Open in new window / Try shogun cloud
--- Log opened Fri Oct 04 00:00:26 2013
-!- lisitsyn [] has quit [Quit: Leaving.]01:34
shogun-buildbotbuild #478 of nightly_all is complete: Failure [failed test]  Build details are at
-!- sonne|osx_ [] has joined #shogun03:49
-!- sonne|osx [] has quit [Read error: Operation timed out]03:50
-!- sonne|osx_ is now known as sonne|osx03:50
shogun-buildbotbuild #562 of nightly_default is complete: Failure [failed test]  Build details are at
-!- iglesiasg [] has joined #shogun04:25
-!- mode/#shogun [+o iglesiasg] by ChanServ04:25
-!- hushell [] has joined #shogun05:04
hushelliglesiasg: hey! still awake?05:04
@iglesiasghushell, well already more than still hehe05:05
@iglesiasghushell, I woke up suddenly in the middle of the night05:05
hushelliglesiasg: sometimes I do like you :)05:05
hushelliglesiasg: I feel better at night05:06
hushellMSG_DEBUG is the same as SG_DEBUG?05:06
@iglesiasgyeah sure05:06
@iglesiasgto activate the prints made inside SG_DEBUG05:06
@iglesiasgone needs to do this shogun_object->io->set_loglevel(MSG_DEBUG)05:07
hushelliglesiasg: I need to include SGIO.h, isn't it?05:08
@iglesiasghushell, where exactly?05:08
hushelliglesiasg: sometimes I got error, say SG_DEBUG is not defined05:08
@iglesiasghushell, aham. I don't remember exactly which header is it bug I think SGIO.h is not necessary05:09
hushellbut SG_SPRINT is safe everytime05:09
@iglesiasgprobably shogun/lib/config.h05:09
hushellhmm, so your point is change every SG_SPRINT to SG_DEBUG?05:10
@iglesiasghushell, well I think that maybe it is not necessary05:10
@iglesiasgsince we are going to keep verbose05:10
@iglesiasgas it does a little bit more than SG_DEBUG05:10
@iglesiasgmaybe it is just fine to keep them with SG_SPRINT or SG_PRINT05:11
hushelliglesiasg: I just try to imitate what Michal did for BMRMs05:11
@iglesiasghushell, sounds reasonable05:11
hushelliglesiasg: so now we have helper member for every SOSVM solvers05:12
@iglesiasghushell, nice!05:12
@iglesiasghushell, are all the objectives  comparable then?05:12
hushelliglesiasg: I think so, they computed by the same function05:13
@iglesiasghushell, that sounds neat05:13
@iglesiasghushell, maybe we could try that for some data set05:13
@iglesiasgplot the objective obtained with different solvers05:13
hushellyeah, sounds great!05:13
hushellI still haven't implement the duality gap function, I'll do it this weekend05:14
hushellokay, let me update the python example05:14
@iglesiasghushell, what is the interpretation of the duality gap?05:14
hushelliglesiasg: The simplest one is primal-dual, but we have different way to compute dual, like Patrick's BCFW paper, they has show a special case of Fenchel duality gap05:17
hushelliglesiasg: I am not expert on this. I need to check in detail before implementing useful statistics05:18
@iglesiasgI see, I have no idea about what the duality gap is really05:19
hushelliglesiasg: The primal solution - lagrangian dual solution is very intuitive, the Fenchel duality is a bit complicate05:22
hushelliglesiasg: Patrick's friend Julien Mairal is an expert on this, maybe try to read his papers :)05:23
hushelliglesiasg: okay, updated the minor issue05:26
@iglesiasgall right, off now05:44
@iglesiasgsee you later05:44
-!- iglesiasg [] has quit [Quit: Ex-Chat]05:44
-!- zxtx [] has quit [Read error: Operation timed out]06:40
-!- hushell [] has quit [Quit: WeeChat 0.3.7]06:50
-!- HeikoS [] has quit [Ping timeout: 240 seconds]08:09
-!- zxtx [] has joined #shogun09:17
wikingmeeting @ 10?09:36
-!- besser82 [~besser82@fedora/besser82] has joined #shogun09:41
-!- lisitsyn- [~lisitsyn@] has joined #shogun09:49
lisitsyn-hey guys09:51
lisitsyn-im on my way still so lets start 10-15 mins later09:51
lisitsyn-so heiko and me could participate09:51
lisitsyn-sorry for that09:52
-!- lisitsyn- [~lisitsyn@] has quit [Remote host closed the connection]09:52
-!- iglesiasg [] has joined #shogun10:02
-!- mode/#shogun [+o iglesiasg] by ChanServ10:03
@iglesiasghi guys, sorry I am late10:03
@iglesiasgah ok, it seems I am fine :)10:03
besser82doesn't meeting start @ 10:00 UTC?!?10:11
@iglesiasgbesser82, I think it was supposed to start 10 minutes ago10:11
@iglesiasgbut we will do it like in 5 minutes10:11
besser82iglesiasg: It should be in ~ 2h afaik10:11
@iglesiasgbesser82, it seems there was some confusion due to doodle managing automatically the conversion to local time10:12
besser82iglesiasg: I've seen it on the emails  ;)10:12
-!- lisitsyn1 [] has joined #shogun10:13
lisitsyn1hey guys10:13
