IRC logs of #shogun for Monday, 2016-02-29

« Sunday, 2016-02-28 Index Tuesday, 2016-03-01 »
--- Log opened Mon Feb 29 00:00:16 2016
@HeikoSlisitsyn: jo00:22
@HeikoSaround?00:22
lisitsynHeikoS: ja00:22
@HeikoSguten abend00:23
@HeikoSlisitsyn: schau mal hier:00:23
@HeikoShttps://github.com/shogun-toolbox/shogun/pull/302100:23
lisitsynuh00:23
lisitsynyou wrote the whole thing00:23
@HeikoSusing shogun serialisation of lists to store multiple lists and vectors00:23
@HeikoSnot general yet00:23
@HeikoSbut it works00:23
@HeikoSand can be integrated in meta language I guess00:23
@HeikoSonly thing is:00:23
@HeikoSthe stored file is a serialised list00:24
lisitsynuh you being very productive last weeks00:24
@HeikoSso if this changes, then all reference file become bad00:24
@HeikoSand the order matters, so cannot be changed after reference file has been created00:24
lisitsynHeikoS: what about two steps00:24
lisitsynfirst step we just compare outputs of different langs00:24
@HeikoSlisitsyn: ah good point00:24
lisitsynmuch later - second step - we start to store them and compare against versions00:25
@HeikoSlike all examples produce the exact same result00:25
@HeikoSlisitsyn: good idea00:25
lisitsynyes like first horizontally00:25
lisitsynacross langs00:25
lisitsynthen vertically00:25
@HeikoSmmh00:25
lisitsynacross versions00:25
@HeikoScan be done on the fly I guess00:25
@HeikoSbut the comparison against reference file is really what I wanted00:25
@HeikoSbut in any case00:25
@HeikoSgood idea00:25
lisitsynHeikoS: what's the meaning of sgmatrixobject?00:26
@HeikoSwhat do you think of file format?00:26
@HeikoSlisitsyn: it is just a SGMatrix wrapper that is a CSGObject00:26
@HeikoSso that I can put it in list00:26
@HeikoSwhich is serialisable00:26
lisitsynHeikoS: what about making it even more generic00:26
lisitsynSGObjectWrapper<T>00:26
lisitsynT m_value;00:26
@HeikoSlisitsyn: SG_ADD might cause problems then no?00:26
@HeikoSthats why I avoided00:27
@HeikoSbut in fact you are right00:27
@HeikoSone class is enough00:27
@HeikoSlet me change00:27
lisitsynwhy? SG_ADD will receive T00:27
lisitsynwhich is either00:27
lisitsynSGMatrix<something>00:27
lisitsynor SGVector or even float00:27
lisitsynHeikoS: anyway, I can go with patching generator maybe?00:30
@HeikoSyeah go ahead00:30
@HeikoSI will use this branch to draft it00:30
@HeikoSpushing in there00:30
@HeikoSyou can adjust the generator00:30
lisitsynHeikoS: ah also, NumericList is not the best name00:31
lisitsynHeikoS: what about storing them by name?00:33
lisitsynso we don't really rely on order00:33
@HeikoSalso ok00:34
@HeikoSgotta change data structure then00:34
@HeikoSok going to bed00:34
@HeikoSsee you00:35
lisitsyncu00:38
@HeikoSlisitsyn: pushed again00:41
@HeikoSbut linker error ;)00:41
@HeikoSyou can fix it00:41
@HeikoSsee you finally!!00:41
lisitsynHeikoS: ok00:44
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Ping timeout: 260 seconds]00:46
-!- travis-ci [~travis-ci@ec2-107-22-51-111.compute-1.amazonaws.com] has joined #shogun00:50
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11248116400:50
-!- travis-ci [~travis-ci@ec2-107-22-51-111.compute-1.amazonaws.com] has left #shogun []00:50
-!- travis-ci [~travis-ci@ec2-54-92-141-248.compute-1.amazonaws.com] has joined #shogun01:42
travis-ciit's Sergey Lisitsyn's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11248329301:42
-!- travis-ci [~travis-ci@ec2-54-92-141-248.compute-1.amazonaws.com] has left #shogun []01:42
-!- travis-ci [~travis-ci@ec2-54-92-141-248.compute-1.amazonaws.com] has joined #shogun02:24
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11248332802:24
-!- travis-ci [~travis-ci@ec2-54-92-141-248.compute-1.amazonaws.com] has left #shogun []02:24
-!- besser82 [~besser82@fedora/besser82] has joined #shogun08:52
-!- mode/#shogun [+o besser82] by ChanServ08:52
-!- besser82 [~besser82@fedora/besser82] has quit [Remote host closed the connection]09:44
-!- besser82 [~besser82@fedora/besser82] has joined #shogun09:45
-!- mode/#shogun [+o besser82] by ChanServ09:45
sonne|work1lisitsyn: so today accepted orgs will be announced?09:59
@besser82sonne|work1, yes, sir! =)10:22
sonne|work1besser82: hey...10:22
sonne|work1besser82: do you know more exactly when?10:22
@besser82sonne|work1, yo!  =)10:23
sonne|work1:P10:23
@besser82sonne|work1, I think around the "usual" time  ;)10:23
sonne|work117-21 hrs?10:23
@besser82sonne|work1, around 6 p.m. UTC10:23
@besser82sonne|work1, until 9 p.m. UTC10:24
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun11:20
-!- mode/#shogun [+o HeikoS] by ChanServ11:20
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit]11:23
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun11:25
-!- mode/#shogun [+o HeikoS] by ChanServ11:25
@besser82lisitsyn, yo?11:25
