[00:21:49] *** bbelt16ag has quit IRC
[00:21:53] *** bbelt16ag has joined #bittorrent
[00:28:49] *** Elrohir has joined #bittorrent
[00:30:56] *** Elrohir has quit IRC
[01:53:38] <TheSHAD0W> http://yro.slashdot.org/article.pl?sid=09/11/08/0010243
[02:30:26] <The_8472> that's why they need ACTA. the WIPO is getting too soft!
[03:29:49] *** init0_ has joined #bittorrent
[03:41:47] *** init0 has quit IRC
[03:57:11] *** The_8472 has quit IRC
[03:57:31] *** wadim has joined #bittorrent
[03:57:33] *** wadim is now known as The_8472
[04:18:45] *** Miller` has quit IRC
[04:39:14] *** inamru has quit IRC
[04:43:09] *** chelz has joined #bittorrent
[05:16:32] *** The_8472 has quit IRC
[05:22:29] *** bt42 has joined #bittorrent
[05:24:43] *** rrr_ has joined #bittorrent
[05:44:16] *** bittwist has quit IRC
[05:44:58] *** bt42 is now known as bittwist
[07:00:53] *** MassaRoddel has quit IRC
[07:19:52] *** K`Tetch has quit IRC
[07:45:25] <sparr> I can download using some trackers, but not using some others.  Is there a client that will help me diagnose that problem?
[07:47:45] *** Miller` has joined #bittorrent
[07:48:12] *** Miller` has quit IRC
[07:48:18] *** Miller` has joined #bittorrent
[08:23:00] *** hydri has quit IRC
[08:24:18] *** hydri has joined #bittorrent
[08:54:45] *** MsDinkAlot has joined #bittorrent
[08:55:28] *** MsDinkAlot has left #bittorrent
[09:18:49] *** burris has quit IRC
[09:18:50] *** stalled has quit IRC
[09:19:08] *** pajlada has quit IRC
[09:19:09] *** jnpplf has quit IRC
[09:19:09] *** mpl has quit IRC
[09:19:09] *** void^ has quit IRC
[09:19:09] *** HandheldPenguin` has quit IRC
[09:19:09] *** uau has quit IRC
[09:19:10] *** mxs has quit IRC
[09:19:10] *** charles has quit IRC
[09:19:11] *** tris has quit IRC
[09:19:40] *** mxs has joined #bittorrent
[09:19:40] *** uau has joined #bittorrent
[09:19:40] *** burris has joined #bittorrent
[09:19:40] *** pajlada has joined #bittorrent
[09:19:40] *** stalled has joined #bittorrent
[09:19:40] *** jnpplf has joined #bittorrent
[09:19:40] *** void^ has joined #bittorrent
[09:19:40] *** mpl has joined #bittorrent
[09:19:40] *** HandheldPenguin` has joined #bittorrent
[09:19:40] *** charles has joined #bittorrent
[09:19:40] *** tris has joined #bittorrent
[09:25:40] *** stalled has quit IRC
[09:32:43] *** stalled has joined #bittorrent
[09:34:44] *** MassaRoddel has joined #bittorrent
[09:38:22] *** burris has quit IRC
[09:38:39] *** stalled has quit IRC
[09:38:40] *** mpl has quit IRC
[09:38:40] *** pajlada has quit IRC
[09:38:40] *** jnpplf has quit IRC
[09:38:41] *** void^ has quit IRC
[09:38:42] *** HandheldPenguin` has quit IRC
[09:38:42] *** uau has quit IRC
[09:38:42] *** mxs has quit IRC
[09:38:44] *** tris has quit IRC
[09:38:44] *** charles has quit IRC
[10:08:52] *** thermal has joined #bittorrent
[10:21:37] *** therma has quit IRC
[10:27:43] *** jnpplf has joined #bittorrent
[10:27:43] *** void^ has joined #bittorrent
[10:27:43] *** mpl has joined #bittorrent
[10:27:43] *** HandheldPenguin` has joined #bittorrent
[10:27:43] *** pajlada has joined #bittorrent
[10:27:43] *** stalled has joined #bittorrent
[10:27:43] *** tris has joined #bittorrent
[10:27:43] *** charles has joined #bittorrent
[10:27:43] *** mxs has joined #bittorrent
[10:27:43] *** uau has joined #bittorrent
[10:27:43] *** burris has joined #bittorrent
[10:27:43] *** L337hium has joined #bittorrent
[10:32:52] *** stalled has quit IRC
[10:35:41] *** chelz has quit IRC
[10:39:17] *** stalled has joined #bittorrent
[12:56:52] *** thermal has quit IRC
[13:00:58] *** waldorf_ has joined #bittorrent
[13:28:49] *** Leoneof` has joined #bittorrent
[13:35:21] <Leoneof`> hi
[13:43:51] <Leoneof`> is there are website can convert torrent link to http link?
[13:44:22] <hlindhe> torrent link?