besser82lisitsyn1: you're ~2h to early  :D10:13
lisitsyn1sonney2k: iglesiasg: thoralf: wiking: besser82: who's around?10:13
@iglesiasghere I am10:13
lisitsyn1so meeting is in 2 hours10:13
@iglesiasgI think it should be now but anyway10:14
@iglesiasgI can wait10:14
lisitsyn1hmm wait10:14
lisitsyn1I got confused by besser8210:14
lisitsyn1I think it is supposed to be now10:14
lisitsyn1according to doodle10:14
lisitsyn1besser82: ah you bad boy10:14
besser82lisitsyn1: Why?10:14
-!- HeikoS [] has joined #shogun10:15
-!- mode/#shogun [+o HeikoS] by ChanServ10:15
@HeikoSsorry for the delay10:15
@HeikoSheavy rain here10:15
besser82HeikoS: meeting starts in ~2h afaik...10:15
@iglesiasgheavy sun here10:15
@iglesiasgI think it should be now...10:15
@sonney2kiglesiasg, dont' get burned10:15
besser82HeikoS: on last email there was written 10:00h UTC10:16
@iglesiasgsonney2k, too cold though :(10:16
@HeikoSbesser82: wasnt that UTC -1 or so10:16
@HeikoSyeah UTC +210:16
@HeikoSwhich is 15 mins ago10:16
lisitsyn1it was a mistake yeah so it should be now10:16
besser82I see...10:16
@HeikoSbesser82: you are here anyways ;)10:16
@iglesiasgbesser82, there was a last mail saying 10:00 UTC+210:17
@HeikoSlisitsyn1: do you have an agenda?10:17
besser82Friday 12:00 UTC+2.10:17
besser82I'll try to prepare some agenda.10:17
@iglesiasganyway, what I do is just to trust the doodle, it shows local time so perfect10:17
besser82HeikoS, lisitsyn:  ^^^^10:17
lisitsyn1HeikoS: well not really strict10:17
lisitsyn1besser82: yeah but I send another mail just along10:17
@iglesiasgbesser82, there was a new mail after that10:17
lisitsyn1sorry if I confused you10:17
@HeikoSis wiking around?10:18
besser82lisitsyn1: np10:18
besser82wiking: ping?10:18
@sonney2kok lets next time only use UTC10:18
@iglesiasgbesser82, anyhow, is it a good moment for you now too?10:18
@sonney2kand everyone has to do the math on her own10:18
lisitsyn1yeah sorry it is always confusing but I didn't found way around10:18
besser82iglesiasg: np, in 5mins i'll be fie  ;)10:19
lisitsyn1ok let me reboot10:19
@iglesiasgbesser82, cool, then we are fine :)10:19
lisitsyn1because windows believes I should in 15 minutes10:19
lisitsyn1otherwise it will slap me right into face10:19
-!- lisitsyn1 [] has quit [Read error: Connection reset by peer]10:20
-!- shogun-notifier- [] has joined #shogun10:20
shogun-notifier-shogun-web: Soeren Sonnenburg :master * b90c977 / / (10 files):
shogun-notifier-shogun-web: add notebooks to homepage10:20
wikingbesser82: pong10:21
wikingbesser82: any news?10:22
besser82wiking: not yet. Will explain on meeting...10:26
* besser82 is on go :D10:27
@HeikoSwhere is sergey?10:27
besser82HeikoS: rebooting his M$ WinCrap  XD10:28
-!- lisitsyn1 [] has joined #shogun10:29
-!- lisitsyn1 is now known as lisitsyn10:31
-!- mode/#shogun [+o lisitsyn] by ChanServ10:31
@lisitsynso what's up10:31
@lisitsynare we ready to start?10:31
@lisitsynHeikoS: as for agenda it looks release thing is overwhelming10:32
@HeikoSlisitsyn: everyone is, go for it!10:32
* besser82 is on go10:32
* iglesiasg is ready10:32
wikingok so when's the meeting on the end?10:32
wikingnow or lateR? :)10:32
@lisitsynwiking: now!10:32
@lisitsynokay lets start10:32
@lisitsynthanks for joining this meeting10:32
@lisitsynwe now have gsoc ended and we've got a lot of new code10:33
@lisitsynand we planned release10:33
* besser82 thanks lisitsyn for chairing10:33
@lisitsynso lets talk how can we prepare for release10:33
@lisitsynand when should it be10:33
@HeikoSThe original plan (github milstone) was now10:33
@HeikoSor a few days ago10:33
@lisitsynyeah lets check milestone10:34
@HeikoSbut there are quite a few open issues remaining10:34
@lisitsynyeah a lot10:34
@HeikoSsome of them are not release relevant though10:34
@lisitsynI wouldn't get to discuss them all though10:34
@lisitsynso how do you think we should handle that?10:34
@HeikoSit would take ages to solve all of them, but we can release with just the most important ones fixed10:34
besser82but there's lot of new code during last days... is it tested well enough?10:35
@lisitsynbesser82: no, of course no ;)10:35
@HeikoSmy suggestion would be re-name milestone 3.0 to 3.1 and move the most important things to 3.010:35
@lisitsynHeikoS: yeah10:35
@HeikoSthen discuss who does them10:35
@lisitsynso everybody just10:35
@HeikoSwe kind of did that last time, but since then, we collected more and more10:35
@HeikoSwiking: what do you think?10:35
wikingthe bugs present in mileston 3.010:35
wikingshould stay10:36
@lisitsyncheck issues and let us know if you think they are not important10:36