@besser82HeikoS, yo! =)11:25
@HeikoSbesser82: jojo whazup?11:26
@besser82HeikoS, How do we want to handle the plugin-loader library?11:26
@HeikoSbesser82: details please!11:26
@besser82HeikoS, shall I keep it in my Github-namespace or do we want to have it in Shogun-namespace?11:27
@HeikoSbesser82: oh, shogun feature branch11:27
@HeikoSor a seperate repo11:27
@besser82Question is just where to keep the repo?  ;)11:27
@HeikoSis it a whole repo?11:27
@HeikoSI see11:27
@HeikoSyeah put it in shogun11:27
@HeikoSwe have a few things in there that are ongoing11:27
@HeikoSsuch as the web211:27
@besser82It's a stand-alone lib...  Okie will init the repo in sg-namespace, then ^^11:28
@besser82HeikoS, any proposals for it's name?11:28
@besser82HeikoS, until now I called it "Dynaplugz"11:28
@HeikoSbesser82: hey its your baby, so you can choose ;)11:28
@HeikoSbesser82: haha ok maybe something else then11:28
@besser82HeikoS, okie  =)11:28
@HeikoSwhat about just sg_plugin?11:29
@HeikoSbut is it only shogun?11:29
@HeikoSor general11:29
lisitsynI am ok with dynaplugz11:29
@HeikoSok then11:29
@HeikoSgo for it11:29
lisitsyncrazier the names the better11:29
lisitsyn:D11:29
@HeikoSlisitsyn: besser82 ok, I am fine with that too11:29
@besser82lisitsyn, my thinking ^^11:29
@HeikoShaha I am just too coservative ;)11:29
@besser82HeikoS, it's pretty versatile11:29
@besser82HeikoS, so one can use the lib out of shogun for own stuff and all11:30
@HeikoSbesser82: great11:30
@besser82lib has platform-independent dynamic-loader and platform-indep. thread-safe logger11:30
@HeikoSbesser82: also put a link in the gsoc proposal for plugins11:30
@HeikoSand a few sentences11:30
@HeikoSwe have a potential student for that already11:31
@HeikoSso he can have a look11:31
@besser82HeikoS, I'll do  Greati =)11:31
@besser82HeikoS, lisitsyn: what about license?11:31
@HeikoSbesser82: the usual bsd! is my opinion11:31
@besser82HeikoS, lisitsyn: I used ASL-2 on the code so far...  Which is pretty BSD-3 clausecompat11:32
@besser82Do we have a license boiler-plate laying around somewhere?11:33
@besser82So I can use that, when initializing the repo?11:33
@besser82HeikoS, lisitsyn ^11:33
@HeikoSbesser82: yes we have11:33
lisitsynlets just refer LICENSE11:34
lisitsynit goes crazy when this license plate is bigger than source11:34
@HeikoSlisitsyn: yeah agreed11:34
@HeikoSbut lets use a proper thing for htat11:34
@HeikoSlike unified acress Shogun11:34
@HeikoSCopyright (c), Shogun Toolbox Foundation11:35
@HeikoSWritten (W) 2016 Heiko Strathmann11:35
@HeikoSDistributed under BSD3 license, see LICENSE11:35
@HeikoSsomething like this11:35
lisitsyn???11:36
lisitsynoops11:36
lisitsynyes pleas11:36
lisitsyne11:36
@besser82HeikoS, lisitsyn: That wouldn't be enough for US jurisdiction...  :(11:37
lisitsynbesser82: really?11:37
lisitsynoh11:37
@besser82HeikoS, lisitsyn: They require full license in every licensed file...11:37
@HeikoSbesser82: I still would do Sergeys suggestion11:37
lisitsynthat's crazy11:37
@HeikoSit is so annoying otherwise11:37
@HeikoSwe are not US anyways ;)11:37
lisitsynok what are dangers?11:38
@HeikoSshort header is so much better11:38
@besser82HeikoS: but people from there can sue us otherwise...  :S11:38
lisitsynsue like?11:38
@HeikoSfor not having a license? :D11:38
@besser82HeikoS: since the warranty disclaimer isn't effective11:38
@HeikoSno they cant we are not on us territory11:38
@HeikoShaha f***ers ;)11:38
@besser82HeikoS: I'm not sure about that...  We need to check the "Schengen-Abkommen" about that...  :S11:39
@HeikoSbesser82: haha11:39
@HeikoSno but seriously, I think lisitsyn is very right with his point11:39
@HeikoSlets have a 3 liner in every file11:39
@HeikoS1.) Copyright Shogun Foundation11:40
@HeikoS2.) Authors of this file11:40
lisitsynHeikoS: lets do this if we find any other example of such thing11:40
lisitsynis anybody doing that?11:40
@HeikoS3.) Licensed BSD3, see LICENSE for details11:40
@HeikoSlisitsyn: I am sure11:40
@HeikoShttps://github.com/SheffieldML/GPy/blob/master/GPy/kern/_src/ODE_UY.py11:40
@HeikoSneil lawrence does it this way11:40
@HeikoSthey dont even have authors in the files11:41
@HeikoSjust reference to AUTHORS and LICENSE11:41
lisitsynoh that's not the best example11:41
lisitsynsomething widely used ;)11:41
@besser82HeikoS, lisitsyn: That doesn't imply it is safe to do so...11:41
@HeikoSthe guy is pretty big and active in opens source11:42
@HeikoSbut I am sure there is another11:42
@HeikoSgo find it ;)11:42
lisitsynapache spark - full license11:42
lisitsynsklearn - Authors: ... License: ....11:43
@HeikoSsecond then!11:43
@HeikoSlisitsyn: c++ question11:44
@HeikoSlisitsyn: I have non-template class11:44
@HeikoSand want a template member function11:44
@HeikoSI get undefined references in linking, what is wrong?11:45