[13:45:14] <Leoneof`> yes, like *.torrent
[13:48:38] <alus> Leoneof`: no
[13:49:00] <alus> Leoneof`: you need a BitTorrent client to download the file which the .torrent refers to
[13:49:15] <Leoneof`> how about furk.net? never used though
[13:49:18] <Leoneof`> o.o
[13:50:28] *** medecau has joined #bittorrent
[13:56:10] <DWKnight> http://btfaq.com
[13:58:24] <Leoneof`> the reason is that bittorrent and utorrent is not able to download , how to fix?
[13:59:57] <DWKnight> who is your internet provider?
[14:00:50] <alus> also maybe the torrent is dead
[14:01:21] <Leoneof`> don't know >_>
[14:01:37] <DWKnight> http://distribution.openoffice.org/p2p/ <-- these torrents aren't dead
[14:01:37] <Leoneof`> alus, it is ubuntu iso
[14:01:52] <DWKnight> you don't know who your internet provider is?
[14:02:38] <Leoneof`> actually i don't know
[14:05:00] <mpl> are you on a campus network or something like that?
[14:05:14] <Leoneof`> at home >_>
[14:05:57] <mpl> so who contracted with the isp so you have internet at home if it's not you?
[14:06:21] <mpl> that person must know who your isp is.
[14:07:45] *** thermal has joined #bittorrent
[14:09:11] *** bbelt16ag1 has joined #bittorrent
[14:09:50] *** bbelt16ag has quit IRC
[14:17:27] *** The_8472 has joined #bittorrent
[14:17:42] *** _rafi_ has joined #bittorrent
[14:32:17] *** Leoneof` has quit IRC
[15:38:06] *** Andrius has joined #bittorrent
[15:55:47] *** medecau has quit IRC
[16:06:21] *** medecau has joined #bittorrent
[16:06:35] *** medecau has quit IRC
[16:31:16] *** Snoopotic has quit IRC
[16:58:06] *** L337hium has quit IRC
[17:02:18] *** therma has joined #bittorrent
[17:02:25] *** thermal has quit IRC
[17:04:45] *** K`Tetch has joined #bittorrent
[17:42:35] *** klapaucjusz has joined #bittorrent
[17:44:10] <klapaucjusz> Oy!
[17:44:30] 
[17:46:57] *** therma has quit IRC
[17:47:20] <The_8472> btw, i have an implementation of BEP 32 running
[17:50:44] <klapaucjusz> Cool, code available?
[17:51:00] <klapaucjusz> Have you put it up on a public address?
[17:52:46] <The_8472> code exists under SVN, it's a plugin for Az
[17:52:58] <The_8472> though i could build a .jar
[17:53:07] <The_8472> not that it does any good, without nodes to test against
[17:53:20] <klapaucjusz> The_8472: SVN is fine.  Pointer?
[17:53:49] <The_8472> sec
[17:54:08] <The_8472> https://azsmrc.svn.sourceforge.net/svnroot/azsmrc/azsmrcplugins/trunk/lbms/plugins/mldht
[17:54:27] * klapaucjusz refrains from ranting about people still using SVN, rather than Darcs or Git.
[17:56:48] <DWKnight> at least it's not CVS anymore
[18:02:34] <klapaucjusz> The_8472: I love your commit messages.  "should be mostly exception-free"
[18:03:03] <DWKnight> you can tell a lot about the mindset of a programmer by their code comments and revision commit messages
[18:04:05] * klapaucjusz realises he doesn't know Java.  What's enum?
[18:04:27] <The_8472> a special class that defines a list of constants.
[18:04:40] <The_8472> not too different from enums in C++
[18:04:56] <The_8472> except that the enumes aren't ints but singleton-objects
[18:05:11] <The_8472> though they can be serialized into ints if necessary
[18:06:40] <The_8472> DWKnight, never let me program realtime embedded systems. planes would probably fall out of the sky and i would say "give me the debug logs, let's see if we can fix this"
[18:06:57] <klapaucjusz> Ah, so you instantiate DHT twice.
[18:07:04] <The_8472> yes
[18:07:23] <klapaucjusz> Where do you dispatch a received packet?
[18:07:43] <The_8472> lbms.plugins.mldht.kad.RPCServer.run()
[18:08:05] *** Elrohir has joined #bittorrent
[18:09:29] <DWKnight> The_8472: I'd be doing manual backtraces to see what went wrong
[18:09:47] <The_8472> "does that plane support step debugging?"
[18:10:03] <klapaucjusz> Binding: have a look at the elegant hack I'm using in http://trac.transmissionbt.com/attachment/ticket/2502/self-ipv6-address.patch .  It allows you to use the kernel's/libc's address selection policies without having to scan for addresses yourself.
[18:10:34] <The_8472> nice. but can't do that in java
[18:12:19] <klapaucjusz> Why not?
[18:12:29] <The_8472> doesn't expose that API
[18:13:02] <The_8472> ahh, wait. you're doing it based on the source address?
