[00:04:23] *** gilles has quit IRC
[00:04:31] *** gilles has joined #bittorrent
[00:07:42] *** KyleK_ has quit IRC
[00:14:23] *** HandheldPenguin is now known as HandheldPenguin`
[00:56:24] *** L337hium has joined #bittorrent
[01:00:34] *** bbelt16ag has joined #bittorrent
[01:39:20] *** Miller` has quit IRC
[01:59:18] *** bpot has quit IRC
[02:07:57] 
[02:08:01] <klapaucjusz> please see http://forum.bittorrent.org/viewtopic.php?pid=724
[02:10:59] <TheSHAD0W> #4 sounds best to me.
[02:11:43] <klapaucjusz> I'd tend to agree.
[02:11:59] <klapaucjusz> I'm a little worried about the PMTU cache, though.
[02:12:33] <TheSHAD0W> If anything, it'll force a better cache implementation.
[02:21:59] *** PN has joined #bittorrent
[02:22:09] 
[02:22:14] <klapaucjusz> With per-byte ACKs.
[02:22:24] <TheSHAD0W> Mmm...
[02:24:21] <DeHackEd> great... TCP all over again..
[02:24:26] *** ProperNoun has quit IRC
[02:32:57] <klapaucjusz> DeHackEd: ?
[02:33:15] 
[02:34:06] <DeHackEd> no, but what you're looking for is something better than TCP yet you failed to learn from the original
[02:34:38] <klapaucjusz> Heh.
[02:34:43] <Nolar> not sure i see the problem
[02:34:59] <Nolar> utp messages always fit in a single packet afaik
[02:35:30] <Nolar> and should be sized appropriately based on pmtu responses
[02:35:45] <Nolar> am i missing something here?
[02:38:44] <klapaucjusz> Nolar: you do make the difference between local MTU and PMTU, right?
[02:38:53] <Nolar> not really
[02:39:03] <klapaucjusz> A == B -- C == D
[02:39:09] <Nolar> it's about the max packet size you can use to get to your destination
[02:39:23] <klapaucjusz> == is Ethernet (1500), -- is ADSL (1480)
[02:39:38] <Nolar> and?
[02:39:39] <klapaucjusz> Local MTU is 1500, for both A and D.  But the PMTU is 1480.
[02:39:56] <Nolar> right, and the point of pmtu discovery is to find the end2end max
[02:40:01] <klapaucjusz> There's no way to find out what the PMTU is without sending a 1500 octet packet and failing.
[02:40:12] <Nolar> at which pooint utp should pick that max size for communication
[02:40:20] <Nolar> and?
[02:40:42] <klapaucjusz> Read my message again.  Third paragraph, scenario that fails.
[02:41:00] <Nolar> ?
[02:41:08] <Nolar> utp is designed to fit into a single packet only
[02:41:22] <Nolar> splitting a message across two goes against the fundamental design
[02:41:36] <Nolar> at which point it becomes a question of finding the max size allowed
[02:42:06] <Nolar> if a packet is too big and gets dropped, that's ok
[02:42:30] <DeHackEd> Bell Canada blocks fragmented packets last time I checked. Stupid as sin, so don't even try item #1
[02:42:31] <Nolar> dropped data can be resent in smaller chunks
[02:43:12] <Nolar> i'd probably do something like testing max size via the SYN packet
[02:43:14] <klapaucjusz> Nolar: how do you do that?
[02:43:22] <klapaucjusz> What if PMTU decreases?
[02:45:21] <Nolar> then the packet fails
[02:45:28] <Nolar> just another lost packet
[02:46:04] <Nolar> given all the logic already handled by a utp client, handling mtu sizing should be part of the logic
[02:46:26] <Nolar> maybe alus could shed light on how that's currently handled in ut
[02:49:31] <klapaucjusz> I've just had a chat with him.
[02:49:37] <klapaucjusz> According to him, it's not handled.
[02:50:24] <Nolar> so no logic to differentiate "normal" packet drops and ones dropped to mtu, which could be fixed by smaller packet sizes?
[02:50:56] <klapaucjusz> Nolar: you cannot repacketise.  You cannot send a smaller packet.