lisitsynHeikoS: to this function?11:45
@HeikoSyes11:45
lisitsynHeikoS: where is it defined?11:45
@HeikoStemplate<class T> void append(T value);11:45
@HeikoSin the class11:46
@HeikoSthat is non template11:46
lisitsynwhere are the impls?11:46
lisitsynin the .cpp?11:46
@HeikoSbesser82: lets do sklearn style11:46
@HeikoSyes11:46
@HeikoSlisitsyn: yes11:46
@HeikoSlisitsyn:11:46
@HeikoStemplate <class T> void CNumericList::append(T value)11:46
@HeikoS{11:46
@HeikoSthis->append_element(new CSGObjectWrapper<T>(value));11:46
@HeikoS}11:46
lisitsynHeikoS: then either this cpp is not being compiled11:46
@HeikoSit is11:47
lisitsynwait11:47
lisitsyndo you implement it in .cpp?11:47
@HeikoSyes11:47
@HeikoSah11:47
lisitsynno you can't11:47
@HeikoSthats the problem right11:47
@HeikoSwith templates11:47
@HeikoSI forgot11:47
lisitsynyou have to put your impl into .h11:47
@HeikoSits a while since I wrote c++11:47
@besser82HeikoS, lisitsyn: I'm still not convinced...   I'll ask the lawyers over at RedHat, what they think about that...  :/11:48
lisitsynbesser82: apparently no one sued sklearn while they are much more popular11:49
@HeikoSbesser82: I think this is merely a detail11:50
@HeikoSbut ok11:50
lisitsynHeikoS: there is a point in that11:50
lisitsynspeaking of file11:50
lisitsynthere is no such LICENSE thing around11:50
@HeikoSlisitsyn: maybe lets put one :)11:51
lisitsynit is something 'not complete'11:51
lisitsynno I mean file itself is self-fulfilling this way11:51
lisitsynbut if we refer LICENSE11:51
lisitsynI am curious if we can refer LICENSE + its md5 hash11:52
lisitsyn:)11:52
lisitsynbesser82: does it sound crazy for you?11:52
lisitsynHeikoS: but apparently it is ok to refer some law stuff without copying it altogether11:53
@HeikoSok then11:55
@HeikoSlets do it11:55
@besser82lisitsyn, HeikoS: just had quick chat to Tom Callaway from RH12:03
@HeikoSbesser82: and?12:04
@besser82lisitsyn, HeikoS: Every source-file *MUST* contain the license individually, since the license itself demands that by itself12:04
@besser82"Redistributions of source code must retain the above copyright notice,12:04
@besser82     this list of conditions and the following disclaimer."12:04
@HeikoSbesser82: well ....12:05
@HeikoSok then12:05
@HeikoSput it in :)12:05
@besser82HeikoS, allrighty12:06
@HeikoSbesser82:  btw12:07
@HeikoSI am not sure I agree still12:07
@HeikoSsee e.g. here http://stackoverflow.com/questions/845895/putting-license-in-each-code-file12:07
@HeikoSaccording to this, copyright needs to be in every file12:07
@HeikoSbut not full license12:07
@HeikoSjust name of license12:07
@besser82HeikoS, they are mostly discussing the GPL, where in fact the Copyright is enough, since the license itself is viral12:10
@HeikoSok12:10
@HeikoSlisitsyn: more c++ questions12:10
@HeikoStemplate<class T> bool append_wrapped_element(T data)12:10
@HeikoS{12:10
@HeikoSreturn this->append_element(new CSGObjectWrapper<T>(data));12:10
@HeikoS}12:10
@HeikoSI put this in CList12:10
@HeikoSbut now I get undefined references again12:10
@HeikoSundefined reference to `shogun::CSGObjectWrapper<shogun::SGMatrix<double> >::CSGObjectWrapper(shogun::SGMatrix<double>)'12:11
lisitsynHeikoS: CSGObjectWrapper constructor is not in .h?12:12
@besser82HeikoS, there is no constructor ;)12:12
@HeikoSthanks! :)12:12
@HeikoSah man12:13
@HeikoSmy brain is a cheese12:13
@HeikoSthings drop out12:13
@HeikoSso quickly12:13
@HeikoScan only remember math ;)12:13
@besser82HeikoS, hehe, lol  =)  Do we have a general mailing-address for Shogun e.V.?12:16
@HeikoSyes we do12:17
@HeikoSgotta find it12:18
@HeikoSits in publicity git I think12:18
@HeikoSwill send later12:18
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Ping timeout: 248 seconds]12:22
lisitsyndynaplugz!12:25
@wiking:>12:25
@besser82lisitsyn, coming soon XD12:37
@besser82wiking, do you know we have common email-addy for shogun?12:38
lisitsynshogun-team12:38
@besser82wiking, like info@shogun-toolbox.org?12:39
@besser82lisitsyn, ^12:39
@besser82lisitsyn, is the mailing list open without subscription?12:39
lisitsynbesser82: shogun-team@shogun-toolbox.org12:39
@besser82lisitsyn, okie, will use that in the LICENSE.txt  :P12:39
lisitsynah12:40
lisitsynyeah12:40
lisitsynok12:40
@besser82lisitsyn, hope this is open for public  :P12:40
lisitsynbesser82: for mails outside?12:40
lisitsynshould be12:40
@besser82yes, like someone can use it as a contact addy12:40
-!- HeikoS [~heiko@host86-160-51-46.range86-160.btcentralplus.com] has joined #shogun12:41
-!- mode/#shogun [+o HeikoS] by ChanServ12:41
-!- lisitsyn [~lisitsyn@37.139.2.75] has left #shogun []12:41
-!- lisitsyn [~lisitsyn@37.139.2.75] has joined #shogun12:42
@HeikoSlisitsyn: so the storing works12:42