wikinglisitsyn: malsar leak ;)10:36
@lisitsynwiking: yeah I remember10:36
wiking1701: is out of our hand10:36
wikingbut actually it's not leaking10:36
wikingso it's ok now10:36
wikingwe could close that one10:36
wiking1694-6 should be fixed in 3.010:36
@lisitsynwiking: lets just glance over it next days and try to remove a few10:37
wiking(1693 included)10:37
wikingi dont know about 1684 and 168010:37
@iglesiasgwiking, what about 1664 finally?10:37
wiking1625 is fixed - or at least it's going to be fixed today10:37
@lisitsynlets rather talk about resources we have ;)10:37
@lisitsynhow much time do you have to work on that?10:38
wiking1664 is something that is known by ubuntu devs10:38
wiking1488 can be closed10:38
@HeikoSwiking, lisitsyn all these smaller things are not sooo important I would say, we can always do a bugfix release10:38
wikingas besser82 did this for us thanx again for that10:38
wikingthe memleaks (1477) are getting in shape10:39
wikingso that can be closed as well soon10:39
wikingwhat really remains10:39
wikingis 'infrastructure' stuff10:39
wikingintegration of demo10:39
wikingsonney2k: just did the notebook hacking10:39
wikingor like ' Restructure and update NEWS '10:40
wikingwhich is all just non-dev releated stuff..10:40
@lisitsynimportant enough though10:40
@HeikoSwiking: yeah NEWS and feature list will be done by the weekend10:40
wiking1590 is going to be fixed as soon as besser82 gets us the cloud in shape10:40
wikingHeikoS: great!10:40
wikingwhat i would really like to be fixed still is: 154810:41
@lisitsynwhat date of release would you like?10:41
wiking Most of the python examples fail with python310:41
@HeikoSbesser82, wiking it would be great to have an access to the cloud stuff for the NIPS application, I will also do that on the weekend10:41
@HeikoSwiking: python3 is not important for now I would say10:41
wikingHeikoS: well it's getting really more and more into major OS distribs10:41
wikinghence it'd be great to have a fix for that10:41
@HeikoSwiking: I know, but just for this release I mean10:41
besser82wiking: most of it can be resolved by using 2to3 (python3)10:41
@lisitsynguys, date10:41
wikingbesser82: yeah those are fixed ;)10:41
wikingbesser82: the remaining problem is more complicated :P10:42
@sonney2kHeikoS, talk to me what you want to do before you do it10:42
wikinglisitsyn: 15th of oct10:42
@HeikoSwe should aim for next week release or so10:42
besser82wiking: allright, will have a look on that....10:42
@HeikoSsonney2k: will send a draft to mailing list10:42
@sonney2kregarding notebooks - we need some .txt file with a short description10:42
@HeikoSsonney2k: news: just a nice text with all the added features10:42
wikingi'd go for a mid october release10:42
@sonney2kHeikoS, no I mean the structure of NEWS needs to stay or else a couple of scripts won't work10:42
@HeikoSsonney2k: features: I will start something from which we can extend10:42
* besser82 will get the CMake refactoring ready over weekend10:43
wikingHeikoS: when's the deadline?10:43
@HeikoSsonney2k: yeah I dont want to change that, just to write a release text10:43
@HeikoSwiking: 8th or so10:43
@HeikoSwiking: I have to do it on the weekend10:43
@sonney2kregarding notebooks again - we need some .txt file for each with a short description!10:43
wikingbesser82: so we need to have really the notebook server up and running this weekend10:43
@HeikoSso one thing: I will disappear more or less after this weekend, phd courses. WIll be back Feb10:43
wikingif that's not doable, let me know and i'll setup a small thing temporarily10:43
@lisitsynHeikoS: feb is february?10:43
wikingHeikoS: hehe not so good news... what's going to be with the 'views' of features? :)10:44
@HeikoSwiking: will try to do that on a weekend ;)10:44
besser82wiking: the problem is sth. is screwed in redhat'S openStack.  That the problem for the persistent storage...10:44
@HeikoSwiking: what changes is that I wont be around during the day anymore10:44
wikingbesser82: since HeikoS really needs to have that server up and running for a conference application10:44
@HeikoSwiking: doesnt need to be perfect though10:44
wikingbesser82: it's alright if we cannot have it by the weekend... we'll have it when we have it10:44
@HeikoSwiking: and NIPS is in december, it should work nicely until then10:45
besser82wiking: if you can get sth. temporary it would be fine10:45
wikingbesser82: i'll just have to setup the other attempt i had last time...10:45
@HeikoSbut for application, just basic stuff is enough10:45