[18:13:09] <The_8472> hrrrm
[18:13:20] <klapaucjusz> Socket.getLocalAddress
[18:13:28] <The_8472> yeah, that exists
[18:13:57] <The_8472> but then prefix policies come into play, like it might pick teredo over something else if a hostname happens to resolve to a teredo address
[18:14:15] <klapaucjusz> The idea is that you create a socket, get the kernel to run its address selection on it, grab the local address, then destroy the socket before you send any packets.
[18:14:33] <klapaucjusz> Yep, that's why you must make sure that you do that against a native address.
[18:14:49] <The_8472> nice idea
[18:15:05] <klapaucjusz> You owe me a beer ;-)
[18:15:26] <klapaucjusz> (And an acknowledgment in the source.)
[18:15:29] * The_8472 DCCs klapaucjusz some beer images
[18:16:27] <The_8472> just binding to any native address, even if it doesn't exist should work, no?
[18:16:38] <The_8472> s/binding/connecting/
[18:16:49] <klapaucjusz> Probably.
[18:16:53] <klapaucjusz> What's call_queue?
[18:17:29] <The_8472> a queue for outgoing requests in case the message IDs are exhausted
[18:17:36] <The_8472> though i think that shouldn't ever happen
[18:17:52] <The_8472> since there are less concurrently allowed tasks than available message IDs
[18:18:14] <klapaucjusz> Eek.
[18:18:23] <klapaucjusz> I'm using 32 bits of message id.
[18:18:29] <The_8472> we're using 16 bits
[18:18:48] <The_8472> 65k concurrent outstanding requests... shouldn't happen
[18:18:59] <klapaucjusz> That's still tens of thousands.  I'd simply drop the message in that case, worse is better.
[18:19:20] <The_8472> those are outgoing requests.
[18:19:25] <The_8472> incoming ones are stateless
[18:19:47] <The_8472> except for the tokens for announces
[18:19:53] * klapaucjusz wishes Java programmers would stop splitting their code into zillions of tiny files.
[18:19:55] <The_8472> but their state is maintained somewhere else
[18:20:24] * The_8472 wishes C(++) programmers would get multi-pass compilers and do away with those silly header files
[18:20:57] * klapaucjusz wishes everyone would just program in Lisp.
[18:21:12] <The_8472> and we're using IDEs, they jump automatically between files. F3 ... open declaration, regardless where it is located
[18:21:54] * klapaucjusz recalls a time when he had to walk eight miles in order to get a text-only terminal.  Barefoot.  In the snow.  Uphill in both directions.
[18:22:19] <The_8472> yeah, these days we get actual work done
[18:22:43] <The_8472> it might cost ten times the memory and twenty times the CPU time. but 1/100th of programmer time
[18:28:03] <The_8472> thread-safe, lockfree collection classes are extremely useful in java for example
[18:29:34] <klapaucjusz> Eek... you're parsing message types non-case-sensitively?  I parse them bytewise.  Any trouble with that?
[18:29:48] <The_8472> hrm?
[18:30:08] <The_8472> some .equalsIgnoreCase somewhere?
[18:30:13] <klapaucjusz> MessageDecoder uses equalsIgnoreCase.  I use memcmp.
[18:30:23] <klapaucjusz> Will I run into trouble, or does everyone use lowercase?
[18:30:36] <The_8472> i haven't seen any problems, no
[18:30:45] <The_8472> it's just one of those "be liberal what you accept" things
[18:30:53] <klapaucjusz> Ack.
[18:31:00] <The_8472> though i'm personally more a fan of fail-fast
[18:31:16] <The_8472> things exploding in a dev's face make him less sloppy
[18:31:42] <klapaucjusz> Where's decodeWant?
[18:32:01] <klapaucjusz> (Got it.)
[18:32:02] <The_8472> lbms.plugins.mldht.kad.messages.AbstractLookupRequest.decodeWant(List<byte[]>)
[18:32:31] <The_8472> and yes, i'm goddamn lazy
[18:32:44] <klapaucjusz> Hmm, you're gonna break when somebody puts a non-string into want.
[18:33:00] <The_8472> the spec says it's a list of strings
[18:33:10] <klapaucjusz> Be liberal...
[18:33:18] <The_8472> they won't get a response if they do that
[18:33:24] <The_8472>  <The_8472> though i'm personally more a fan of fail-fast <- ^^
[18:33:36] <The_8472> but ok
[18:36:53] <klapaucjusz> where do you add nodes to your routing table?
[18:37:11] <The_8472> ughh... that's a non-trivial process
[18:38:01] <The_8472> basically any method calling lbms.plugins.mldht.kad.KBucket.removeAndInsert(KBucketEntry, KBucketEntry)
[18:38:09] <The_8472> but there are several call points for that
[18:39:18] <The_8472> guess this one is the most important one: lbms.plugins.mldht.kad.Node.recieved(DHTBase, MessageBase)
[18:51:55] *** Elrohir has quit IRC
[18:53:45] *** bittwist has quit IRC
[19:04:28] *** L337hium has joined #bittorrent
[19:26:29] <TheSHAD0W> Um.