@HeikoSlisitsyn: I also had another thought about the two levels of testing you suggested12:43
@HeikoSlisitsyn: not sure we need the comparison across all target languages: if we compare against referencee file, we get that automatically12:43
@HeikoSlisitsyn: thoughts?12:43
@HeikoSlisitsyn: re the map data structure, this might be more work as CMap is not serializable12:43
@HeikoSlisitsyn: and not idea how much work it is to make it12:44
@besser82lisitsyn, HeikoS:  pushed the bsd3-templates to publicity-repo:  https://github.com/shogun-toolbox/shogun-publicity/tree/master/bsd3-license12:48
@HeikoSbesser82:  nice!12:50
@HeikoSbesser82: any idea how to approach changing the licenses of all files?12:50
@HeikoSbesser82: might have to update this:12:50
@HeikoShttps://github.com/shogun-toolbox/shogun/issues/295812:50
@HeikoSbesser82: would be cool to do this soon and before next 5.0 release12:51
@besser82HeikoS, will try to find a way to shift them without much problems...12:51
@HeikoSbesser82: great!12:51
@HeikoSlet me know12:51
@HeikoSbesser82: whats with the encoding in the license file?13:23
@besser82HeikoS, that's to make sure the compiler / interpreter uses the correct file-encoding13:24
@besser82HeikoS, I have seen python-stuff going boom, when python ran on ASCII and the file was UTF-813:25
@besser82HeikoS, same with some compilers on Windows13:25
@besser82HeikoS, it also make sure, the files keeps the correct encoding, when it is edited by people having a strange enconding in their env13:26
@HeikoSok cool13:27
@besser82HeikoS, those two lines make sure 99.8% of possible sh*t happening is prevented13:27
@besser82^^13:27
@HeikoSok great then :)13:30
@HeikoSused it13:30
@HeikoSbesser82: looking forward to plugin13:30
@HeikoSever tried to modify List.h?13:30
@HeikoShave to re-compile all of shogun to test13:30
@HeikoSthe changes13:30
-!- HeikoS [~heiko@host86-160-51-46.range86-160.btcentralplus.com] has quit [Quit: Leaving.]13:34
-!- travis-ci [~travis-ci@ec2-54-147-132-119.compute-1.amazonaws.com] has joined #shogun14:29
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11257333014:29
-!- travis-ci [~travis-ci@ec2-54-147-132-119.compute-1.amazonaws.com] has left #shogun []14:29
-!- HeikoS [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has joined #shogun14:41
-!- mode/#shogun [+o HeikoS] by ChanServ14:41
@wikingbesser82: around?14:51
@besser82wiking, yo14:52
@wikingso i've just got back to my libarchive feature branch14:53
@wikingbut14:53
@wikinghave you been working with libarchive before?14:53
@wikingmy problem is seeking14:53
@wikingbecause we usually process files in 2 pass14:53
@wikingmaybe we should remove that :)14:54
@besser82Hmmm...  Never directly used libarchive before...  :S  But I think there might be some good good-examples in MATE-archiver...15:04
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has joined #shogun15:04
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11257407215:04
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has left #shogun []15:04
@besser82https://github.com/mate-desktop/engrampa15:05
@besser82wiking, ^15:05
-!- humility [uid138660@gateway/web/irccloud.com/x-ncecteifyfsjaezg] has joined #shogun15:05
@besser82Maybe with libarchive double-pass processing isn't needed anymore...15:05
@HeikoSbesser82:  what is this?15:13
@besser82HeikoS, which? the fail on Travis or libarchive?15:14
@wikingbesser82: the thing is that currently we do double pass in order to know what size of an SGMatrix we should allocate15:37
@besser82wiking, for the file? Or it's contents?15:37
sonne|work1and that double pass is extremely cheap if the file fits in memory :)15:37
@wikingcontent15:43
@wikingsonne|work1: the problem here is that with libarchive seeking is not so straightforward :)15:44
sonne|work1wiking well there are 2 options: either we open twice and read twice or we read once and re-alloc as needed...15:44
@wikingsonne|work1: heheh yeah i know :)15:45
@wikingi was just wondering if maaaaybe somebody knwos a good trick for15:45
@wikinglibarchive seeking15:45
@wikingso that i dont have to refactor our code :)15:45
@wikingas other than that the code is ready :)15:45
sonne|work1wiking: the lazy guy in you is speaking :P15:46
@wikingindeed15:46
@wiking:)))15:46
@wikingbtw the circular buffer has a bug :)15:47
sonne|work1wiking: didn't libarchive work with FD's instead of FILE?15:47
sonne|work1wiking: did you solve this?15:47
sonne|work1wiking: circular* - what bug?15:48
@wikingyeah that's not a problem (libarchive)15:48
@wikingbbl15:48
-!- HeikoS [~heiko@pat-191-250.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.]15:56
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has joined #shogun15:59
-!- mode/#shogun [+o HeikoS] by ChanServ15:59
@wikingsonne|work1: so the circular buffer (ringbuffer) has a little bug16:07
