IRC logs of #shogun for Thursday, 2013-12-19

--- Log opened Thu Dec 19 00:00:13 2013
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: freedom, friends, features, first ---> fedoraproject.org]00:25
-!- bb_ [458fc8de@gateway/web/freenode/ip.69.143.200.222] has quit [Ping timeout: 272 seconds]03:20
shogun-buildbotbuild #652 of nightly_default is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/65204:06
-!- Saurabh7 [~Saurabh7@117.200.187.127] has joined #shogun05:26
Saurabh7hello, can anyone tell me why the java modular thing fails in https://travis-ci.org/shogun-toolbox/shogun/builds/15686637?05:29
-!- lisitsyn [~lisitsyn@80.252.20.67] has quit [Ping timeout: 252 seconds]06:22
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has quit [Ping timeout: 250 seconds]06:25
-!- tricksy_ is now known as tricksy07:10
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun07:11
-!- Saurabh7 [~Saurabh7@117.200.187.127] has quit [Ping timeout: 240 seconds]07:17
-!- Saurabh7 [~Saurabh7@59.88.34.245] has joined #shogun07:29
lisitsynwiking: we are on the way to win hah07:38
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has joined #shogun07:48
-!- Saurabh7 [~Saurabh7@59.88.34.245] has quit [Ping timeout: 248 seconds]07:51
-!- Saurabh7 [~Saurabh7@117.212.159.94] has joined #shogun08:03
-!- Saurabh7 [~Saurabh7@117.212.159.94] has quit [Quit: Leaving]08:55
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has quit [Quit: Leaving.]09:26
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has joined #shogun09:26
-!- xanthus [uid15721@gateway/web/irccloud.com/x-ntgkbkbxkgsbammv] has joined #shogun09:56
wikinglisitsyn: \o/10:11
wikinglisitsyn: seems ppl are not so keen on the licensing story :P (only 3 votes till now)10:45
lisitsynwiking: I can't decide between MIT and BSD10:47
lisitsyn:D10:47
wikinglol10:47
wikingjust pick one10:47
wikinganyhow bsd is not really specified which one of that :P10:48
wiking3 liner 2 liner etc.10:48
lisitsynwiking: I like 2-clause more10:50
lisitsynwiking: I am quite tired of beliefs any company will ever commit anything10:50
lisitsyndue to lgpl/gplness10:50
lisitsynthis has nothing to do with 2/3-clause, just recalled :)10:51
wikingyeah sure10:52
wikingsee webkit10:52
wikingthat was working nicely for a while :P10:52
lisitsynwiking: yeah if you are webkit10:53
lisitsynor linux10:54
lisitsynwiking: still believe gpl would only scare away and that's it10:54
lisitsynno commitment to commit anything :)10:54
wikingyeah well now we only have to decide which way to go11:02
wikingand send out ~90 emails :)11:02
wikingand in worst case drop out some code :P11:02
lisitsynI like that11:02
lisitsyndropping code is the most effective thing11:02
wikingwell this way we might be forced to do so :)11:02
lisitsynwiking: it is good to be forced to do good!11:03
wikingheheh well yeah i think we'll need about 2-3 months to get this thing done11:03
wikingso i think we could barely finish it before next gsoc :P11:04
wikingseems the cmake improvements are optimized for fedora and like distors as nothing works from that on a normal distribution11:05
-!- besser82 [~besser82@2a02:8108:8840:1800:e8b:fdff:fe16:bb33] has joined #shogun11:19
-!- besser82 [~besser82@2a02:8108:8840:1800:e8b:fdff:fe16:bb33] has quit [Changing host]11:19
-!- besser82 [~besser82@fedora/besser82] has joined #shogun11:19
lisitsynwiking: it doubled the number of website visitors haha12:12
wiking:D12:23
wikinglol12:23
wikingwell reddit effect12:23
wikinglol github tracker works!12:25
-!- bb_ [458fc8de@gateway/web/freenode/ip.69.143.200.222] has joined #shogun13:16
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun13:24
