyways, they post really cool stuff regarding pretty much everything important in cycling. Even if it does involve carbon fiber.

Then there's cyclingwmd. This guy is famous for his sentence to exclamation mark ratio. He posts "crazy bike shit that will blow your mind and leave you begging for more". It's fantastic.

This guy is my hero. He posts the best of the best of craigslist NYC, takes pictures of trucks in bike lanes, hating (snobbing) on triathletes, and just plain wacky stuff.

Anyways, just robert karl checking in. Add those cycling blogs to your google reader and I assume you will get the other 50 really neat cycling blogs that it suggested I read. Adiós...