[19:31:41] <TheSHAD0W> Heh.
[19:32:58] <The_8472> ?
[19:33:29] <TheSHAD0W> Nothing, just thought I'd seen something strange in utorrent's operation.
[19:34:48] * klapaucjusz is running Transmission over TCP-LP.  Up: 4.8 KB/s.
[19:35:04] <klapaucjusz> That's definitely less than best-effort.
[19:35:11] <The_8472> congested line?
[19:36:05] <The_8472> because a less-than-best-effort controller still should utilize the line if there is no best-effort traffic competing with it
[19:36:43] <klapaucjusz> Looks like congestion at the receiving end -- I'm getting large throughput if I try seeding a different torrent.  (I've got just one leecher.)
[19:36:56] <The_8472> ah
[19:37:19] <klapaucjusz> s/large/reasonable/
[19:37:30] <The_8472> you can pick the congestion controller via socket opts under linux?
[19:37:59] <klapaucjusz> Yep. http://trac.transmissionbt.com/attachment/ticket/2338/congestion-control.patch
[19:38:29] <The_8472> nice
[19:39:38] <klapaucjusz>   # modprobe tcp_lp
[19:39:38] <klapaucjusz>   # echo reno cubic lp > /proc/sys/net/ipv4/tcp_allowed_congestion_control
[19:39:54] <klapaucjusz> ...or it won't work :-(
[19:40:22] <The_8472> well, i doubt anything like that exists on windows...
[19:41:00] <klapaucjusz> They do have switchable controllers -- you can choose NewReno or CTCP using a registry tweak.
[19:41:22] <klapaucjusz> So adding the right setsockopt should be trivial if somebody asks Microsoft nicely.
[19:41:43] <klapaucjusz> They've been doing some remarkable work with their networking stack (which was crap before Windows 2003 or so).
[19:42:03] <The_8472> i know about CTCP, but i mean them being pluggable and choosable at the application level
[19:42:15] <klapaucjusz> But of course it's more sexy to add the code in user space, and get a lot of press coverage in the process.
[19:42:50] 
[19:44:58] 
[19:46:13] <klapaucjusz> I'm sure they could actually get Ledbat included in the stock kernel.
[19:48:00] *** Elrohir has joined #bittorrent
[19:48:25] <The_8472> i suggested the same...
[19:48:29] <The_8472> but, nobody listens
[19:51:10] <klapaucjusz> Heck, you're the one who pointed that out to me initially.
[19:51:40] <klapaucjusz> Unfortunately, you didn't want to start an argument, so you left it to me to speak up publically and get flamed to a crisp ;-)
[19:52:56] <The_8472> actually, we did argue about it. here on this channel
[19:53:18] <The_8472> everyone went "uhh. internet standards bodies move too slow. adopting it into kernels would take years" etc. etc.
[19:53:19] * klapaucjusz is wondering when Poland joined the Berne convention, and whether his mass-testing of Transmission and Hekate with Polish movies from the 1950s is legal.
[19:54:58] <The_8472> i doubt anyone will care
[19:55:31] <klapaucjusz> The_8472: conversations on IRC are not quotable -- you cannot easily publish a link to an IRC chat.  If you want people to listen, you need to go through a public forum.
[19:55:53] <The_8472> actually, i think there's a bot logging and publishing this channel
[19:55:53] <klapaucjusz> They're good movies, though.
[19:56:13] * The_8472 points at echelog
[19:56:46] <The_8472> an apt name i must say
[20:13:37] <K`Tetch> I log this channel
[20:14:01] <K`Tetch> so much good damned info, i don't want to risk missing it
[20:14:09] *** linux_jon has joined #bittorrent
[20:14:46] <linux_jon> is the some way I could command ktorrent from a shell script to search and download a file name?
[20:15:05] <The_8472> no, that's not how bittorrent works
[20:15:38] <klapaucjusz> linux_jon: BitTorrent is a file transfer protocol.
[20:15:47] <klapaucjusz> Searching is outside the scope of the torrent.
[20:16:06] <linux_jon> ok how would I search for a file and then download it?
[20:16:07] <klapaucjusz> Er., protocol.
[20:16:29] <linux_jon> I understand what a protocol is and I know bittorrent is this.
[20:16:31] <TheSHAD0W> http://btfaq.com
[20:16:43] <klapaucjusz> Query your favourite BT search engine, download the .torrent, then pass it to your client.
[20:17:00] <klapaucjusz> Alternatively, find the torrent's info-hash and pass the magnet link to your client.
[20:17:11] <linux_jon> I was wondering how I could execute a automated script that would search for a file name that is a in a list in a text file. Then download the correct one.
[20:18:34] <klapaucjusz> http://isohunt.com/forum/viewtopic.php?p=433527#433527
[20:20:40] *** fgader has joined #bittorrent
[20:23:15] <klapaucjusz> Heh.