@wiking:)16:07
sonne|work1and which?16:07
@wikingahhaha16:08
@wikingit actually has more than one problem16:08
@wikinga) it's not thread safe16:08
@wikingb) if you use that and process a file bigger than the buffer16:08
@wikingit's counter is not correct16:09
sonne|work1sigh16:09
@wikingwhy wasn't this implemented thread safe? :)16:09
sonne|work1so we need that + tests for this or an alternative impl.16:09
@wikingyeah16:09
@wikingcan i just use c++11 for this?16:10
@wikingit'd be fairly easy to do this whole story using std::atomic etc16:10
@wikingbtw shouldn't we actually as Heiko say have Eigen as a required library16:10
@wikingas well as c++11 ?16:10
@wikingor we wanna work with old compilers?16:11
@wikingHeikoS sonne|work1 besser82 lisitsyn ^^^ ? :)16:12
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has joined #shogun16:13
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11259103816:13
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has left #shogun []16:13
sonne|work1wiking: ask the boss16:13
@wikingwho's the boss? :)16:13
* sonne|work1 is pointing fingers to HeikoS16:13
@wikingHeikoS: BOSS!16:13
@besser82wiking, I'm fine with support for let's say gcc >= 4.4, so we can build at least on RHEL616:23
@besser82wiking, since that will be actively supported until 202216:24
@besser82wiking, but we should step up on CMake >= 3.116:24
@besser82wiking, so we can use CMake's internal target_compile_features16:25
@besser82wiking, https://cmake.org/cmake/help/v3.1/prop_gbl/CMAKE_CXX_KNOWN_FEATURES.html#prop_gbl:CMAKE_CXX_KNOWN_FEATURES16:26
@wikinggcc 4.4 has c++11 or only c++0x?16:29
@wikingoh ok16:30
@wikingi see16:30
@wikinghttps://gcc.gnu.org/projects/cxx0x.html16:30
@wiking:)16:30
@wikingit has enough stuff :)16:30
@HeikoShahaha16:33
@HeikoSwiking, sonne|work1 I am all for requiring eigen3 (as can be bundled) and requiring c++11 (we did already I think)16:35
@HeikoSkeeps us moving16:36
@HeikoSand the old compiler ppl have to use old version16:36
@wikingmmm c++11 we dont require16:38
@wikingwe just use it if we can16:38
@HeikoSlisitsyn: https://github.com/shogun-toolbox/shogun/pull/3021/files16:39
@wikingbut still provide the alternative if needed16:39
lisitsynHeikoS: good16:39
@HeikoSwiking: I dont really like all these #IFDEFS floating around and breaking the build, as for eigen316:39
@HeikoSwiking: but you decide16:39
@HeikoSlisitsyn: this works16:39
@HeikoSlisitsyn: stores the things named16:39
lisitsyneigen is not optional16:39
lisitsynit is header only16:39
lisitsynlets just always use specific eigen version16:39
@HeikoSlisitsyn: thats why I think we should remoive the #ifdef HAVE_EIGEN3 stuff16:40
lisitsynlike ignoring system one by default16:40
@HeikoSlisitsyn: agreed16:40
@HeikoSwiking: ^16:40
lisitsynHeikoS: I had some prototype before I went to sleep yesterday16:40
@HeikoSlisitsyn: cool16:40
@HeikoSlisitsyn: shall we continue to do this in the integration branch?16:40
lisitsynHeikoS: I create virtual 'Init' expressions16:40
lisitsynno sorry, I use Init16:41
lisitsynand then create MethodCall16:41
lisitsynso in the end I just put all the things16:41
@wikingHeikoS: ok so i'll go ahead and fix with c++11 the circular buffer16:44
@wikingdunno what to do with libarchive for now16:44
@HeikoSwiking: cool!16:44
@HeikoSlisitsyn: nice, let me know when you pushed it :)16:45
@HeikoSwiking: okok16:45
@wikinganybody knows how to get the open mode of a FILE* f?16:45
@HeikoSwiking: thats needed for the CSV right?16:45
@wikingHeikoS: yes16:45
@wikingbtw it's used in LibSVMFile.h16:46
@wikingas well16:46
@wikingso i suppose that's broken as well :)16:46
@HeikoSwiking: so lisitsyn and me are working on this meta-example wide integration testing16:46
@HeikoSthat is, integration tests for numerical results for all target languages16:46
@HeikoSwhich first ensures that they all give the same, and second avoids errors so slip if not covered by unit tests16:47
@wikingyep16:47
@wikingthat's sounds great!16:47
@HeikoSalmost there16:47
@HeikoSwiking: we use shogun serialisation though16:47
@HeikoShttps://github.com/shogun-toolbox/shogun/pull/3021/files16:47
@HeikoSascii16:47
@wikingthat's ok16:48
@besser82wiking, when we are pluginizing and all:  Can we reconfigure the buildbot?17:14
@besser82Since I want to save the cpu of the buildbot and split out builds to OBS17:15
@besser82It's a public accessible thing and we can do amazing CI on it, like automatically rebuild dependencies and everything17:16
@besser82and we can ship our own debian / ubuntu / SUSE / Arch repos there17:16
@HeikoSbesser82: that sounds cool17:18
@besser82HeikoS, thought so, too  =)  Free and useful  :D17:19
@wikingbesser82: yeah sure of course17:19