[02:50:58] <Nolar> or do they just use a safe max and dont worry about it
[02:51:05] <Nolar> *sigh*
[02:51:11] <Nolar> utp is a single packet msg
[02:51:37] <Nolar> you just send again, this time with a smaller payload
[02:52:33] * klapaucjusz gives up
[02:52:49] <Nolar> heh
[02:53:15] <Nolar> i'll follow the forum posts :)
[02:53:19] *** chelz has joined #bittorrent
[02:53:32] <Nolar> i'm sure someone at bt inc will jump in
[02:55:08] <Nolar> but i've written udp apps before, so i'd be highly surprised if they didnt come across the mtu problem long ago
[02:55:25] <The_8472> nobody has commented in over a week. despite there being more activity on the forum than usual
[02:55:44] <klapaucjusz> The_8472: I just posted this message.
[02:55:57] <The_8472> i mean nobody from bt inc ^^
[02:56:08] <The_8472> see ipv6 dht thread
[02:56:16] <The_8472> the bep hasn't been updated
[02:56:20] <Nolar> The_8472 well, if they are anything like us, it's a weekly check through of the forums :)
[02:56:29] <The_8472> lazybums ^^
[02:56:38] <Nolar> overworked more like it
[02:56:45] <The_8472> shush
[02:56:54] <Nolar> :)
[02:56:56] <The_8472> don't destroy my arguments
[02:57:18] *** waldorf_ has quit IRC
[02:57:27] <Nolar> nah, just my own personal excuses
[02:58:16] *** waldorf_ has joined #bittorrent
[02:58:37] *** waldorf_ has quit IRC
[02:59:22] 
[02:59:58] <The_8472> TCP exists, it has the features it has for a reason (well, most of them) and we only really need to replace the congestion controller
[03:00:34] <The_8472> but no. yet-another UDP-transport springs to life
[03:00:40] <The_8472> like dozens before it
[03:00:58] * klapaucjusz switches into rant mode
[03:01:21] 
[03:01:54] <klapaucjusz> http://forum.utorrent.com/viewtopic.php?id=59150
[03:02:06] <chelz> if it makes sense i think other implementors would either follow suit or come out with their own brand. similar to how the DHT stuff has worked out.
[03:02:24] <Nolar> i disagree that moving to something like utp is unnecesary
[03:02:37] <Nolar> there are biiig wins
[03:02:40] <chelz> moving sounds bad, but supporting it in addition sounds ok
[03:02:47] <Nolar> but ya, it does suck that it was developed in secret
[03:02:48] <The_8472> then plug ledbat into tcp.
[03:03:05] <chelz> it would be better if they engaged other implementors in the beginning, i agree
[03:03:05] <Nolar> makes it less likely that we'd impl it
[03:03:40] <Nolar>  <chelz> if it makes sense i think other implementors would either follow suit or come out with their own brand.  <<< that's my worry
[03:04:03] <Nolar> i wont comment on the name itself ;)
[03:04:22] 
[03:05:03] <Nolar> well, i consider them as one
[03:05:11] 
[03:05:46] <Nolar> what exactly differentiates them for you?
[03:06:09] 
[03:06:27] <Nolar> potatoes potatoes :)
[03:06:29] <Nolar> one impl
[03:06:53] <Nolar> not like they'd swap out utp for something else
[03:07:00] <The_8472> plug ledbat into tcp. give apps a socket option to use it. problem solved.
[03:07:07] <Nolar> hah
[03:07:17] <Nolar> good like with getting that pushed through committee ;)
[03:07:25] <The_8472> worked with vegas
[03:07:27] <Nolar> lUck
[03:07:34] <klapaucjusz> Eh?
[03:07:45] <Nolar> and tcp itself has flaws when bt'ing
[03:07:49] <klapaucjusz> You don't need a committee decision to just add a congestion control algorithm to your favourite kernel.
[03:07:57] <chelz> well ideally everyone would support the same protocol. but i think it would be better for there to be different support rather than no support at all, like the DHT things
[03:08:21] <Nolar> klapaucjusz and alll other peers' kernels....