shogun-notifier-shogun: Saurabh7 :develop * 0512023 / tests/unit/evaluation/ (2 files): https://github.com/shogun-toolbox/shogun/commit/0512023bfb9cf4dacb4f888598b5ddcb2b5a5e4d13:24
shogun-notifier-shogun: test cross validation fold indices13:24
shogun-notifier-shogun: Saurabh7 :develop * 0f9ddbd / tests/unit/evaluation/ (2 files): https://github.com/shogun-toolbox/shogun/commit/0f9ddbd05c3bce46d3432a6b21df8834841f26a813:24
shogun-notifier-shogun: modified one file to include all tests13:24
shogun-notifier-shogun: Saurabh7 :develop * 32725f5 / tests/unit/evaluation/ (2 files): https://github.com/shogun-toolbox/shogun/commit/32725f50f23ece9771b7fce70821f2007da9903413:24
shogun-notifier-shogun: remove file13:24
shogun-notifier-shogun: Saurabh7 :develop * ef6347d / tests/unit/evaluation/CrossValidation_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/ef6347dd3fe73fe6d04963541e4b9d5e2445a42513:24
shogun-notifier-shogun: fixed typos in crossvalidation tests13:24
shogun-notifier-shogun: Saurabh7 :develop * f4f8f07 / tests/unit/evaluation/CrossValidation_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/f4f8f07ec14e5007781c389594d3c91484e45d8213:24
shogun-notifier-shogun: test covering of all indices13:24
shogun-notifier-shogun: Saurabh7 :develop * 7458c16 / tests/unit/evaluation/CrossValidation_unittest.cc: https://github.com/shogun-toolbox/shogun/commit/7458c1696515b99217be8c08873bdce2955084fb13:24
shogun-notifier-shogun: test covering of all indices13:24
shogun-buildbotbuild #2543 of deb1 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/2543  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Saurabh7 <saurabh.mahindre@gmail.com>13:34
wikinghuh that's a failure13:45
-!- iglesiasg [~iglesiasg@c-3415e555.08-88-73746f24.cust.bredbandsbolaget.se] has joined #shogun13:45
-!- mode/#shogun [+o iglesiasg] by ChanServ13:45
-!- bb_ [458fc8de@gateway/web/freenode/ip.69.143.200.222] has quit [Ping timeout: 272 seconds]13:59
-!- Saurabh7 [~Saurabh7@117.200.185.218] has joined #shogun14:01
-!- iglesiasg [~iglesiasg@c-3415e555.08-88-73746f24.cust.bredbandsbolaget.se] has quit [Quit: Leaving]14:06
-!- travis-ci [~travis-ci@ec2-23-22-78-97.compute-1.amazonaws.com] has joined #shogun14:19
travis-ci[travis-ci] it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: http://travis-ci.org/shogun-toolbox/shogun/builds/1570455114:19
-!- travis-ci [~travis-ci@ec2-23-22-78-97.compute-1.amazonaws.com] has left #shogun []14:19
-!- Saurabh7 [~Saurabh7@117.200.185.218] has quit [Ping timeout: 240 seconds]14:25
shogun-buildbotbuild #140 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20undefined%20behaviour%20analysis/builds/140  blamelist: Saurabh7 <saurabh.mahindre@gmail.com>14:27
-!- Saurabh7 [~Saurabh7@117.200.185.218] has joined #shogun14:30
shogun-buildbotbuild #138 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/138  blamelist: Saurabh7 <saurabh.mahindre@gmail.com>14:46
shogun-buildbotbuild #157 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/157  blamelist: Saurabh7 <saurabh.mahindre@gmail.com>15:34
-!- Saurabh7 [~Saurabh7@117.200.185.218] has quit [Ping timeout: 246 seconds]15:38
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.]15:38
-!- iglesiasg [~fjig@red-17.csc.kth.se] has joined #shogun16:11
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout]16:24
-!- iglesiasg [~fjig@red-17.csc.kth.se] has quit [Quit: leaving]16:59
-!- lisitsyn [~lisitsyn@80.252.20.67] has joined #shogun17:31
besser82hey lisitsyn!17:53
lisitsynbesser82: hey17:53
besser82are you proof in da pure ANSI-C ???17:53
lisitsynbesser82: well know a bit17:54
besser82perhaps you know...17:54
lisitsynbesser82: just ask and lets see if I know the answer ;)17:54