[20:23:16] <klapaucjusz> draft-welzl-ledbat-survey-00
[20:23:27] <klapaucjusz> " Intention: Avoid reinventing the wheel"
[20:23:34] <The_8472> rofl
[20:23:58] * TheSHAD0W scratches his head
[20:24:01] <The_8472> well, ledbat yes...
[20:24:04] <TheSHAD0W> What's wrong with reinventing the wheel?
[20:24:05] 
[20:24:22] <TheSHAD0W> If we couldn't reinvent the wheel, we'd still be driving around on wooden wagon wheels.
[20:24:27] <The_8472> TheSHAD0W, if you don't reuse you're incompatible. that's usually the problem
[20:24:27] <klapaucjusz> It's great if you reinvent it right.
[20:24:37] <klapaucjusz> Somewhat less so if you reinvent it square.
[20:25:06] <TheSHAD0W> Wooden wagon wheels are incompatible with modern cars.  Is that wrong?
[20:25:39] <The_8472> no, but let's say continental invents a super awesome new wheel that saves 30% fuel. but it's incomatible with all existing cars
[20:25:58] <The_8472> so you'd have to swap out all cars to deploy it
[20:26:04] <klapaucjusz> Your car analogy is flawed.
[20:26:13] <The_8472> couldn't come up with a better one
[20:26:27] <TheSHAD0W> All analogies are flawed on some level or other, otherwise they wouldn't be analogies.
[20:26:39] <klapaucjusz> Support GM invents a new wheel that's 30% *less* efficient than current wheels, is incompatible with all current cars, and causes other cars to crash if the road is shared.
[20:26:53] <TheSHAD0W> It's been done.
[20:26:55] <TheSHAD0W> :-P
[20:26:58] <klapaucjusz> Er, I meant "legacy cars".
[20:27:02] <The_8472> now you're overstating things ^^
[20:27:07] <klapaucjusz> Sue me.
[20:27:13] <The_8472> ledbat itself is a good idea.
[20:27:20] 
[20:27:37] <klapaucjusz> Agreed.
[20:27:47] <klapaucjusz> That's not a reason for not overstating things :-P
[20:28:11] <The_8472> well, the 30% fuel efficiency was refering to the benefits of ledbat
[20:28:12] * klapaucjusz needs to port Stochastic Fair Blue to OpenWRT.
[20:28:30] <klapaucjusz> The_8472: yes, but.
[20:28:32] <The_8472> RED already exists for TC
[20:28:44] <The_8472> you can probably adapt that
[20:29:11] <klapaucjusz> (I'm already using RED on my router.  BLUE is the new RED, and SFB rocks.)
[20:29:35] <The_8472> with ECN or drop?
[20:29:52] <klapaucjusz> SFB is able to use different mark/drop rates for different flows, detect inelastic flows and rate-limit them.
[20:29:58] *** bittwist has joined #bittorrent
[20:30:18] <klapaucjusz> ECN, of course, although ECN must be supported on both endpoints to be effective.
[20:30:38] <klapaucjusz> Coming back to Ledbat... I'll read a couple of papers before expressing an opinion, if that's all right with you.
[20:31:03] <klapaucjusz> You'll find a link to the SFB paper on http://www.pps.jussieu.fr/~jch/software/sfb/
[20:31:20] <The_8472> since you linked to TCP-LP... what are the differences to ledbat?
[20:32:01] <klapaucjusz> I have no idea. There's also TCP-Nice.
[20:32:31] * klapaucjusz is reading.
[20:33:15] <The_8472> well, the key goal of ledbat is to keep queues empty... or at least at ~100ms delay above the baseline
[20:33:44] <The_8472> other less-than-best-effort controllers may work with differnet goals
[20:34:07] <The_8472> just yielding to reno doesn't make any latency guarantees for example
[20:34:32] * TheSHAD0W has never used OpenWRT
[20:35:46] <TheSHAD0W> dd-wrt is too easy and convenient.
[20:36:33] <bittwist> you devious hackers with your custom firmware
[20:36:36] <The_8472> dd-wrt is hell if you try to get 6to4 on a dynamic v4 ip to work
[20:36:43] <bittwist> you would use vanilla if you had nothing to hide!
[20:36:44] * TheSHAD0W shrugs
[20:36:48] *** fgader has left #bittorrent
[20:36:54] <klapaucjusz> bittwist: try it, you'll be convinced.
[20:37:04] <bittwist> klapaucjusz: i was being a bit sarcastic
[20:37:13] <klapaucjusz> And I'm dead serious.
[20:37:20] <bittwist> in light of the heat brought on the custom cable modem firmware guy
[20:37:20] <TheSHAD0W> That'd be a great project; making OpenWRT with a better install interface.
[20:37:22] <bittwist> by the feds
[20:37:36] <klapaucjusz> Heh.
[20:37:49] <TheSHAD0W> Maybe having it mount a directory on a local machine, to use for workspace.