[03:08:26] <Nolar> it's ipv6 all over again
[03:08:51] <Nolar> i'm just saying, i understand why it was built as it was
[03:09:08] <Nolar> and i agree with the core idea
[03:09:17] <The_8472> microsoft would be the biggest fish to fry. idk if you can just hook a new congestion controller into their kernel. they CAN change congestion controllers (like the regular one vs. compound tcp).
[03:09:43] <Nolar> the approach to introducing it i dont lilke so much
[03:10:00] <Nolar> i reserve judgement on the impl
[03:10:53] <The_8472> well, and i just see yet-another UDP transport. everyone is brewing their own. not exactly what we need
[03:11:28] <klapaucjusz> Just to calm everyone down...
[03:11:33] <Nolar> wait until you see my version!
[03:11:48] * The_8472 glares at Nolar
[03:12:01] <chelz> haha
[03:12:05] 
[03:12:06] <Nolar> :)
[03:12:25] <klapaucjusz> They've just been implemented too early, without consulting people.
[03:13:03] <klapaucjusz> No problem with having two distinct encapsulations for Ledbat, one over UDP for legacy systems, and one native.
[03:14:25] <The_8472> well, imo more effort should have been put into getting it adopted as TCP option (like writing a linux kernel module) instead of cementing it into some other UDP transport
[03:14:41] <The_8472> why add it to TCP if clients already use it over UDP anyway.
[03:14:44] <The_8472> lack of incentive
[03:19:04] <The_8472> <Nolar> and tcp itself has flaws when bt'ing <- so, could you point out some advantages of using UDP instead of TCP?
[03:20:28] <The_8472> <Nolar> it's ipv6 all over again <- not necessarily. tcp already supports timestamps. so even non-compatible peers could be used. assuming their precision is sufficient.
[03:21:04] <Nolar> lower connection overhead, far easier to multiplex thousands of peer msgs
[03:21:13] <klapaucjusz> Eh?
[03:21:31] <Nolar> it just fits better with bittorrent's chunk oriented approach
[03:21:38] <klapaucjusz> Eh?
[03:22:08] <Nolar> there's no need for a serial stream of data
[03:22:17] <The_8472> uhm. you'd still have to reassamble PIECE messages (tcp reassembly does that for you). you still need connection states. (like... socket pairs!)
[03:22:21] <klapaucjusz> Eh?
[03:22:21] <Nolar> since it is reassembled in order and verified client side
[03:22:28] *** bbelt16ag has quit IRC
[03:22:53] <Nolar> klapaucjusz same reason i dont see issue with mtu
[03:22:55] 
[03:23:03] <Nolar> a lost packet can be made up from some other peer
[03:23:20] <Nolar> tcp requires maintened connections
[03:23:22] 
[03:23:24] <Nolar> maintained
[03:23:38] <Nolar> eh?
[03:23:49] 
[03:23:56] 
[03:24:02] <klapaucjusz> If you lose packet n, you cannot advance beyond n + window_size
[03:24:10] <klapaucjusz> The_8472: right.
[03:25:13] <The_8472> really. it's just a tcp-knockoff. some things shaved off. its only purpose is to serve as a vehicle for a new congestion controller as far as i can see.
[03:25:25] <The_8472> that's what i've been ranting about the whole time ^^
[03:25:50] <Nolar> klapaucjusz but packets are independent
[03:26:08] <Nolar> you should be able to lose 1 in the middle and still use all the other ones
[03:26:21] <Nolar> anyway
[03:26:34] <Nolar> it's out, it exists, and it's not going anywhere
[03:26:48] <Nolar> despite The_8472's best efforts :)
[03:27:28] <The_8472> well, imo it should be a temporary solution. like using teredo as a form of nat traversal...
[03:27:36] <Nolar> that works
[03:27:44] <The_8472> and needs to be replaced
[03:28:07] <Nolar> but if you think tcp would adopt ledbat in any reasonable time, i think you're mistaken
[03:28:16] <The_8472> problem with temporary solutions is that they take away the incentive to build a real one. see NAT vs. ipv6
[03:28:23] <Nolar> true
[03:28:35] <The_8472> without NAT we'd have ipv6 today
[03:28:48] <Nolar> but you have to weigh that against it not happening at all
[03:28:56] <Nolar> maybe
[03:30:02] <Nolar> i dont think bt inc saw waiting many years before they could deliver the feature as a realistic option
[03:30:09] *** init0 has joined #bittorrent
[03:30:15] <Nolar> i'd like proper world wide multicast
[03:30:25] <Nolar> i could wait years and years for it to maybe happen
[03:30:27] <The_8472> ISP-local would be a start
[03:30:34] <Nolar> or i could invent bittorrent and solve the problem today ;)
[03:30:37] <klapaucjusz> ...and a Pony.
