--- Log opened Tue Jul 31 00:00:17 2012 | ||
-!- heiko1 [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has joined #shogun | 00:06 | |
-!- heiko1 [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has quit [Quit: Leaving.] | 00:25 | |
-!- needsch [~user@ip-176-198-229-75.unitymediagroup.de] has quit [Ping timeout: 252 seconds] | 00:58 | |
CIA-18 | shogun: Sergey Lisitsyn master * r3bda438 / (4 files in 2 dirs): A bunch of fixes for multitask algoritmhs - http://git.io/tUKTlQ | 01:24 |
---|---|---|
-!- blackburn [~blackburn@109.226.100.87] has quit [Quit: Leaving.] | 01:26 | |
-!- ckwidmer [8ca3fe9d@gateway/web/freenode/ip.140.163.254.157] has joined #shogun | 02:35 | |
-!- puffin444_ [62e3926e@gateway/web/freenode/ip.98.227.146.110] has quit [Quit: Page closed] | 03:06 | |
-!- emrecelikten [~emre@176.40.234.211] has quit [Ping timeout: 240 seconds] | 03:50 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 246 seconds] | 04:51 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 06:58 | |
-!- uricamic [~uricamic@2001:718:2:1634:e506:de07:312c:a435] has joined #shogun | 07:35 | |
-!- av3ngr [~av3ngr@60-241-222-244.static.tpgi.com.au] has joined #shogun | 08:57 | |
-!- av3ngr [~av3ngr@60-241-222-244.static.tpgi.com.au] has left #shogun [] | 08:58 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Quit: leaving] | 09:02 | |
-!- pluskid [~pluskid@111.120.64.205] has joined #shogun | 09:22 | |
-!- pluskid [~pluskid@111.120.64.205] has quit [Ping timeout: 260 seconds] | 09:35 | |
-!- pluskid [~pluskid@108.171.196.83] has joined #shogun | 09:35 | |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has joined #shogun | 11:00 | |
yoo | hi al | 11:01 |
-!- heiko [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has joined #shogun | 11:38 | |
-!- blackburn [~blackburn@109.226.100.87] has joined #shogun | 12:03 | |
-!- emrecelikten [~emre@176.40.234.211] has joined #shogun | 12:03 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 12:06 | |
-!- pluskid [~pluskid@108.171.196.83] has quit [Ping timeout: 260 seconds] | 12:11 | |
-!- pluskid [~pluskid@111.120.64.205] has joined #shogun | 12:31 | |
-!- needsch [~user@ip-176-198-229-75.unitymediagroup.de] has joined #shogun | 12:38 | |
-!- heiko [~heiko@host86-181-156-213.range86-181.btcentralplus.com] has left #shogun [] | 14:36 | |
pluskid | blackburn: what's the working cycle if I want to add some multiclass learning stuff to the shogun tutorial? | 15:07 |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Read error: Operation timed out] | 15:32 | |
-!- pluskid [~pluskid@111.120.64.205] has quit [Ping timeout: 260 seconds] | 15:53 | |
-!- pluskid [~pluskid@108.171.196.83] has joined #shogun | 16:14 | |
-!- pluskid [~pluskid@108.171.196.83] has quit [Quit: Leaving] | 16:30 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 16:41 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 252 seconds] | 16:48 | |
-!- hoijui [~hoijui@141.23.65.251] has joined #shogun | 17:12 | |
-!- gsomix [~gsomix@95.153.166.127] has joined #shogun | 17:13 | |
-!- emrecelikten [~emre@176.40.234.211] has quit [Ping timeout: 264 seconds] | 17:16 | |
-!- emrecelikten [~emre@176.42.217.59] has joined #shogun | 17:18 | |
-!- uricamic [~uricamic@2001:718:2:1634:e506:de07:312c:a435] has quit [Quit: Leaving.] | 17:22 | |
-!- yoo [2eda6d52@gateway/web/freenode/ip.46.218.109.82] has quit [Quit: Page closed] | 17:32 | |
-!- gsomix [~gsomix@95.153.166.127] has quit [Ping timeout: 248 seconds] | 17:35 | |
-!- heiko [~heiko@dhcp-184-16.internal.eduroam.ucl.ac.uk] has joined #shogun | 18:17 | |