wikingHeikoS: yeah that's what i thought10:45
besser82wiking: You need some V-Server for that?10:45
wikingbesser82: naaah the whole squashfs is setup10:45
besser82wiking: I might have one in spare....10:45
@HeikoSbut what about the release?10:45
wikingso it's really a bit more hacking and that's it10:45
wikingHeikoS: you mean release + cloud10:45
@sonney2kthis is how the notebooks look like w/o description
wikingHeikoS: hopefully the thing with redhat's openstack is going to be fixed by then10:46
wikingsonney2k: is this autogen?10:46
@sonney2kI am extracting the first image from the notebook10:46
besser82wiking, HeikoS: the RedHat guys are working on it...10:46
wikingsonney2k: i would say that people should write a meaningfull title for the notebooks10:46
@sonney2kwiking, autogen == done live10:46
@iglesiasgthat looks nice10:46
@HeikoSI mean the cloud thing should not be part of the release, but something we are doing internally10:46
wikingi.e. does not have a title set at all iglesiasg10:47
@sonney2kwiking, no rather a text file with the same prefix10:47
@HeikoSthe release should focus on exposing the GSoC projects in a nice looking way10:47
@sonney2kwiking, ahh ok well yeah that too10:47
wikingsonney2k: we woudl store it within the repo? those txt files10:47
@HeikoSiglesiasg: oh btw I think you should link your github account, and the gsoc project page in the notebook10:47
@sonney2kwiking, yes sure10:47
@sonney2kdirectly in the notebook dirs10:47
@iglesiasgHeikoS, sure10:47
wikingsonney2k: as we can really set like a meaningful <abstract></abstract> tag in the generated html and just add that to
@sonney2kwiking, then I can display that test instead of the filename10:48
@lisitsynokay guys lets get back10:48
@lisitsynso wiking votes for 15th of oct10:48
@lisitsyndoes it work for you?10:48
@iglesiasgwiking, I don't get that thing of the title, why there is none, but I will try to have a look10:48
* besser82 15th Oct +110:48
@lisitsynother thinng is10:48
@lisitsynwho does the release?10:48
@sonney2kohh and our release script is R.I.P10:49
@lisitsynhaha nice10:49
@sonney2kso no more make release10:49
@sonney2k== hours of work10:49
@sonney2kto do a release10:49
@lisitsynI am afraid wiking10:49
besser82sonney2k: I can fix that within cmake10:49
@lisitsynoh besser8210:49
besser82sonney2k, lisitsyn: like make tarball10:49
wikingbesser82: cpak10:49
@sonney2kit is not that easy10:49
@sonney2ktarball cpack is easy yes10:50
@lisitsynso besser82 & wiking can you approach to repair 'make release'?10:50
besser82wiking: I know  ;)10:50
wikinglisitsyn: yeah10:50
@lisitsynthat's for you then ok10:50
besser82lisitsyn: sure10:50
@sonney2kbut not signing everything with gpg + upload + automagic news update on website etc10:50
wikingwe'll the the autosign + hashing etc as well...10:50
wikingso that should not be a problem at all10:50
wikingbesser82: so you're going to have the refactored cmake done during the weekend?10:51
@sonney2k..uploading to freshmeat, writing descr's there10:51
besser82wiking: yes10:51
@sonney2k...updating the website etc10:51
wikingsonney2k: should be alriht10:51
wikingsonney2k: we still have the access to the script10:51
wikingin the master branch10:51
wikingso we'll just use that10:51
wikingas a guideline10:51
wikingand then just we'll create a new target: 'make release'10:52
wikingor make dist10:52
@lisitsynokay so the plan10:52
besser82wiking, sonney2k: I think the script should be sth. external from cmake10:52
@HeikoSlisitsyn: you should create a priority list of things needed for the release, linked with github issues and who is responsible, like wiking did last time10:52
wiking(maybe that's more appropriate)10:52
@lisitsynHeikoS: yeah I know10:52
@lisitsynso the plan is10:52
@lisitsynwe should have all issues assigned by tomorrow10:52
@sonney2kthat script is obsolte in some parts too - the website is django now ... anyway we should try to at least get the tarball upload & signing automated10:53
@lisitsynwiking & besser82 port the make release script10:53
@sonney2kthen later news -> news on website10:53
@sonney2klisitsyn, well some issues are days work :D10:53
@lisitsynsonney2k: yes10:53
wikingwe need demo integration10:53
@lisitsyns what?10:53
@lisitsynsonney2k: we just have to assign them10:54
wikingdemos are way too cool10:54
wikingbut we need to add them to the website10:54
@lisitsynI am responsible for assigning milestone issues10:54
wikingHeikoS: any update on that parameterMap issue with thread data race condition?10:55