[03:30:55] <The_8472> well, in the case of bittorent it's sortof inverse
[03:31:10] <The_8472> bittorrent makes an excellent argument why SSM should be deployed
[03:31:43] 
[03:31:58] <Nolar> i dont see the difference
[03:32:12] <Nolar> the "proper" route takes years and may never come to pass
[03:32:32] <Nolar> so people come up with less-than-proper ways to solve the problem
[03:32:51] 
[03:32:59] <klapaucjusz> The_8472: http://www.ece.rice.edu/networks/TCP-LP/
[03:33:08] <Nolar> i see bittorrent as a Today-solution for a problem that would otherwise not get solved for years
[03:33:15] <Nolar> utp is similar.
[03:33:54] <The_8472> yes, you're talking about the now
[03:34:01] <The_8472> i'm talking about the then
[03:34:12] <Nolar> waiting 3 years to get out of committee, another few for it to get implemented in stacks, and a few more before it's got enough market penetration to be useful
[03:34:24] <The_8472> lack of foresight got us such nice things as the reserved bitfield in bittorrent. or NAT...
[03:34:32] <Nolar> we call that feature dead on arrival :)
[03:35:18] <The_8472> yesyes. and years later people realize they created tons of legacy problems and curse about them
[03:36:24] <Nolar> and meanwhile the problem goes totally unsolved?
[03:36:31] 
[03:37:10] <The_8472> Nolar, not unsolved. you'd have the proposal and could point anyone at it. "if you want a solution, here is it. get it adopted"
[03:37:22] <Nolar> ok, well back in the real world...
[03:37:48] <The_8472> oh, it does work
[03:37:56] <The_8472> see the DNS exploit
[03:38:10] <The_8472> things can move fast if necessary
[03:40:15] *** init0_ has quit IRC
[03:41:19] <The_8472> in fact, that one was the most impressive industry-wide coordination i've ever seen when it comes to the internet. they got all major players to deploy a patch simultanously in such a small timeframe.
[03:43:22] <The_8472> and even if it's slowmoving
[03:43:27] <The_8472> bittorrent existed for years
[03:43:42] <The_8472> what bad would waiting a year more to get a proper solution do?
[03:43:51] <Nolar> it's not a year
[03:44:00] <The_8472> or a few years. details.
[03:44:11] <Nolar> at least 5-10 to get it live
[03:44:59] <Nolar> longer, if Comcast gets wind of it ;)
[03:45:18] <The_8472> it should be in their interest to shorten the process oO
[03:45:28] <TheSHAD0W> I think that it's not our responsibility to solve infrastructure problems, but only to give the best experience to our users.
[03:47:03] <The_8472> TheSHAD0W, UDP connections tend to fry home routers a lot more than TCP ones for starters
[03:47:21] <The_8472> so it's not just positive user experience
[03:47:25] <klapaucjusz> Let them eat cake.  (Or install OpenWRT.)
[03:47:35] <The_8472> i have ddwrt ^^
[03:49:14] <klapaucjusz> That's just OpenWRT with some makeup.
[03:50:05] <The_8472> i thought they parted ways years ago?
[03:52:38] <klapaucjusz> Ah?  Possible.  At any rate DD-WRT is okay too.