-!- hoijui [~hoijui@141.23.65.251] has quit [Ping timeout: 264 seconds] | 18:19 | |
-!- emrecelikten [~emre@176.42.217.59] has quit [Ping timeout: 255 seconds] | 18:23 | |
-!- emrecelikten [~emre@92.44.119.104] has joined #shogun | 18:25 | |
-!- emrecelikten [~emre@92.44.119.104] has quit [Ping timeout: 260 seconds] | 18:29 | |
-!- emrecelikten [~emre@92.44.119.53] has joined #shogun | 18:43 | |
-!- n4nd0 [~nando@s83-179-44-135.cust.tele2.se] has joined #shogun | 18:56 | |
-!- hoijui [~hoijui@dslb-092-078-244-135.pools.arcor-ip.net] has joined #shogun | 19:07 | |
@sonney2k | !seen gsomix | 19:18 |
@sonney2k | droopy is dead - oh noooo | 19:19 |
-!- emrecelikten [~emre@92.44.119.53] has quit [Ping timeout: 260 seconds] | 19:29 | |
blackburn | sonney2k: did you check chiyuan's PR? | 19:34 |
-!- heiko [~heiko@dhcp-184-16.internal.eduroam.ucl.ac.uk] has left #shogun [] | 19:56 | |
-!- gsomix [~gsomix@95.153.161.255] has joined #shogun | 19:58 | |
gsomix | report! | 19:59 |
gsomix | just a moment | 19:59 |
gsomix | I'm writing :) | 19:59 |
n4nd0 | gsomix: what news this week? :) | 20:03 |
gsomix | n4nd0, it seems, that 'no news'. I'm resting a little at sea. ^__^ | 20:05 |
n4nd0 | gsomix: nice! enjoy it; where? | 20:05 |
gsomix | n4nd0, http://bit.ly/McuADd | 20:05 |
blackburn | yellow one | 20:06 |
n4nd0 | cool | 20:06 |
blackburn | I've been to black sea - it is yellow believe me | 20:06 |
blackburn | :D | 20:06 |
gsomix | n4nd0, thanks. :) | 20:06 |
blackburn | gsomix: and I wdn't believe it is not | 20:06 |
gsomix | blackburn, green at now. | 20:07 |
blackburn | basically because russians are dirty pigs | 20:07 |
blackburn | with no culture and nothing | 20:07 |
gsomix | blackburn, huh. this is very much in the resorts... pigs, I mean | 20:09 |
blackburn | omg snail eating cucumber here | 20:09 |
blackburn | the ugliest thing I've ever seen | 20:10 |
n4nd0 | blackburn: dirty pigs? | 20:13 |
n4nd0 | I bet you don't like that much Russia :P | 20:14 |
blackburn | n4nd0: I don't like russians | 20:18 |
blackburn | n4nd0: most of russians are totally rude, uneducated and angry - there is no reason to love this people | 20:20 |
blackburn | n4nd0: well I generalize too much but russians you know in sweden are probably the best because they managed to leave :D | 20:21 |
n4nd0 | hehe | 20:21 |
-!- gsomix_ [~gsomix@95.153.184.16] has joined #shogun | 20:25 | |
-!- gsomix [~gsomix@95.153.161.255] has quit [Ping timeout: 255 seconds] | 20:26 | |
@sonney2k | gsomix_, around now? | 20:29 |
@sonney2k | blackburn, HUGE PR! | 20:29 |
@sonney2k | n4nd0, yeah do PLiFs first then COFFIN stuff | 20:29 |
@sonney2k | n4nd0, I hope you finally have some idea why this can be useful... | 20:30 |
n4nd0 | sonney2k: I have faith more than idea right now :O | 20:30 |
@sonney2k | n4nd0, think of you want to train a 10 million dimensional problem with 50 million examples on your notebook | 20:31 |
@sonney2k | that is when computing features on demand is the only way | 20:32 |
n4nd0 | yes, I understand that | 20:33 |
n4nd0 | so COFFIN is a way to manage features on demand? or something else? | 20:33 |
@sonney2k | only that yes | 20:34 |
@sonney2k | you compute *non-zero* features on demand | 20:34 |
n4nd0 | ok | 20:34 |
@sonney2k | this can be much faster than precomputing features | 20:34 |
@sonney2k | but can also be slower | 20:34 |
@sonney2k | depends on you Phi(x) | 20:34 |
@sonney2k | and the sparsity of x / Phi etc | 20:35 |