[20:37:57] <bittwist> http://news.google.com/news/more?pz=1&cf=all&ned=us&cf=all&ncl=dAhHYmHILX8r3yMg-L0iBkJHJ1yfM
[20:38:13] <bittwist> Feds Charge Cable Modem Modder With 'Aiding Computer Intrusion'
[20:38:18] <klapaucjusz> TheSHAD0W: nah.  That's not the way towards world domination.
[20:38:37] <klapaucjusz> Sell a router with custom firmware and easily installable new packages.  That's fon's new business model.
[20:38:39] <TheSHAD0W> If I were interested in world domination, I'd probably have done it already.
[20:38:55] <bittwist> klapaucjusz: and get busted by the feds later :P
[20:39:03] <bittwist> poor tcsino
[20:39:19] <klapaucjusz> bittwist: no, because you're only selling a wifi router, no cable modem.
[20:39:37] <bittwist> TheSHAD0W: there, i contributed a story to the chan :D
[20:39:55] <bittwist> your not the only one that pulls link weight hehe
[20:40:24] <bittwist> klapaucjusz: remember when several US ISPs said hooking up routers to your modems to split internet between pc's was breaking TOS
[20:40:33] <bittwist> and considered abusive behaviour?
[20:40:36] <bittwist> :S
[20:40:45] <The_8472> "An Oregon man faces up to to 20 years in prison for allegedly selling modding tools that allowed his customers to swipe high-speed internet access without paying."
[20:40:47] <The_8472> lolwut?
[20:40:48] <bittwist> att was like that for a moment
[20:40:58] <The_8472> build infrastructure that can only be used when you're authenticated
[20:41:05] <The_8472> ... their own fault if they don't secure their systems
[20:41:11] <klapaucjusz> Not really.
[20:41:20] <klapaucjusz> My pockets are not secured.
[20:41:29] <klapaucjusz> This doesn't mean you're welcome to scan them when I take the metro.
[20:41:30] <The_8472> your pockets aren't infrastructure
[20:41:37] <bittwist> klapaucjusz: well comcast has a shitty infrastructure on purpose
[20:41:56] <bittwist> nearly every cable system in western europe is modern
[20:41:59] <klapaucjusz> Okay, think about it this way.
[20:42:02] <bittwist> comcast stayed behind because it was cheap
[20:42:12] <bittwist> because it was Comcastic
[20:42:15] <The_8472> compare it to those bag tresor things at the metro where you can store your bags. then think of building them withoug locks
[20:42:22] <klapaucjusz> The electricity network is not secure, the electricity counter in my flat ensures I pay for the electricity I draw.
[20:42:29] <klapaucjusz> Should hacking electricity counters be legal?
[20:42:39] <bittwist> klapaucjusz: uh yea
[20:42:45] <TheSHAD0W> "Meters"?
[20:42:46] <bittwist> using the device to avoid payment is wrong
[20:42:48] <bittwist> mkay
[20:43:00] <The_8472> the difference is that syphoning off electricity hard to defend against as you can do it anywhere
[20:43:03] <bittwist> but thats not what he was arrested for
[20:43:08] <linux_jon> fuckin with a electric meter is kinda dangerous.. hah
[20:43:13] <The_8472> cryptographically securing a network on the other hand is easy
[20:43:18] <The_8472> no service unless you have a key
[20:43:44] <The_8472> it's like sueing someone for providing software that lets you access open wifi networks oO
[20:43:48] <linux_jon> could you in some way use the network still, without a key?
[20:43:59] <linux_jon> communicate with those who dont have a key as well?
[20:44:12] <The_8472> you could get cable tv maybe
[20:44:16] <klapaucjusz> linux_jon: I hope someday you'll join us.
[20:44:17] <TheSHAD0W> Qwest uses PPPoE with their network.
[20:44:23] *** Elrohir has quit IRC
[20:44:28] <klapaucjusz> http://wifi.pps.jussieu.fr/
[20:44:39] <linux_jon> Thats what we have here, fibre optic lines with ppoe
[20:44:44] <linux_jon> buyncha BS i think
[20:44:46] * TheSHAD0W had a hell of a time with Qwest's internet
[20:45:05] <TheSHAD0W> They gave me a modem with a built-in and really crappy router.
[20:45:06] <linux_jon> yea, I get my sever booted every so often, and wuite alot sometimes
[20:45:19] <linux_jon> yea, same here... shitty router....
[20:45:45] <TheSHAD0W> I wound up having to disable the modem's router, and then program my own router to log in with PPPoE through the disabled modem.
[20:46:53] <linux_jon> hmm, how did that work out?
[20:47:06] <linux_jon> Were you having trouble loosin your connection?
[20:47:16] <TheSHAD0W> No, it's pretty stable.
[20:48:20] <linux_jon> I keep loosing my PPoE connection, could this be a user side problem?
[20:48:38] <TheSHAD0W> It's possible.