besser82is it safe to cmp some double with da memcmp?17:55
lisitsynbesser82: safe in means of?17:55
besser82like da if ( memcmp ( double a, double b, sizeof(double) )  == 0 )  // doubles are equal17:55
besser82sry, worng...17:56
lisitsynit doesn't sound like a good idea17:56
besser82like da if ( memcmp ( (void*) double a, (void*) double b, sizeof(double) )  == 0 )  // doubles are equal17:56
lisitsynI mean comparing doubles exactly is kind of bad17:56
lisitsynbesser82: apart from that it should work ok17:57
besser82i know, but i need da generic code for working with da int int64 float double ....17:57
besser82complex17:57
besser82because da FORTRAN linpack is da mut too slow17:58
besser82lisitsyn: ^^17:58
lisitsynbesser82: okay I see17:58
lisitsynbesser82: but still it could be an issue17:58
lisitsyndouble a == double b is like a rare case17:59
besser82from mem fooprint or from value?17:59
besser82footprint, i guess....17:59
lisitsynbesser82: isn't it the same? ;)17:59
besser82lisitsyn: \o/ NOOOOOOOOOOO!18:00
lisitsynbesser82: you can have similar values but different footprints that's true18:00
lisitsynbut similar not the same right?18:00
besser82lisitsyn: likely on not IEEE conform hw18:00
besser82lisitsyn: you gont me  :-P18:01
lisitsynbesser82: so what do you compare?18:01
besser82lisitsyn: ordinals of eval()18:02
lisitsynbesser82: are you sure you want to rewrite anything from linpack? ;)18:03
besser82and I need a way to do the in da ANSI-C with da generic-code18:03
lisitsynyou gonna have a lot of troubles with that I guess18:03
besser82for da f*ck**g embedded device  :-P18:03
lisitsynbesser82: macroses for the win!18:03
lisitsynbesser82: just use CMP(a,b)18:03
lisitsynCMP(double, a, b);18:04
lisitsynCMP(int, a, b);18:04
lisitsynetc18:04
lisitsynwould that work?18:04
besser82lisitsyn:  That's not in da ANSI-C  :(18:04
lisitsynbesser82: why not?18:04
lisitsyn#define CMP(type, a, b) type##cmp_implementation(a,b)18:05
lisitsynintcmp_implementation(a,b) { return a==b; }18:05
besser82that would be too much size in binary  ;)18:05
lisitsynbesser82: ehmmmm what's the other way?18:06
besser82lisitsyn: i dunno, but memcpm18:06
besser82lisitsyn: memcmp isn't picky about types18:06
lisitsynbesser82: yes it doesn't care18:06
lisitsynbesser82: but floating point?18:06
besser82lisitsyn: i'm currently checkin if it would matter ---> p_implementation(a,b)18:07
besser82lisitsyn: i'm currently checkin if it would matter ---> http://paste.fedoraproject.org/63243/8747232418:07
lisitsynbesser82: in means of == it could be ok but do you need == for doubles?18:08
lisitsynit would work for 0.0 or whatever18:08
besser82lisitsyn: yes somewhere in fortran code of linpack there is such18:08
lisitsynbut once you get 0.75 + 0.5 it is not the same as 0.5 + 0.7518:09
besser82lisitsyn: so i need to check for that in the loop, tooo....  thanks for pointing that out  ;)18:10
besser82lisitsyn: seems that on recent compiler 0.5 + 0.75 is the same as 0.75 + 0.5 ...  :D18:14
lisitsynbesser82: what about 0.76 and 0.51?18:14
besser82from memory compare18:15
besser82lemme check18:15
besser82lisitsyn: same memory18:16
besser82lemme do a loop check for vals 0.00001 .... 1000.000018:17
lisitsynbesser82: looks like it works18:19
lisitsynhttp://ideone.com/qA3Vb518:19
lisitsynbesser82: http://ideone.com/qA3Vb5 modified a bit18:20
lisitsynbesser82: IEEE floating numbers are not commutative ;)18:21
besser82lisitsyn: haha18:21
besser82lisitsyn: a != b  :-P18:21
besser82lisitsyn: should be memcmp ( ba, ab)   ;)18:21
lisitsynoops18:21
lisitsynhttp://ideone.com/qA3Vb518:22
lisitsynokay let me break it18:22
besser82lisitsyn:  same mem  :D18:22
besser82lisitsyn:  looks like compilers are today smarter that the ones from 1985  :-P18:23