@sonney2k | n4nd0, in any case I can tell that the example I used in the coffin paper got reproduced on a 1000 node cluster at yahoo :) | 20:35 |
@sonney2k | with explicitly computed features :) | 20:36 |
n4nd0 | nice | 20:36 |
blackburn | hmm we need cluster modelselection | 20:36 |
n4nd0 | the problem is that here the Phi depends both on features and labels | 20:36 |
gsomix_ | sonney2k, will be in 30 minutes. ok? | 20:37 |
@sonney2k | gsomix_, I might be gone by then... | 20:37 |
@sonney2k | n4nd0, what is the problem? I don't see it (yet?). | 20:38 |
-!- gsomix_ is now known as gsomix | 20:38 | |
gsomix | sonney2k, ok. I'm here. | 20:38 |
@sonney2k | gsomix, so what is the state? | 20:39 |
@sonney2k | blackburn, one way would be to store parts of the modsel tree / together with everything else (on disk) and then run that part on some node | 20:39 |
@sonney2k | shouldn't be too hard | 20:40 |
blackburn | sonney2k: technologically? | 20:40 |
blackburn | concept is clear for sure | 20:40 |
@sonney2k | blackburn, well we can serialize the machine with everythingneeded | 20:40 |
@sonney2k | so all that is needed is dump parts of the tree too | 20:40 |
@sonney2k | ...and load it back in | 20:41 |
gsomix | sonney2k, there is problem with pointers. they breaks between get_buffer and release_buffer. | 20:41 |
gsomix | and it's strange | 20:41 |
blackburn | well map reduce may be | 20:41 |
n4nd0 | sonney2k: we need them a CFeatures object that holds both features and labels? | 20:41 |
gsomix | and only for SGVector. | 20:41 |
@sonney2k | blackburn, maybe not :D | 20:42 |
blackburn | sonney2k: what else? MPI? handmade solution | 20:42 |
blackburn | ? | 20:42 |
@sonney2k | gsomix, what does get/releasebuffer point to for sgvectors? | 20:43 |
@sonney2k | blackburn, I only have access to sun grid engine based clusters... | 20:43 |
n4nd0 | dinner time! | 20:43 |
-!- n4nd0 is now known as n4nd0_away | 20:44 | |
blackburn | sonney2k: I am pretty lame - does it mean you can't run say hadoop? | 20:44 |
@sonney2k | n4nd0, where do you have features and labels accessible currently? | 20:44 |
@sonney2k | blackburn, exactly | 20:44 |
blackburn | sonney2k: what can be ran? | 20:44 |
@sonney2k | blackburn, I have to call qsub <some executable / script > | 20:44 |
@sonney2k | then it will be run on some node | 20:44 |
blackburn | ohh qsub | 20:45 |
gsomix | sonney2k, in get_buffer we allocate memory for shapes and strides. | 20:45 |
blackburn | sonney2k: which technology would work on that kind of clusters? | 20:45 |
gsomix | and put it in PyView | 20:46 |
@sonney2k | gsomix, but you point to the memory of the SGVector? | 20:46 |
gsomix | nope | 20:46 |
@sonney2k | gsomix, where is that done | 20:46 |
gsomix | internal memory, only for buffer stuff. | 20:46 |
@sonney2k | gsomix, hmm - could the operators of SGVector += etc conflict with the ones you have in your pybuf * | 20:47 |
n4nd0_away | sonney2k: the features labels are accesible in the CStructuredModel, see https://dl.dropbox.com/u/11020840/shogun/SOFramework_classes.pdf | 20:48 |
gsomix | sonney2k, hardly. buffer stuff is independent of the operators. | 20:48 |
@sonney2k | n4nd0_away, maybe a structureddotfeatures class that has access to both labels and features would be the right thing then - it could supply you with the operations you usually do in structuredmodel | 20:51 |
@sonney2k | gsomix, well if pointers `break` someone else must be modifying them - did you check with valgrind? | 20:52 |
gsomix | sonney2k, nope. can valgrind detect it? cool. | 20:53 |