[20:48:39] <linux_jon> I beleive it to be a ISP problem.,...
[20:48:50] <linux_jon> hmm
[20:48:53] <TheSHAD0W> Doing what I did might be a good way to find out.
[20:49:11] <The_8472> not sure, but my modem regularly hangs up if i flood its conntrack table
[20:49:19] <TheSHAD0W> Hehe.
[20:49:51] <The_8472> 5 DHTs, 6to4, teredo and regular TCP connections... so yeah
[20:49:53] <linux_jon> what did you do shaddow?
[20:50:03] <linux_jon> disabled the modem and router and then what?
[20:50:29] <TheSHAD0W> First, I copied down the PPPoE login data that was in the modem.
[20:51:01] <klapaucjusz> Okay, still looking at Ledbat and TCP-LP.
[20:51:04] <TheSHAD0W> Then, there were options in the router configuration that shut down routing and disabled PPPoE login.
[20:51:15] <klapaucjusz> sorry, forget it.
[20:51:32] <TheSHAD0W> Then I put the login data in my WRT54GL.
[20:52:14] * klapaucjusz has given up on the WRT54GL.  He's very happy with the older models by Accton.
[20:52:22] <klapaucjusz> (Rebranded by Fon.)
[20:52:42] <The_8472> GL works fine for me
[20:53:06] <TheSHAD0W> Same here.
[20:53:13] *** Andrius has quit IRC
[20:53:14] <klapaucjusz> The Accton/Fon models are tiny, have more memory, are cheaper, and use Atheros for the Wifi.
[20:53:25] <linux_jon> so you just used the modem as a hum to the network
[20:53:31] <klapaucjusz> Slow flash, though.
[20:53:48] <TheSHAD0W> Right.
[20:55:05] <TheSHAD0W> klapaucjusz: This thing?
[20:55:06] <TheSHAD0W> https://shop.fon.com/FonShop/shop/US/ShopController?view=product&product=PRD-001
[20:56:41] <klapaucjusz> No, that's the new model, and it's not by Accton.  Unsupported with OpenWRT.
[20:57:00] <TheSHAD0W> https://shop.fon.com/FonShop/shop/US/ShopController?view=product&product=PRD-022 - this one then?
[20:57:05] <klapaucjusz> There's the old 2100, which is lovely if you don't need Ethernet.
[20:57:39] <TheSHAD0W> Doesn't look like it's still available.
[20:58:10] <klapaucjusz> And there's the 2200 and 2201 which have two Ethernets instead of just one.  The latter has USB.
[20:58:30] <klapaucjusz> https://shop.fon.com/FonShop/shop/FR/ShopController?product=PRD-018&view=product&partner=home2plus
[20:58:39] <klapaucjusz> Unfortunately, the 2201 is no longer available.
[20:59:26] <klapaucjusz> (I'm speaking to you through a 2201.)
[20:59:48] <TheSHAD0W> (...And my GL.)
[20:59:49] <TheSHAD0W> ;-)
[20:59:55] <klapaucjusz> Okay, differences between TCP-LP and Ledbat.
[21:00:44] <klapaucjusz> TCP-LP drops down to cwnd=1 when congestion persists over two RTTs, Ledbat doesn't.
[21:01:53] <klapaucjusz> The equation that governs cwnd increase/decrease is not the same, it looks prettier in Ledbat.  I'm not competent to determine the difference.
[21:02:06] <klapaucjusz> Other than that, they're the same protocol, I think.
[21:02:34] 
[21:03:21] <linux_jon> is there a way I can scan for address on my local network starting with 10.*.*.*
[21:03:23] <linux_jon> ??
[21:03:45] <TheSHAD0W> ...
[21:03:46] <klapaucjusz> TheSHAD0W: not 2201, 2202. Sorry.  http://wiki.fon.com/wiki/Image:FON2202.gif
[21:04:27] <klapaucjusz> linux_jon: nmap
[21:04:48] 
[21:04:58] <alus> klapaucjusz: perhaps it was just left out of the ledbat spec by accident
[21:06:00] <klapaucjusz> alus: 4.8 of the Ledbat draft.
[21:07:13] <klapaucjusz> I don't see it in BEP-29 either.
[21:07:35] <The_8472> i think i've seen it somewhere in the ledbat draft
[21:08:27] <klapaucjusz> Okay, found it.  Ineed, apologies.
[21:08:47] <klapaucjusz> It will set its  packet_size and max_window to the smallest packet size (150 bytes)
[21:08:51] <klapaucjusz> Very aggressive.
[21:09:01] <klapaucjusz> Er, aggressively conservative :-)
[21:09:43] <klapaucjusz> alus: do you have anyone competent to explain the difference between Ledbat and TCP-LP/
[21:11:14] <The_8472> tbh, as long as they both achieve similar results the major is not that they're different but that one is deployed over TCP while the other is totally incompatible with everything.