[03:55:26] *** gilles has quit IRC
[03:56:31] *** klapaucjusz has quit IRC
[03:57:34] *** The_8472 has quit IRC
[03:57:55] *** wadim has joined #bittorrent
[03:57:57] *** wadim is now known as The_8472
[03:59:02] *** Miller` has joined #bittorrent
[04:10:33] *** bbelt16ag has joined #bittorrent
[04:12:37] *** bpot has joined #bittorrent
[04:31:08] *** The_8472 has quit IRC
[06:28:42] *** Andrius has joined #bittorrent
[06:30:13] *** MassaRoddel has quit IRC
[07:03:00] *** MassaRoddel has joined #bittorrent
[07:49:50] *** gilles has joined #bittorrent
[08:20:16] *** gilles has quit IRC
[09:00:42] *** goussx has quit IRC
[09:02:32] *** cyb2063 has joined #bittorrent
[09:06:28] *** thermal has quit IRC
[09:06:37] *** therma has joined #bittorrent
[09:21:22] *** goussx has joined #bittorrent
[09:22:53] *** bbelt16ag has quit IRC
[10:06:32] *** waldorf_ has joined #bittorrent
[10:36:43] *** chelz has quit IRC
[11:26:18] *** pajlada has joined #bittorrent
[13:10:23] *** L337hium has quit IRC
[14:10:27] *** HandheldPenguin` is now known as HandheldPenguin
[14:16:56] *** The_8472 has joined #bittorrent
[14:17:11] *** betadoctor has joined #bittorrent
[14:30:13] *** _rafi1_ has joined #bittorrent
[14:42:07] *** HandheldPenguin is now known as HandheldPenguin`
[15:05:30] *** _rafi_ has joined #bittorrent
[15:09:43] *** HandheldPenguin` is now known as HandheldPenguin
[15:11:07] *** ycy has quit IRC
[15:15:45] *** ycy has joined #bittorrent
[15:20:30] *** HandheldPenguin is now known as HandheldPenguin`
[15:20:39] *** HandheldPenguin` is now known as HandheldPenguin
[15:23:31] *** _rafi1_ has quit IRC
[15:44:46] *** bittwist has quit IRC
[15:46:10] *** HandheldPenguin is now known as HandheldPenguin`
[15:50:51] *** _rafi1_ has joined #bittorrent
[15:56:56] *** _rafi_ has quit IRC
[16:04:56] *** cyb2063 has quit IRC
[16:19:33] *** DWKnight has joined #bittorrent
[16:19:38] *** DWKnight has quit IRC
[16:20:06] *** DWKnight has joined #bittorrent
[16:24:40] *** DreadWingKnight has quit IRC
[16:28:23] *** Snoopotic has joined #bittorrent
[16:32:29] *** pevangelista has joined #bittorrent
[16:32:55] <pevangelista> Hello to all!
[16:33:58] <pevangelista> I am studying the rarest first policy and I want to know if I got it right
[16:34:09] *** Elrohir has joined #bittorrent
[16:36:16] *** HandheldPenguin` is now known as HandheldPenguin
[16:36:27] <pevangelista> So, among all the peers that are connected to the the client, only a subset can upload pieces to it (because the client is interested in these peers and they are not choking him)
[16:37:47] <pevangelista> I am thinking that the client considers only the pieces that are readily available when applying the rarest policy
[16:37:52] <pevangelista> Is that right?
[16:46:53] <burris> no
[16:47:49] <burris> when deciding which piece to request from a client, you download the piece they have that is the rarest among all connected clients, including ones that are currently choking you...
[16:48:00] <TheSHAD0W> burris: Wrong also.
[16:48:06] <TheSHAD0W> It's actually pretty complicated.
[16:49:02] <TheSHAD0W> You do want to pick rare pieces, but it's also important to make sure you complete pieces you've already partially downloaded.  Finding the right balance is difficult.
[16:50:04] <burris> of course there is a lot more to it but you don't just consider pieces among peeers that aren't choking you, which is what pevangelista wanted to know
[16:50:19] *** _rafi_ has joined #bittorrent
[16:50:31] <TheSHAD0W> Yup.
[16:50:55] <TheSHAD0W> The idea is to get peers to unchoke you, which means you have to pick pieces they desire also.