@sonney2k | gsomix, valgrind will detect any memory leaks / accidental memory access | 20:53 |
gsomix | btw python docs says, that this pointers should be const for other code. | 20:54 |
@sonney2k | (it seems this one is not intentional so probably) | 20:54 |
gsomix | ok, I will check it. | 20:54 |
@sonney2k | gsomix, anyway back to the big picture of the project - the last task (that I consider very important) are the typemaps for the model selection | 20:54 |
@sonney2k | gsomix, IIRC you have some intense university studies starting again in september so it is better to start very soon (in 1-2 days) with that ... | 20:56 |
gsomix | sonney2k, is there specification of syntax for model selection? | 20:56 |
@sonney2k | and postpone the debugging again | 20:56 |
@sonney2k | gsomix, well I sent you two emails with examples | 20:56 |
gsomix | sonney2k, aha | 20:57 |
@sonney2k | gsomix, btw pybuffer access for Labels should be much easier... | 21:00 |
gsomix | sonney2k, yes, of course | 21:01 |
@sonney2k | and that together with densefeatures is probably the most used. sgvector* stuff will only be used by people doing directors for rather complex stuff I guess | 21:01 |
gsomix | sonney2k, and working with numpy without copying... | 21:01 |
gsomix | so what's plan? | 21:01 |
@sonney2k | gsomix, just do it all :D | 21:03 |
gsomix | yay! | 21:03 |
@sonney2k | gsomix, but lets prioritize - for pybuffers just do the stuff with labels for now (that should be done by end of tomorrow or?) | 21:04 |
@sonney2k | gsomix, any estimates on how long numpy w/o copying would take? | 21:04 |
blackburn | sonney2k: octave celebrates JIT | 21:04 |
@sonney2k | blackburn, did you see my msg? | 21:04 |
blackburn | yeah | 21:04 |
blackburn | but they STILL celebrate | 21:04 |
blackburn | :D | 21:04 |
gsomix | sonney2k, labels - tomorrow. | 21:04 |
@sonney2k | blackburn, heheh | 21:05 |
gsomix | sonney2k, w/o copying - 1-2 days. | 21:05 |
blackburn | sonney2k: measure it in vodka/hour | 21:05 |
@sonney2k | gsomix, ok then just do it | 21:05 |
blackburn | i.e. | 21:05 |
blackburn | how much vodka/hours would it take, gsomix? | 21:05 |
@sonney2k | gsomix, IIRC we have a meeting on friday | 21:05 |
@sonney2k | gsomix, so after that meeting it would be good if you can discuss with heiko about the modsel typemaps | 21:06 |
gsomix | sonney2k, oh. P____p | 21:06 |
gsomix | what's time? | 21:06 |
@sonney2k | gsomix, because he knows how to create the parameter trees | 21:06 |
blackburn | we really need a hack for C1/C2 | 21:07 |
blackburn | it *should* be in 2.0 | 21:07 |
@sonney2k | blackburn, full ack | 21:07 |
blackburn | everybody complains | 21:07 |
@sonney2k | blackburn, can't we have the SGVector based solution with len=1 -> C / len=2 -> Cp Cn / len = n -> C[i] | 21:07 |
blackburn | sonney2k: we could but it would take a while to get it work | 21:08 |
@sonney2k | blackburn, only other alternative I see is to have a C member variable and a Cp / Cn member variable and a Cs member variable | 21:08 |
@sonney2k | and then some flag what is to be used :-[ | 21:09 |
blackburn | I'd suggest to add C for now | 21:09 |
@sonney2k | gsomix, 3rd of august 11 UTC | 21:09 |
@sonney2k | (as usual) | 21:09 |
gsomix | sonney2k, thanks. | 21:09 |
@sonney2k | gsomix, can you make it this time? | 21:09 |
@sonney2k | blackburn, or alternatively C2 = 0 indicates that C1 is to be used solely | 21:10 |
blackburn | sonney2k: well and the other one is to add something that ties parameters to each other | 21:11 |