lisitsynbesser82: it has nothing to do with compilers18:23
besser82lisitsyn:  but with serialization  ;)18:24
besser82lisitsyn:  and it looks recent compilers do these on real / double18:24
besser82with da automagick  ;)18:24
besser82lisitsyn:  let's try clang and see if  that f*cks it up  ;)18:25
besser82lisitsyn: nope even clang does it right  ;)18:26
besser82lisitsyn: like using serialization18:26
lisitsynbesser82: what serialization?18:26
besser82lisitsyn: making shure there is always the same mem-repres of one possible value18:27
lisitsynbesser82: well in this snippet of mine18:27
lisitsynit just computes the same value18:27
besser82lisitsyn: yes18:27
lisitsynlet me force it to compute it non-associative way somehow18:28
besser82lisitsyn: kk18:28
lisitsynbesser82: compilers are smart but IEEE stays the same18:29
besser82lisitsyn: proof?18:29
lisitsynbesser82: what makes it different? :)18:29
lisitsynit is a standard18:30
besser82lisitsyn: but then compiler are not compliant to IEEE18:31
lisitsynno it just optimizes this synthetic test18:31
besser82lisitsyn: it look like it optimizes the whole loops18:31
besser82when doing a cmp in a loop18:32
besser82like loop:  a_ptr = &b ;  a=a+i ; b=i+b; memcmp (a,b,sizeof(double))18:32
lisitsynyeah sure18:48
lisitsynbesser82: you've got to compare assembly to make sure ;)18:49
lisitsynbesser82: but the truth is the same still, comparing doubles is a bit dangerous18:49
besser82lisitsyn: yes, but i'm the search for doing all with the least possible code....18:50
besser82all == int, int64, real, double, complex, ...18:50
lisitsynbesser82: and == is faster ;)18:50
besser82lisitsyn: yes, but more different code18:50
lisitsynI am pretty sure it won't inline memcmp18:50
besser82lisitsyn: the matter is speed && minimum binsize18:51
besser82lisitsyn: otherwise i could stick with that fortran-shit18:52
lisitsynbesser82: fortan is considered faster for algebra ;)18:57
lisitsyndue to good vectorization and alignment18:58
besser82lisitsyn: really?19:02
besser82lisitsyn: but it actually takes too much mem19:02
besser82lisitsyn: for linking in gfortran19:03
besser82lisitsyn: and for f-linpack i will need f-blas, which isn't faster that say atlas or openBlas19:03
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has quit [Ping timeout: 248 seconds]21:01
besser82lisitsyn: found my solution when studying the code of openBLAS  :D21:07
besser82lisitsyn: many thanks21:07
-!- iglesiasg [~iglesiasg@c-3415e555.08-88-73746f24.cust.bredbandsbolaget.se] has joined #shogun21:10
-!- mode/#shogun [+o iglesiasg] by ChanServ21:10
-!- zxtx [~zv@129-79-241-148.dhcp-bl.indiana.edu] has joined #shogun21:39
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has quit [Ping timeout: 240 seconds]21:55
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has joined #shogun21:56
-!- FSCV [~FSCV@fixed-203-250-138.iusacell.net] has joined #shogun22:27
-!- bb [458fc8de@gateway/web/freenode/ip.69.143.200.222] has joined #shogun22:47
besser82bb: Are all your problems solved now?23:14
bbbesser82 yes.. i have got shogun installed on osx! which was my primary choice.. i did not get any time today to test applications code etc23:15
bbbesser82 will do that in the next few days..23:15
besser82bb: Nice to hear  :)23:15
bbbesser82 where are you located..23:16
besser82bb: Germany, near Hannover23:16
besser82bb: and youss23:17
besser82bb: and you?23:17
bbbesser82 okay good in DC, USA23:17
besser82bb: How is it going there?  Snow and stuff?23:18
bbit does not snow much in dc .. around 4-5 times for the year23:20
besser82bb: actually it isn't much different over here.  Temperatures are still > 0?C23:21