[16:52:07] <pevangelista> But the client should not wait these rarest pieces to become available, or else he will never start downloading anything
[16:53:15] <pevangelista> I mean, if the interesting peers that unchoked the client have only common pieces, the client should start downloading something, and when a rarest piece becomes available for download, the client prioritizes it
[16:53:18] *** Elrohir has quit IRC
[16:53:27] <pevangelista> That's my guess
[16:54:11] <burris> yes, of cours, you download the rarest piece that the client is offering you, but you consider all peers including ons that are choking you when determining what is rare, but if all the peer has is "common" pieces you still dl, just whichever one is most rare
[16:54:56] <pevangelista> ah, I got it! Now I saw the problem with restricting the set of peers when defining what is rare
[16:55:15] *** _rafi1_ has quit IRC
[16:55:16] *** HandheldPenguin is now known as HandheldPenguin`
[16:55:40] <pevangelista> This is one of those times when you need to say out loud (or write) to realize something =)
[16:55:45] <burris> if the peer offers you two pieces, one that every other connected peer has and one that no other connected peer has then you download the one that no other peer has, but if the peer is offering you two pieces and all the other peers have those pieces, then you just pick one at random and download it anyway
[16:56:26] *** bittwist has joined #bittorrent
[16:56:32] <burris> assuming you want that piece too... :-)
[16:57:54] <pevangelista> yup! Thanks, burris! You too, TheSHADOW! It was of great help! Rest assured that if I have other questions, I will come here ;)
[16:59:39] <mpl> that's was very informative, thx.
[16:59:44] <mpl> *that
[17:00:13] *** uau has quit IRC
[17:00:25] *** uau has joined #bittorrent
[17:00:46] *** sparr has quit IRC
[17:00:56] *** andar has quit IRC
[17:00:56] *** hlindhe has quit IRC
[17:00:59] *** hlindhe_ has joined #bittorrent
[17:01:32] *** andar has joined #bittorrent
[17:01:32] *** sparr has joined #bittorrent
[17:01:59] <burris> anyone tried BT's self service DNA thingy?
[17:03:17] *** Nikty has joined #bittorrent
[17:03:24] <Nikty> hi
[17:03:57] <Nikty> what's the simplest way to set up own tracker and forum?
[17:04:23] <TheSHAD0W> Separately?
[17:04:28] <Nikty> together
[17:04:39] <Nikty> i need something like TPB, but local
[17:04:43] <Nikty> for intranet
[17:04:55] <TheSHAD0W> No, I mean, set up a tracker, and set up a forum.
[17:05:28] <TheSHAD0W> For an intranet, a script tracker should be plenty.  Look up phpbtt.
[17:05:29] <Nikty> i think, it would be ok
[17:06:06] <TheSHAD0W> For a forum, well, there's a few to choose from.
[17:06:22] <burris> for intranet you could just use the one built into mainline then post the torrent files wherever, like e-mail IM or Notes or whatever you already use
[17:06:22] <TheSHAD0W> Look up what your favorite forum uses.
[17:06:37] <Nikty> aha.
[17:07:21] <Nikty> i need comments
[17:07:29] <Nikty> not onlu a tracker
[17:07:32] <Nikty> *only
[17:07:37] <The_8472> you need an indexing site
[17:07:47] <The_8472> that's orthogonal to a tracker
[17:14:04] <Nikty> yeah, it's hard to do :(
[17:14:12] <Nikty> thanks
[17:15:02] *** bpot has quit IRC
[17:28:09] *** KyleK_ has joined #bittorrent
[17:28:48] *** bpot has joined #bittorrent
[17:31:09] <betadoctor> nikty .. try xbtit .. this should do ure trick
[17:31:38] <betadoctor> indexer + built in tracker + built in forums
[17:37:03] *** bpot has quit IRC
[17:41:05] *** KyleK_ has quit IRC
[17:53:20] <burris> friis and zennstrom are geniuses
[17:55:24] <betadoctor> yeah they are ... and criminals ;)
[17:56:25] <burris> many criminals do very well for themselves
[17:56:33] <burris> just ask bush or blair
[17:57:13] <The_8472> how is creating software a criminal act?
[17:57:55] <burris> because the legislature or sovereign says it is?
[17:58:09] <The_8472> time to leave the country then
[17:58:15] <burris> which one?
[17:58:26] <The_8472> the one you're talking about
[18:09:17] *** _rafi1_ has joined #bittorrent
[18:17:41] *** cgreco has quit IRC
[18:19:32] *** cgreco has joined #bittorrent
[18:26:18] *** _rafi_ has quit IRC
[18:27:34] *** goussx has quit IRC
[18:33:41] *** Miller` has quit IRC
[18:41:45] *** klapaucjusz has joined #bittorrent
[18:43:33] <klapaucjusz> alus: yo!