besser82wiking, sonney2k:  The testsuite-data should be within release tarball? What do oyu think?10:55
wikingthat's separate10:55
* iglesiasg brb10:55
@sonney2kbesser82, separate10:55
@sonney2ktoo big10:55
wikingoooh and btw: static analysis error fixes!10:55
@sonney2kand not required if you don't want to run demos10:55
besser82wiking, sonney2k:  is it really that big?10:55
wikingwaaaaty too many of them...10:55
@sonney2kbesser82, 200 MB ?10:55
wikingbesser82: yep10:55
@lisitsynokay is that all about release?10:56
@HeikoSwiking: no sorry, I havent looked into these race conditions10:56
thoralfwiking: Is there a list of errors somewhere?10:56
@HeikoSwiking: I am not sure how much they will hurt us?10:56
@HeikoSthoralf: see github milestones10:56
wikingHeikoS: ok please get somebody assigned to that one as it really hurts us atm ;)10:56
@HeikoSthoralf: ehm 3.0 milestone10:56
wikingthoralf: yes milestone .3010:56
besser82wiking, sonney2k:  I can see ~90 MBytes /data/testsuite10:56
@HeikoSwiking: what happens?10:56
@HeikoSwiking: I might de-activate threads in GP stuff, that will solve quitre a feq10:56
wikingHeikoS: well it's just destined to fail ;)10:56
@HeikoSwiking: ok :)10:57
thoralfwiking, HeikoS: Which ticket?10:57
@HeikoSwiking: will check parameter map10:57
wikingthoralf: any ;)10:57
@HeikoSthoralf: filter for milestone 3.010:57
@HeikoSyou get like 40 or so issues10:57
thoralfwiking: No, I mean: Which ticket for "oooh and btw: static analysis error fixes"10:57
wikingthoralf: aaah that's this10:57
wikingthere's way tooo many of them10:57
@lisitsynI don't think we need to resolve all of them10:58
wikingand just take the last one:
wikingthoralf: ^10:58
@lisitsynI looked into a few and they are just about code rewrite10:58
@lisitsynthat's not what we should do now10:58
thoralfwiking: Thanks.10:58
wikinglisitsyn: we should do it one day...10:58
@lisitsynokay guys10:58
@lisitsynNIPS application10:58
@lisitsynHeikoS: what help do you need?10:58
@HeikoSlisitsyn: I will write the abstract (1 page)10:59
@lisitsyndo you need help with reviewing it?10:59
@HeikoSyou have to give me a nice webpage with notebooks and the cloud basic10:59
@HeikoSlisitsyn: yes, and then I will send it to you guys for review10:59
* iglesiasg re10:59
@lisitsynwho's on notebooks?10:59
@lisitsynHeikoS: what's cloud basic?10:59
@HeikoSlisitsyn: they are more or less done, it would just be cool to have them on the webpage10:59
@lisitsynhow do you want to do that?10:59
@HeikoSlisitsyn: I want to include a cloud shogun link in the abstract11:00
@lisitsynah so11:00
@lisitsynipython console11:00
wikinglisitsyn: i've already created a small testbed for it... needs some extensions, but all in all we'll move to besser82's system as soon as it's ready11:00
@lisitsynwith shogun installed11:00
wikinglisitsyn: yesp11:00
@lisitsynso it is like some job for wiking and besser82 again?11:01
@lisitsynsorry ;)11:01
@lisitsynokay then NIPS plan is11:01
wikinglisitsyn: more for me now... we are still waiting for besser82 and fedora ppl to fix some stuff now on their system11:01
@lisitsynHeiko writes abstract11:01
@lisitsynwe review it whether possible11:01
wikingbut a basic thing will be ready by weekend for HeikoS11:01
@lisitsynwiking and besser82 collaborate on cloud stuff11:01
wikingwhat i'll do is to add opencv as well to the image i created11:01
@lisitsynis that it?11:02
wikingso we can show the cool opencv demo that kevin created11:02
@HeikoSwiking: good, if that doesnt work, deadline is 7th or so so thats latest :)11:02
besser82wiking: +1  :)11:02
@HeikoSwiking: yeah that opencv thing will be a boig plus for the submission11:02
wikingas one of the chairman of that workshop is a main opencv guy11:02
@lisitsynhaha yeah11:02
@HeikoSwiking: since the red line is "integrating different software"11:02
wikingHeikoS: as well as that gerry whoever11:02
wikingor what's his name is opencv guy ;P11:03
@lisitsyngarry bradski I think11:03
wikingGary Bradski11:03
@lisitsynor gary11:03
wiking"Gary Bradski is the Vice President of Computer Vision and Machine Learning, Magic Leap Inc. and President and CEO of OpenCV Foundation, which manages the development of the OpenCV machine learning and computer vision library"11:03
@lisitsynI actually would like to see shogun replacing opencv's ml module :D11:03
@lisitsynnot possible though11:03
wikingso if we do a demo with shogun+opencv i think that's big plus11:03
besser82wiking, sonney2k:  you tried to compress the testsuite-dir from data and it is ~15 MBytes after compression  :D11:03
@lisitsynwiking: have you ever used opencv's ml?11:04