@sonney2k | blackburn, btw I have *never* seen C1 / C2 to improve accuracy in practice... | 21:11 |
@sonney2k | blackburn, which is tough ... | 21:11 |
blackburn | sonney2k: why to have it then :D | 21:11 |
@sonney2k | heh | 21:11 |
blackburn | sonney2k: well actually nothing works in ML :D | 21:12 |
@sonney2k | blackburn, seems like droopy is on vacations | 21:12 |
@sonney2k | except for SVMs ;-) | 21:12 |
gsomix | sonney2k, ok, I'll try... right now. I'm resting a little with gf at sea. sorry for that. :) | 21:12 |
blackburn | sonney2k: I'm currently trying to model select multitask method on cancer data to achieve 0.62 AUC | 21:13 |
gsomix | ok, will be in 30 minutes and will work. | 21:13 |
blackburn | 30 minutes is not enough | 21:13 |
blackburn | ok I'm quiet :D | 21:13 |
blackburn | sonney2k: what I meant is that nothing works :D | 21:14 |
blackburn | svm too however | 21:14 |
@sonney2k | blackburn, which is already sth! | 21:15 |
@sonney2k | blackburn, I need help with the code review of pluskids patch | 21:15 |
@sonney2k | it is too big | 21:15 |
blackburn | sonney2k: 0.1 better than random uhhhh :D | 21:15 |
blackburn | sonney2k: well I've seen one thing there | 21:15 |
blackburn | I didn't get | 21:15 |
@sonney2k | gsomix, relax and have fun and then have fun again :D | 21:15 |
blackburn | I do not understand why to have kernel factory | 21:15 |
@sonney2k | blackburn, I asked that too | 21:16 |
blackburn | I am a big fan of factories | 21:16 |
@sonney2k | I've written ~10 comments or so | 21:16 |
-!- gsomix [~gsomix@95.153.184.16] has quit [Quit: Ex-Chat] | 21:16 | |
blackburn | I live in city full of factories | 21:16 |
blackburn | but still why to have it | 21:16 |
@sonney2k | blackburn, no idea | 21:16 |
@sonney2k | didn't make sense to me either | 21:16 |
@sonney2k | alright gtg too | 21:17 |
@sonney2k | cu | 21:17 |
blackburn | see you | 21:17 |
blackburn | well you commented on most things already | 21:17 |
blackburn | factories were the only confusing things here | 21:18 |
@sonney2k | well and we have a lot of std:: stuff in there.. | 21:49 |
@sonney2k | queue | 21:49 |
@sonney2k | algoritm | 21:49 |
@sonney2k | whatever | 21:49 |
@sonney2k | blackburn, pluskid is like an std:: virus | 21:49 |
@sonney2k | ;-) | 21:49 |
blackburn | sonney2k: heh I use std sometimes too | 21:50 |
@sonney2k | blackburn, you bastard! | 21:50 |
blackburn | sonney2k: I hate dynamicarray or how it is called | 21:50 |
blackburn | :D | 21:50 |
@sonney2k | blackburn, I guess you will never explain why :D | 21:54 |
blackburn | sonney2k: will you ever explain why to use handmade data structures? :) | 21:55 |
n4nd0_away | serialization was an answer once I think | 22:01 |
-!- n4nd0_away is now known as n4nd0 | 22:01 | |
blackburn | but internally? | 22:01 |
blackburn | we do not serialize everything | 22:01 |
n4nd0 | but we do serialize vectors, matrices, etc? | 22:02 |
n4nd0 | right? | 22:02 |
blackburn | sure but I use it only for temporary stuff like dynamic sets and dynamic vectors | 22:03 |
n4nd0 | ok | 22:03 |
n4nd0 | maybe there were more difficulties with std ones? | 22:03 |
n4nd0 | blackburn: let me ask you something | 22:06 |
blackburn | n4nd0: ask me | 22:06 |
n4nd0 | in our multiclass machines, the user is forced to use labels within {0, 1, ..., num_classes-1} | 22:06 |
n4nd0 | right? | 22:06 |
blackburn | exactly | 22:07 |
n4nd0 | ok | 22:07 |
n4nd0 | I am making right now the same assumption for my HMSVMLabels and the features I use for the HMSVM | 22:08 |
n4nd0 | and I feel a bit lazy to change it :) | 22:08 |