[18:43:53] 
[18:54:11] *** goussx has joined #bittorrent
[18:55:28] *** _rafi_ has joined #bittorrent
[19:12:15] *** _rafi1_ has quit IRC
[19:13:25] *** Nikty has left #bittorrent
[19:29:35] *** pevangelista has left #bittorrent
[19:37:25] *** _rafi1_ has joined #bittorrent
[19:42:37] *** ProperNoun has joined #bittorrent
[19:43:13] *** PN has quit IRC
[19:44:22] *** klapaucjusz has quit IRC
[19:55:27] *** _rafi_ has quit IRC
[20:04:54] *** hlindhe has joined #bittorrent
[20:10:46] *** andar2 has joined #bittorrent
[20:26:35] *** pevangelista has joined #bittorrent
[20:27:31] <pevangelista> Hello again! I have a question about choking. A Peer can receive a choke message from another Peer when in the middle of a download with it?
[20:28:36] <pevangelista> I remember that the choking cicle is time-based (period of 4s, for instance), so what happens if a Peer decides to choke another in the middle of a download?
[20:28:47] <pevangelista> Is the piece download canceled?
[20:41:57] <Nolar> clients grab pieces in ~16kb chunks
[20:42:06] <Nolar> i.e a message = 1 chunk
[20:42:41] <Nolar> so if you get a choke, you assume 1) that there wont be any more pieces coming down the pipe, even though you might have requested them
[20:43:04] *** DWKnight has quit IRC
[20:44:33] <Nolar> 2) no more requests allowed to that peer until they unchoke you again
[20:45:15] <Nolar> 1) that there wont be any more *chunks* coming down the pipe
[20:45:50] <Nolar> there are some subtle timing issues involved, but that's the basic idea
[20:48:00] <pevangelista> ah, I see...
[20:48:29] <pevangelista> Actually, now that I am thinking about it, my question doesn't make much sense
[20:48:56] <pevangelista> the choke message will be the last one the choked peer will receive
[20:49:42] <pevangelista> Not that I wasn't thinking before, its just that discussing with others help me understand better ;)
[20:51:48] *** bittwist has quit IRC
[20:52:51] <Nolar> ya, you should not receive any more Piece msgs after a choke
[20:54:56] <pevangelista> thanks Nolar
[20:59:52] <Nolar> np
[21:13:15] *** PN has joined #bittorrent
[21:19:36] *** bbelt16ag has joined #bittorrent
[21:33:48] *** ProperNoun has quit IRC
[21:33:50] *** bbelt16ag has quit IRC
[21:33:56] *** ProperNoun has joined #bittorrent
[21:36:05] *** bbelt16ag has joined #bittorrent
[21:47:10] *** DWKnight has joined #bittorrent
[21:48:00] *** bbelt16ag has quit IRC
[21:50:07] *** bittwist has joined #bittorrent
[21:50:13] *** bbelt16ag has joined #bittorrent
[21:52:38] *** PN has quit IRC
[21:56:08] *** pevangelista has quit IRC
[22:01:43] *** bbelt16ag has quit IRC
[22:03:33] *** bbelt16ag has joined #bittorrent
[22:07:46] *** bbelt16ag has quit IRC
[22:09:55] *** bbelt16ag has joined #bittorrent
[22:14:02] *** bbelt16ag has quit IRC
[22:23:15] *** PN has joined #bittorrent
[22:23:16] *** medecau has joined #bittorrent
[22:27:33] *** burris has quit IRC
[22:27:45] *** burris has joined #bittorrent
[22:28:59] *** waldorf_ has quit IRC
[22:31:36] *** bbelt16ag has joined #bittorrent
[22:33:26] *** _rafi1_ has quit IRC
[22:34:01] *** ProperNoun has quit IRC
[22:34:13] *** _rafi_ has joined #bittorrent
[22:50:20] *** ivan` has left #bittorrent
[23:08:05] *** L337hium has joined #bittorrent
[23:23:40] *** waldorf_ has joined #bittorrent
[23:42:50] *** medecau has quit IRC
[23:44:21] *** waldorf_ has quit IRC





top