wikinglisitsyn: yeah it's a shit11:04
besser82wiking, sonney2k:  s/^you/I/11:04
@lisitsynI wouldn't say API is something convenient11:04
wikinglisitsyn: and some of their algos are classify only11:04
wikingso it's really stupid...11:04
@lisitsynHeikoS: I think we are not getting to something you want to push11:04
wikingindeed replacing opencv's ml would be great but i think that's really a long run...11:04
@lisitsynthat reminds me HeikoS wants funding ;)11:05
wikingyeah funding would be great11:05
wikingand here comes the question for11:05
wikingHeikoS: and sonney2k11:05
wikingwhat's the status of shogun fundation?11:05
@HeikoSlisitsyn, wiking I think its quite realistic to get that11:05
@sonney2kwiking, waiting for HeikoS to do sth11:05
@HeikoShowever, I think we should do that after the release11:05
@lisitsynHeikoS: do what?11:06
wikingHeikoS: what's the status of fundation papers?11:06
@HeikoSah sorry11:06
wikingHeikoS: do u need some help?11:06
@HeikoSfoundation will be done, no its easy11:06
@sonney2klisitsyn, foundation 'satzung'11:06
thoralfwiking: The static list contains 101 errors - almost all of them are easy to fix (but to make sure we don't break anything last-minute, we should unit-test things first).11:06
@HeikoSbut then, we should discuss these other things, what we want, how we do it etc.11:06
@HeikoSbut I would do that later11:06
wikingif it's tooo complicated to get it done in germany11:06
@sonney2kHeikoS, I think you promised 3 months ago11:06
@HeikoSand sorry for the delay on the foundation, also on top of my list11:06
thoralfwiking: Can we fix the "Argument with 'nonnull'" errors by assert(var != NULL)?11:06
@HeikoSI know, gsoc came in between11:06
@sonney2kwiking, same like you with workshop video uploads11:06
wikingwe can choose another place for shogun fundation and have it done somewhere11:07
wikingsonney2k: *nod*11:07
@lisitsynphew it seems I escaped 'you promised' gun11:07
wikingthoralf: i think so11:07
@HeikoSfoundation is not important before the release is done, so I will do it once all release stuff is solved11:07
@lisitsynHeikoS: okay but what's about funding11:08
@lisitsynhow serious we are11:08
@lisitsynand if we are how should we do that11:08
@HeikoSI would go for as serious as possible11:08
@HeikoSbut after release11:08
besser82lisitsyn: I'm on it to get SHOGUN to publicity11:08
@lisitsynwell HeikoS I know11:08
@HeikoSI think we should aim for something like a GSoC which we found ourselves11:08
@HeikoSbut like 3-times is a row11:08
@lisitsynHeikoS: so you wouldn't talk about it yet?11:08
besser82lisitsyn: Already talked with some Heise && linux magazine guys  ;)11:08
@lisitsynbesser82: I have no idea what is it ;)11:09
@HeikoSlisitsyn: no I would wait a little11:09
@lisitsynHeikoS: I just wanted to may be discuss what exactly we should fund11:09
@HeikoSuntil we have the cloud, the notebooks on webpage, bugs solved, and heise stuff maybe11:09
@HeikoSah I see11:09
besser82lisitsyn: Heise is a BIG German "verlag" for IT focussed magazines  ;)11:09
@lisitsynverlag verlag11:09
@lisitsynI see11:10
wikingbesser82: is the rpm generator file/script is public already?11:10
@HeikoSso if we could find a person who is willing to work fulltime for us for say 1 year11:10
wikingi mean can we include that in our repo?11:10
@lisitsynbesser82: I would have fun to write anything for them ;)11:10
@HeikoSthat would be a good investment11:10
@HeikoSlike gsoc but more involved11:10
besser82lisitsyn: Sounds good  ;)11:10
wikingHeikoS: i have the time ;)11:10
@HeikoSwiking: I was thinking about you ;)11:10
@sonney2kwiking, only if you upload the videos11:10
wikingHeikoS: meaning i'm not booked up for anything serious till next september11:10
wikingsonney2k: :DDD11:11
wikingsonney2k: ayay captain!11:11
besser82wiking:  the spec is avail:  but not finished though...11:11
@HeikoSyeah, so if we could pay you for that, that would be amazing11:11
wikingbesser82: can we start including that in our repo?11:11
besser82wiking: not yet11:11
besser82wiking: will add that along cmake with autogeneration  ;)11:11
wikingbesser82: but we can update it later... i mean it's already in a 'working' state or?11:11
wikingah ok11:12
wikingthen i guess it's after this weekend11:12
@HeikoSsince you could a) push all new ideas wrt to cloud stuff b) polish some things on the fly11:12
besser82wiking: right  ;)11:12
wikingbesser82: ok cool11:12
@HeikoSand then we could also focus a bit more on presentation11:12
wikinglooking forward for that cmake refurbish PR11:12