@besser82wiking, great!  So let me finish the dynaplugz lib and prepare all for it to be the first thing built on the OBS-service  =)17:20
@wikingmmm has it been pushed?17:20
@besser82wiking, not fully yet currently just enough to build an empty lib in the repo17:21
@besser82wiking, I want to have CMake fully working17:21
@wikingoh ok i see17:21
@wikingkk17:21
@besser82when I'm done with all, I'll give a shout  ^^17:21
@wikingping me anytime17:22
@besser82will do so! Thank you =)17:22
@wikinguntil then i'll try to work on the ringbuffer and the libarchive stuff17:22
@besser82great!17:22
@besser82btw...  can we teach buildbot to create and push git-tags?17:23
@wikingyes17:26
@besser82Yay! That will come in handy then, too17:27
-!- humility [uid138660@gateway/web/irccloud.com/x-ncecteifyfsjaezg] has quit [Quit: Connection closed for inactivity]17:37
-!- travis-ci [~travis-ci@ec2-107-22-55-79.compute-1.amazonaws.com] has joined #shogun17:52
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11259435517:52
-!- travis-ci [~travis-ci@ec2-107-22-55-79.compute-1.amazonaws.com] has left #shogun []17:52
@HeikoSlisitsyn:  jo18:12
shogun-buildbotbuild #3572 of deb1 - libshogun is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/357218:46
-!- travis-ci [~travis-ci@ec2-107-22-55-79.compute-1.amazonaws.com] has joined #shogun19:02
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11261658519:02
-!- travis-ci [~travis-ci@ec2-107-22-55-79.compute-1.amazonaws.com] has left #shogun []19:02
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun19:02
shogun-buildbotbuild #329 of trusty - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/329  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>19:14
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Ping timeout: 252 seconds]19:15
shogun-buildbotbuild #2767 of deb3 - modular_interfaces is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/276719:18
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun19:20
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Ping timeout: 252 seconds]19:25
shogun-buildbotbuild #329 of CentOS7 - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/CentOS7%20-%20libshogun/builds/329  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>19:33
shogun-buildbotbuild #446 of trusty - libshogun - viennacl is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/446  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>19:40
@HeikoSlisitsyn: around=?19:47
@HeikoSlisitsyn:  see the latest version: https://github.com/shogun-toolbox/shogun/pull/302119:48
@HeikoSlisitsyn: the last part of the .sg file has to be autogenerated. The ball is with you19:49
@HeikoSlisitsyn: once this works, we can write a python script to compare results19:49
@HeikoSintegration will be a bit messy19:49
@HeikoSwiking: latest class list is 40419:56
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has joined #shogun20:06
sanujHeikoS: Congrats!!20:06
sanujGSoC accepted :)20:07
@HeikoSsanuj: ha really?20:07
@HeikoStotally forgot about the deadline20:07
@HeikoSseems like good news then :)20:07
sanujyeah see20:07
sanujHeikoS: https://summerofcode.withgoogle.com/organizations/5104067932061696/20:07
sanujIt was announced just now.20:07
@HeikoSwhoooo!20:07
@HeikoS :) :) :)20:07
@HeikoSgreat news20:08
sanujGreat :)20:08
sanujIndeed20:08
@HeikoSthanks for notifying us :)20:08
sanujmy pleasure20:09
sanujHeikoS: can't wait to start contributing back again20:09
sanujhave an exam tomorrow :P20:09
@HeikoSsanuj: oh good luck for that20:10
rcurtinHeikoS: congratulations!  mlpack got in too so maybe we can have an ML summit at the mentor summit :)20:10
@HeikoSsanuj: yeah looking forward to your patches20:10
@HeikoSrcurtin: thanks20:10
@HeikoSrcurtin: and congrats to you too! :)20:10
@HeikoSrcurtin: could even think again about that joint project now ;)20:10
@HeikoSrcurtin: definitely hang out at the summit20:10
rcurtinalso Orange is in this year20:12
rcurtinbut no other ML libraries (other than R) as far as I can see20:13
@HeikoSrcurtin: ml is on the hype so there is room for many libs20:19
@HeikoSsurprised all the new projects are not in20:19
rcurtinyeah20:19
@HeikoStensorflow20:19
@HeikoSsklearn20:19
@HeikoStorch20:19
@HeikoSautograd20:19
@HeikoSetc etc20:19
rcurtintensorflow at least I think already has google's support :)20:19
@HeikoStrue20:19
rcurtinI would not be surprised if many of those organizations did not apply20:19
rcurtinlots of ML people don't know about GSoC, I'm constantly explaining what it is to people at conferences20:20
sanujHeikoS: tensorflow and torch probably don't require funding20:20
sanujrcurtin: gsoc is very popular in india.20:20
rcurtinI met the caffe maintainer, Evan something I think was his name, he wasn't familiar with GSoC and didn't seem to be very familiar with the open-source world20:20