[21:11:33] <klapaucjusz> The_8472:
[21:11:48] <The_8472> *major poin
[21:11:48] 
[21:11:50] <The_8472> t
[21:12:00] <The_8472> i know
[21:12:17] <klapaucjusz> So we need to make sure they don't break the Internet.
[21:12:40] <klapaucjusz> s/0.78/0.5/g, unless they can prove it doesn't break BLUE.
[21:12:48] <The_8472> well, i think they won't. at worst it will cause minor disruptions
[21:13:02] <klapaucjusz> Not breaking in this sense.
[21:13:35] <klapaucjusz> We want low delay in the Internet.
[21:14:04] <The_8472> well, afaik that's not a problem at the core routers. the last mile or shoddy isp-level infrastructure causes most of the problems
[21:14:24] <klapaucjusz> End-to-end congestion control is not enough to achieve that, because there'll always be some eegit who send CBR data.
[21:14:46] 
[21:15:44] <TheSHAD0W> FLASH NEWS: Massive disruption in the intertubes has been blamed on BitTorrent Inc's new transfer protocol.
[21:16:00] <klapaucjusz> You forgot "news at eleven".
[21:16:40] * klapaucjusz really needs to get his act together and put SFB into the Linux kernel.
[21:18:39] <TheSHAD0W> http://www.wired.com/epicenter/2009/11/4-ways-live-and-digital-music-are-teaming-up-to-rock-your-world/
[21:19:16] <TheSHAD0W> http://www.wired.com/threatlevel/2009/11/breach-laws
[21:26:18] 
[21:26:38] 
[21:27:29] <The_8472> as i said, i think i read about reducing it to a 1 packet window in ledbat too
[21:29:26] <K`Tetch> The_8472 - is there a page on your wiki about half-open patches?
[21:29:40] <The_8472> hrrm, i don't think so
[21:29:50] <The_8472> there's really no need for those anyway
[21:30:13] <K`Tetch> trying to educate a whole bunch of users
[21:30:18] <The_8472> since a) vista/7 don't do that anymore b) it just slows down connection establishments. not the total number of connections
[21:30:37] <The_8472> and linux never did btw
[21:30:44] <The_8472> it's just an XP thing
[21:31:14] <K`Tetch> any newb knows that xp limits the open connections 10 for security issues and that is why you apply this patch and set it 50 not 50, 000000000. and i will get back on this later. btw i had also patched my windows server with this and of course any newb will again know that you have to open your net max connections in UT for the full speeds to kick in.
[21:31:55] <K`Tetch> people still reccomending the hack, leading to stack instability, etc
[21:32:08] <klapaucjusz> K`Tetch: it only allows it to ramp up faster.
[21:32:11] <The_8472> speed is not correlated to half open connections, only the ramp up time for a torrent is. but that should be less than 2 minutes in either case
[21:32:12] <K`Tetch> ive dumped some ut forum links, but 2nd-source confirmation can help
[21:32:20] <klapaucjusz> Once its stationary (has enough connected peers) it's fine.
[21:32:34] <klapaucjusz> I'm actually thinking of modifying transmission to establish connections less aggressively.
[21:32:42] <The_8472> and incoming connections aren't affected by this. so if you're not NATed you won't see much of a difference
[22:00:45] <klapaucjusz> Okay, I think I understand LEDBAT.
[22:01:13] <klapaucjusz> It's just a PD controller, with halving on loss added.
[22:06:16] <klapaucjusz> Looking at the equations some more...
[22:06:52] *** Andrius has joined #bittorrent
[22:07:11] <klapaucjusz> If there is a very aggressive AQM in the way, and delay-based congestion control doesn't kick in, you get growth governed by GAIN.  If GAIN is larger than 1/MSS, then LEDBAT is going to starve Reno.
[22:08:48] <klapaucjusz> Of course, the draft doesn't say what GAIN is.
[22:11:27] <klapaucjusz> alus: what's the value of MAX_CWND_INCREASE_PACKETS_PER_RTT?
[22:33:09] *** gilles has joined #bittorrent
[22:36:27] *** K`Tetch_ has joined #bittorrent
[22:55:17] *** K`Tetch has quit IRC
[23:11:11] *** gilles has quit IRC
[23:17:40] *** _rafi_ has quit IRC
[23:19:17] *** L337hium has quit IRC
[23:27:48] <klapaucjusz> Experimental analysis of LEDBAT: http://arxiv.org/abs/0908.0812v1
[23:28:22] <klapaucjusz> Tail-drop buffers, of course.  Exhibited some serious inter-flow fairness issues, but other than that excellent results.
[23:30:20] <The_8472> i think there as a lengthy discussion on the late-start issue on the ledbat mailing list
[23:34:01] <klapaucjusz> Look at Fig. 3.  Pretty colours.
[23:34:46] <klapaucjusz> The difference between 3(2) and 3(3) is interesting.
[23:36:17] <klapaucjusz> It shows that being able to react to loss remains necessary.
[23:53:33] *** klapaucjusz has quit IRC





top