blackburn | I wish we had everything general | 22:08 |
n4nd0 | I will issue the PR like that and will hear from the lords | 22:08 |
blackburn | but we would end being lisp interpreter in the case | 22:09 |
n4nd0 | why? | 22:09 |
blackburn | I mean next step is to allow some gaps here | 22:10 |
blackburn | next is to allow strings | 22:10 |
blackburn | objects | 22:10 |
blackburn | :) | 22:10 |
blackburn | and so on | 22:10 |
n4nd0 | is it in the plans to support list? | 22:12 |
n4nd0 | lisp* | 22:13 |
blackburn | ?? | 22:13 |
blackburn | :D | 22:13 |
blackburn | why to have it? | 22:13 |
n4nd0 | haha I don't know | 22:13 |
n4nd0 | I am just trying to understand your previous comment :D | 22:13 |
blackburn | I mean tries of making everything general have no end | 22:14 |
n4nd0 | aham ok! I understand know | 22:14 |
blackburn | if we go for generality where to stop? | 22:14 |
n4nd0 | I understood your comment completely the other way round hehe | 22:14 |
blackburn | :) | 22:15 |
n4nd0 | blackburn: another question :) | 22:18 |
n4nd0 | let's say I have a local variable in a function, an SGVector | 22:19 |
blackburn | right | 22:19 |
n4nd0 | this function is going to be called lot of times during the execution of my algorithm | 22:19 |
n4nd0 | I want to avoid the allocation of memory for the structure every time is called | 22:19 |
blackburn | preallocate that vector then | 22:20 |
n4nd0 | I think that in java this could be done using static, for a local variable | 22:20 |
n4nd0 | I am not sure though | 22:20 |
n4nd0 | do I need to make this local variable a class member? | 22:20 |
n4nd0 | I don't really like this solution :S | 22:20 |
blackburn | yeah I see no other way | 22:20 |
blackburn | or put it to arguments :) | 22:21 |
n4nd0 | class member would be then | 22:21 |
blackburn | I think I have an idea | 22:22 |
n4nd0 | blackburn: do you think this kind of considerations do actually have impact at the end of the day? | 22:22 |
n4nd0 | tell me | 22:22 |
blackburn | we should implement some pool | 22:22 |
blackburn | which is global | 22:22 |
n4nd0 | summer is the right time for them | 22:22 |
blackburn | heh sure | 22:22 |
blackburn | so you do | 22:22 |
blackburn | VectorPool::get(size) | 22:22 |
blackburn | or so | 22:22 |
blackburn | and it returns preallocated stuff | 22:22 |
blackburn | allocations are not cheap but I don't know whether it really impacts there | 22:24 |
n4nd0 | I can test it before doing | 22:24 |
n4nd0 | check how long it takes and memory used | 22:24 |
n4nd0 | the idea of the pool is ok | 22:25 |
n4nd0 | but I don't see directly how it is avoided to use extra class members? | 22:26 |
n4nd0 | the VectorPool is required right? | 22:26 |
n4nd0 | but indeed it is an improvement | 22:27 |
n4nd0 | in my case I won't need just one SGVector | 22:27 |
blackburn | n4nd0: vector pool can be global | 22:32 |
-!- hoijui [~hoijui@dslb-092-078-244-135.pools.arcor-ip.net] has quit [Quit: Leaving] | 22:47 | |
blackburn | n4nd0: I just received swedish spam letter :D | 23:11 |
n4nd0 | blackburn: oh, not from me :P | 23:11 |
n4nd0 | what does it say? | 23:11 |
blackburn | that I've won in UK lottery | 23:11 |
blackburn | :D | 23:11 |
n4nd0 | nice | 23:12 |
blackburn | google spam filters work baaad | 23:12 |
n4nd0 | they tell to a Russian in Swedish that he has won the lottery in UK? | 23:12 |
blackburn | exactly they did | 23:12 |
n4nd0 | they are pretty bad spammers I think :DD | 23:12 |
--- Log closed Wed Aug 01 00:00:17 2012 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!