sanujamong the students atleast20:20
rcurtinsanuj: absolutely, that's where the majority of the applications I see come from :)20:20
rcurtinbut most of the researchers I work with aren't familiar with it20:21
rcurtinso maybe lots of these organizations just aren't aware of the opportunity, or maybe they don't have people with enough spare cycles to do mentoring20:21
sanujyeah.20:21
rcurtinbut I dunno, I am not them, I'm just idly guessing :)20:21
lisitsynare we accepted?20:21
lisitsynuhm interesting20:21
lisitsyn:D20:21
sanujlisitsyn: yeah20:21
@HeikoSlisitsyn: yeah :D20:21
@HeikoSI gotta leave now20:22
@HeikoSsee you guys later20:22
sanujcool20:22
sanujgoodnight20:22
-!- mode/#shogun [+o sonney2k] by ChanServ20:23
@besser82sonney2k, He's alive! \o/20:26
-!- HeikoS [~heiko@untrust-out.swc.ucl.ac.uk] has quit [Ping timeout: 246 seconds]20:26
@besser82sonney2k, You're preparing the invitations for GSoc2016-mentors?  :P20:27
shogun-buildbotbuild #330 of trusty - libshogun is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/33020:34
shogun-buildbotbuild #447 of trusty - libshogun - viennacl is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/44720:35
-!- ebadawy [~ebadawy@197.163.32.61] has joined #shogun20:44
@besser82lisitsyn, yo?20:51
lisitsynbesser82: hey20:51
@besser82lisitsyn, you know any generic way <tm> to list the files in a dir?20:52
@besser82using C++20:52
@besser82without using POSIX or Windows specific system-calls?20:52
lisitsynbesser82: C++17 haha20:53
lisitsynhttp://en.cppreference.com/w/cpp/experimental/fs20:53
@besser82lisitsyn, okie, so boost::filesystem  :P20:53
lisitsynyeah probably20:53
-!- nelson-liu [~nelsonliu@nelsons-macbook-pro.dyn.cs.washington.edu] has joined #shogun20:58
-!- sanuj [0e8bc402@gateway/web/freenode/ip.14.139.196.2] has quit [Quit: Page closed]21:00
nelson-liuhi! I was checking out shogun on gsoc, and noticed that you mentioned python in your description. However, I looked at your "Getting Involved" page on github, and it seemed to place more of an emphasis on c++21:01
nelson-liuI'm pretty experienced with general ML theory and such, and contribute a lot to scikit-learn. Was just wondering if I'd be able to contribute substantially if I'm completely unfamiliar with cpp21:01
-!- mlenthu [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun21:01
mlenthuHi, I am Shashank. I am a MS By Research, CSE student at IIIT Hyderabad working in applications of Deep Learning in NLP. I came across your organization's profile in GOSC 2016 list. There are some very exciting Deep Learning projects in description. I want to contribute in some project. How should I start?21:03
@besser82lisitsyn, any objection, when I'll make Boost a dependency of Dynaplugz?21:12
@besser82lisitsyn, I mean, every distro and Cygwin/MinGW ship some version of it...21:12
-!- nelson-liu [~nelsonliu@nelsons-macbook-pro.dyn.cs.washington.edu] has quit [Remote host closed the connection]21:16
@besser82brew has it, too...21:18
@wikingso what?21:22
@wikingwhat happened? :)21:22
@wikingreading logs21:22
@wiking:)21:22
@wiking:>>>>>>>>>>>>21:23
@wikingoh man21:23
@wikinghere we go!21:23
-!- nelson-liu [~nelsonliu@D-173-250-146-196.dhcp4.washington.edu] has joined #shogun21:25
@wikingbtw somebody who has access to the twitter account retweet my status or send out that we've got accepted21:27
-!- nelson-liu [~nelsonliu@D-173-250-146-196.dhcp4.washington.edu] has quit [Remote host closed the connection]21:30
-!- lordon [~Thunderbi@2a02:810a:13c0:8494:584a:5b07:c89:caa6] has joined #shogun21:35
@wikingchannel infestation official has begun21:38
@wikingwelcome all! :)21:38
@wikingsonne|work1: around?21:45
@sonney2kwiking, no22:13
shogun-buildbotbuild #586 of debian wheezy - memcheck is complete: Failure [failed memory check]  Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/586  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>22:24
@wiking:>22:31
@wikingok22:31
-!- nelson-liu [~nelsonliu@D-69-91-185-195.dhcp4.washington.edu] has joined #shogun22:34
nelson-liubtw, where are most of the devs located?22:34
lisitsynrandom points22:36
@wiking:)22:37
@wikingbesser82: no boost22:37
lisitsynremember no line in boost22:38
@besser82lisitsyn, wiking:  Y?22:39
@besser82lisitsyn, wiking: Boost::filesystem is really fun22:39
@wikingno22:39
lisitsynjust like any other boost library22:39
lisitsyn;)22:39
@wikingno boost22:40
@wikinglisting ifles22:40
@wikinglisting files is easy22:40
@wikingno need boost fs for that22:40
@besser82wiking, yes, but not if you need to be platform-indep...22:40
@besser82wiking, I know how to do that in POSIX, too...  But POSIX != Windows22:40
@wikingyes22:41
@wikingyou write ifdef22:41
@wikingnot that complicated22:41
@wikingand btw22:41
@wikingshogun does not run on window22:41