bbbesser82 that is good.. i used to work for a german company23:22
besser82bb: Which one?23:22
bbbesser82 qiagen sciences23:23
besser82bb: never heard of them....23:23
bbit is a biotech company23:23
besser82bb: i see  just googled them  ;)23:24
bbbesser82 what kind of work do you do..23:25
besser82bb: I'm a freelancer, focussed on riskcalculations for individual insurances23:26
besser82bb: stuff like JLo wants to insure her bum-bum or Hugh Heffner wants to insure himself not needing viagra  :-P23:26
bbbesser82 okay good... that sounds interesting23:27
besser82bb: actually just sounds...23:27
bbthese days they can grow or trip anything23:27
bbi meant trim23:28
besser82bb: from me pov it is all the same23:28
bbbesser82 i understand23:28
besser82bb: just mining huge amounts of data and applying some `magic` on that  :D23:28
bbbesser82 you will get rich someday and not show up on irc anymore23:29
besser82bb:  :-P23:29
besser82bb: That would actually not stop me from hanging out in irc  ;)23:29
besser82bb: biotech actually seems to be more interessting, btw.23:30
bbbesser82 were you in the shogun workshop in summer23:31
besser82bb: unfortunalely not,  i just hit the project ~ sept23:31
bbbesser82 okay23:32
besser82bb: but i think next year i will be there23:32
besser82bb: you, too?!?23:33
bbbesser82 good.. i would like to attend too just for fun and vacation lets see23:33
besser82bb: nice!23:33
besser82bb: but i don't know the scedule, yet...23:34
besser82bb: looks like sonney2k needs to publish...23:35
besser82bb: hope it won't interfere with some of my other visits23:35
besser82bb: like fosdem, brussels or devconf.cz in Brno,CZ23:35
bbbesser82 yeah.. i have very slim chance to make it23:36
besser82bb: Why?23:36
bbbesser82 good i dont have any such nice plans...just leaving work is hard for me23:36
besser82bb:  so much to do?23:37
besser82bb:  for me those are essential, because of catching up those new movements in some projects / products have in use on my infra23:38
bbbesser82 i am not a freelancer.. to make the big invention/bucks23:38
besser82bb  ^^23:38
besser82bb:  Those are actually part of my work23:39
besser82bb:  not directly, but inderectly, because I'm using some nifty redhat stuff on my infra  like oVirt / open Stack / openShift23:39
bbbesser82 good i do not know these terms.. may be some day when i get out of my routine i can23:40
besser82bb:  those are some middleware / cloud-oriented things23:41
besser82bb:  oVirt is low-level datacenter management23:41
besser82bb:  openStack actually make private cloud like putting some butter on your bread23:42
bbbesser82 okay.. so who are the commericial players in this23:42
besser82bb:  http://www.openstack.org/foundation/companies/23:43
besser82bb: says everything  ;)23:43
besser82bb: ...brb...  pc23:44
besser82bb: lots of global players23:44
bbbesser82 okay23:44
besser82bb re23:51
besser82bb: openShift is actually for deploying Software-as-a-Service23:51
bbbesser82 okay so do you get paid to this stuff as these are funded by these companies23:52
besser82bb: no, i just use it, it's opensource23:53
besser82bb:  because it makes my work a lot easier23:54
besser82bb:  so i can focus on important stuff...23:54
besser82bb:  which doesn't bring me $$$23:54
bbbesser82 okay may be not $$$ but $$$$$$$$$$$$$$$$$$$$$$$$23:55
besser82bb:  the stuff which brings me $$$$$$$$ isn't important, just annoying neccesity23:55
besser82bb: important stuff == making the world a better place, e.g.23:56
bbbesser82 that is good .. are you married by any chance23:57
besser82bb: nope23:57
besser82bb: have been but that's ~2 yo ago...23:58
--- Log closed Fri Dec 20 00:00:14 2013

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!