@wiking*windows22:41
@wikingso that'll take some loooonger time22:41
@besser82wiking, when we're done it will :P22:41
@wikingto get that running on none-posix place22:41
@wikinghahahha22:41
@wikingthere are tooo many22:41
@wikingcode that doesn't compile with vc++22:42
@besser82wiking, and with plugins we can stomp them one after another in little time  :P22:42
lisitsyn vc++ is not a C++ compiler22:43
@besser82lisitsyn, I know...22:43
@wikingbesser82: have u tried22:43
@wikingto compile the code in shogun22:43
lisitsynit is some magic world22:43
@wikingwith vc++22:43
@wiking?22:43
@wikingbecause i did22:43
@wikingand trust me22:43
@wikingmaking plugins22:43
@wikingwill not get you anywhere22:43
@besser82wiking, not yet...22:43
@wikingin that effort22:44
@wikingi did some patches and fixes22:44
@wikingbut there's still a significant amount of code that needs 'fixing'22:44
@besser82wiking, there are other compilers for windows than just VC++22:44
@wikingyes22:44
@wikingwith mingw it works22:45
@wiking:)22:45
@wikingdunno if there's still borland22:45
@wikingc++ ? :))))22:45
@besser82wiking, If I remeber correctly there are even more than Borland22:45
@wikingmmm22:45
@wikingok please let's not get into nitpicking22:45
@wikingif you want windows support22:45
@wikingyou better have vc++ support22:46
@wikingand that is a significant effort22:46
@besser82wiking, that's true...22:46
@wikingbecause there's a lot of stuff in shogun22:46
@wikingthat is none-vc++ ready22:46
@wikingand no22:46
@wikingno boost dependency22:46
@wikingthat has been a rule for a loooooong time now22:46
@wiking:)22:46
lisitsynwhat about no exceptions ;)22:47
@wiking:)22:47
@wikingfuck try catch22:47
@wiking:D22:47
@wikingthat is for pussies :)))22:47
lisitsyngoogle uses this rule22:47
lisitsynO_o_O22:47
@wikingin java?22:47
@wikingbecause it makes sense in jvm22:48
lisitsynhmm not in java I think hah22:48
@wikinggetting stack in jvm i actually a horror22:48
lisitsynjava without exceptions is something interesting22:48
@wikingso it actually makes sense not to have exceptions in jvm22:48
@besser82wiking, What is the real problem with boost?22:48
@wikingbesser82: too big22:48
@wikingtoo bloated22:48
@wikingnot necessary22:48
@wikingyou can do fs22:48
@wikingnatievely22:48
lisitsynwiking: well it is a good practice to avoid checked exceptions22:48
@besser82wiking, boost::filesystem is exactly what c++17 will be, same author22:49
lisitsynhave seen production code doing checked->unchecked conversion22:49
@wikingbesser82: will be22:49
@wikinguntil then22:49
@wikingthere are many things22:49
@wikingthat was boost22:49
@besser82guys c'mon...  :P22:49
@wikingthat became c++1122:49
@wikingno come on22:49
@wikingthis has been like this22:49
@wikingfor a long time22:49
@wikingworked out just fine22:49
@wikingwrite that code22:49
@wikingor ask somebody to write it22:49
@besser82I have POSIX for that here...22:50
@wikinggood22:50
@wikinguse that22:50
@besser82ready code, but it won't do on win...  :(22:50
@wikinggood enough22:50
@wikingnobody case for that22:50
@wikingif we have a gsoc for win this summer22:50
@wikingthen we do it22:50
@wikingif not22:50
@wikingthen we will not22:50
@wikingagain aas said there are many more problems with win support for shogun22:51
@wikingthe listing of files and dirs is the least of them all22:51
-!- lordon [~Thunderbi@2a02:810a:13c0:8494:584a:5b07:c89:caa6] has quit [Quit: lordon]23:05
@besser82okie23:08
shogun-buildbotbuild #330 of CentOS7 - libshogun is complete: Success [build successful]  Build details are at http://buildbot.shogun-toolbox.org/builders/CentOS7%20-%20libshogun/builds/33023:09
-!- nelson-liu [~nelsonliu@D-69-91-185-195.dhcp4.washington.edu] has quit [Remote host closed the connection]23:18
-!- alaaelnouby [292f264e@gateway/web/freenode/ip.41.47.38.78] has joined #shogun23:23
-!- alaaelnouby [292f264e@gateway/web/freenode/ip.41.47.38.78] has quit [Client Quit]23:26
-!- ghoshbinayak [~AndChat29@117.251.210.78] has joined #shogun23:39
shogun-buildbotbuild #1121 of precise - libshogun is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1121  blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>23:39
-!- ghoshbinayak [~AndChat29@117.251.210.78] has quit [Read error: Connection reset by peer]23:55
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has joined #shogun23:56
travis-ciit's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/11266005523:56
-!- travis-ci [~travis-ci@ec2-54-147-13-163.compute-1.amazonaws.com] has left #shogun []23:56
-!- psitae [a5e6e0df@gateway/web/freenode/ip.165.230.224.223] has joined #shogun23:56
--- Log closed Tue Mar 01 00:00:18 2016
« Sunday, 2016-02-28 Index Tuesday, 2016-03-01 »

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