@HeikoSlisitsyn, but yeah, after release, I would do an own meeting for that, I can prepare things I have in mind etc11:12
wikingok so one more question11:12
@sonney2kguys can we do a little who does what for the release?11:12
besser82wiking: it will be incredible  ;)11:12
@sonney2kI mean I will integrate notebooks and demose11:13
besser82wiking: and huge   ;)11:13
wikingwhere would the demos run?11:13
@sonney2kbut there is 0 more I can do11:13
@sonney2kwiking, fatbot11:13
wikingsonney2k: in that image?11:13
@sonney2kwiking, for example11:13
wikingsonney2k: ok11:13
@lisitsynI'll get to issues and try to manage resolving them11:13
wikingsonney2k: then i'll prepare that11:13
@lisitsynI mean I'll approach some issues by myself and bug other people to resolve issues assigned to them ;)11:14
wikingsonney2k: nightly_all i know the last fix... so will be finished today11:14
@sonney2kI mean basically everyone who does not infrastructure should resolve issues the buildbot has / that are on github11:14
@sonney2kwhat I would suggest is not that you lisitsyn go through the issues11:14
@sonney2kbut that everyone assigns himself to an issue11:15
@sonney2kor to multiple with some deadline11:15
@HeikoSsonney2k: would still be good if lisitsyn made an overview or priorities and categories of issues11:15
wikingHeikoS: there's things like this with parameter.cpp as well:11:15
@HeikoSotherwise agreed11:15
wiking/home/buildslave/clang34_-_undefined_behaviour_analysis/build/src/shogun/base/Parameter.cpp:2349:23: runtime error: member access within address 0x00000c25e4f8 which does not point to an object of type 'SGSparseVector<char>'11:15
wiking0x00000c25e4f8: note: object is of type 'shogun::SGSparseVector<double>'11:15
@HeikoSwiking: is that in an issue?11:15
wikingHeikoS: i'll check11:16
@HeikoSwiking: if not, assign to me, I will try to solve a few tomorrow and sunday11:16
wikingHeikoS: if not i'll put it up11:16
@sonney2kHeikoS, well you can decide your own what is important right? So just do that stuff and yes we can tag things 'minor'11:16
wikingactually what would be great is to get the buildbot all green if possible11:16
wikingbut that's a lot of work11:16
@sonney2kwiking, that is HeikoS hack for serialization11:16
wikingso we might not have that11:17
@HeikoSwiking: yeah I second that, at least the basic ones11:17
@HeikoSwhat about the warning hunt?11:17
@HeikoSjust on the fly?11:17
@sonney2kHeikoS, I fixed them11:17
@HeikoSsonney2k: sweet!11:17
wikingi think we have quite some schedule set up11:18
wikingi guess it's going to be some discussions here anyways11:19
wikingduring the weekend11:19
wikingso if any of you can plz stick around and join the bughunt11:19
@HeikoSoh one thing, lisitsyn I think we should aim for translating the shogun-tutorial stuff into notebooks on the long run, already did that for the MMD one11:19
@sonney2kHeikoS, iglesiasg - write notebook descriptions and push everyone else to do so if you cannot do them all11:19
wikinglisitsyn: assigining is cool but i guess the common rule should be 'if u have time and want to fix a bug take one and fixit'11:19
wikingand please11:20
@lisitsynwiking: well my plan was to talk about issue with assignee11:20
@iglesiasgsonney2k, these descriptions will go directly to the website, right? Next to the links we have for the notebooks I guess11:20
wikingif u are assigned to a bug then either work on it, but if u cannot just mention it in the issue comment that u cannot do it, so others know that is not being worked on11:20
wikingiglesiasg: in the repo11:21
@iglesiasgwiking, all right. txt in the same dir as the ipynb with the same name11:21
wikingiglesiasg: but i think if we find a way to generate a special html tag (html5 <abstract>) within the notebook, then we can have a simple parser that extracts that section of the notebook ;P11:21
wikingiglesiasg: or even better: mark and abstract section in the notebook itself ;P11:21
wikingiglesiasg: if u or anybody have a good idea for this how to do it11:22
@iglesiasgwiking, ok11:22
wikingjust make an example and mention it in an issue11:22
wikingsince json parsing is quite easy11:22
wikingand then we just extract that json variable and we are all set11:22
@sonney2kiglesiasg, yes11:23
wikingthen the description of the notebook is self-included11:23
wikingwhich would be better imo11:23
@iglesiasgwiking, it sounds neat like that yes11:23
@iglesiasgwiking, let's see if there is an easy way to make a special cell in the notebook for this11:24
wikingbtw guys11:24
wikingHeikoS: especially:
wikinghave fun ;)11:24
@HeikoSwiking: ah thats cool11:25
wikingand let's just beat them :P11:25
@sonney2kwiking, yes would be nice11:25
@HeikoSwiking: we will dont worry ;)11:25
thoralfwiking: Where do I find information how to generate the static analysis?  So I can check if I fixed something...11:25
@sonney2kHeikoS, iglesiasg, wiking so what about having a mandatory 'abstract' as the first thing in a notebook?11:25
wikingthoralf: either ask shogun-buildbot11:26
@sonney2kor wait for him :)11:26
@iglesiasgsonney2k, yes that's the tihng11:26
@HeikoSsonney2k: up for that! author contacts included and short summary11:26
@iglesiasgsonney2k, I am searching if there is a way to mark a cell in the notebook11:26
wikingthoralf: or if u  dont want to push it before then just: build cmake -D_HAS_CXX11_FLAG=OFF -D_HAS_CXX0X_FLAG=OFF -DBUILD_EXAMPLES=OFF -DENABLE_CCACHE=OFF -DCMAKE_BUILD_TYPE=Debug ..11:26
@lisitsyndoes anyone have any other issues to discuss?11:26
wikingthoralf: ok sorry so: scan-build cmake -D_HAS_CXX11_FLAG=OFF -D_HAS_CXX0X_FLAG=OFF -DBUILD_EXAMPLES=OFF -DENABLE_CCACHE=OFF -DCMAKE_BUILD_TYPE=Debug ..11:27
wikingthoralf: scan-build -analyze-headers -o static_analysis --status-bugs make -j 2 -C build11:27
@sonney2kiglesiasg, how about I just use the first cell_type markdown?11:27
wikingthoralf: u need clang installed11:27
thoralfwiking: Okay, thanks.11:27
wikingthoralf: note that it'll take quite some time to generate the output :S11:27
@iglesiasgsonney2k, easy enough11:27
@sonney2kiglesiasg, if it is too long I can stop after say 1000 chars11:28
@sonney2kalright then11:28
@iglesiasgwe just should write somewhere so that when someone writes a new notebook, keeps that in mind11:28
wikingiglesiasg: a notebook template? :)11:28
wikingiglesiasg: shogun notebook template and just make it somewhere available (either within repo or the website)11:29
thoralfwiking: No worries.  Doing it on the fastest machine I can get. ;)11:29
wikingthoralf: alrighty ;)11:29
@sonney2kiglesiasg, could you please do such notebook template?11:29
@iglesiasgwiking, so this template is basically a notebook right?11:29
@sonney2kthat would indeed be nice!11:29
@iglesiasgwe save it as ipynb and that's all11:29
@sonney2kiglesiasg, yes with some title and a standard format11:29
wikingiglesiasg: yes11:29
@iglesiasgI will do that11:30
wikingiglesiasg: just add an issue for this then11:30
@iglesiasglatest this Sunday is ready (got a busy day tom)11:30
wikingthnx heaps!11:30
@iglesiasgno problem at all :)11:30
@HeikoSwiking, iglesiasg, sonney2k it will be easy to edit all notebooks and make them have the format once decided on that :)11:30
@HeikoSanything else to discuss?11:33
@sonney2kdo work!11:34
@lisitsynbrief indeed but anything I missed?11:34
besser82HeikoS, sonney2k: Would it be possible for me to get membership of "shogun-toolbox" on gh?11:35
besser82HeikoS, sonney2k: So I can assign all build-sys related issues to me ?11:35
@HeikoSbesser82: sure, I have no problem with that, ask sonney2k, he is the master ;)11:36
@HeikoSlisitsyn: looks good, quite short mr lazy ;)11:36
@HeikoSI have to go now, see you all tomorrow!11:36
besser82HeikoS: allrighty  ;) see you tomorrow then  ;)11:37
* besser82 waves to HeikoS11:37
@lisitsynsee you11:37
@lisitsynHeikoS: I will fill in anythin else once I re-read logs11:37
@HeikoSok :)11:38
-!- HeikoS [] has left #shogun []11:38
besser82sonney2k, sonne|osx:  Can you please add me to the shogun-toolbox group on gh?  So I can assign all build-sys related issues to me?11:38
thoralflisitsyn: No date?11:39
besser82sonney2k, sonne|osx: my gh-acc is "besser82", too ;)11:40
@lisitsynthoralf: date of11:40
@sonney2kbesser82, try11:43
besser82sonney2k: will do, thanks  :)11:43
besser82sonney2k: seems like my perms a re issufficient to assign issues  :(11:45
-!- iglesiasg [] has quit [Quit: Ex-Chat]11:57
@sonney2kbesser82, ahh github doesn't support this12:07
@sonney2kbesser82, so you are root now12:08
@sonney2kbesser82, or god or $DEITY as you wish12:08
besser82soneey2k:  many thanks for your trust  ;)12:08
shogun-notifier-shogun-web: Soeren Sonnenburg :master * 563f380 / / (6 files):
shogun-notifier-shogun-web: add abstracts12:40
@sonney2kso now with some abstract12:41
-!- iglesiasg [] has joined #shogun14:23
-!- mode/#shogun [+o iglesiasg] by ChanServ14:24
thoralfsonney2k: The MathJax formulas for the last two notebooks on the page are broken.15:10
thoralfChromium, v2815:11
-!- shogun-notifier- [] has quit [Quit: transmission timeout]15:40
-!- iglesiasg [] has quit [Ping timeout: 240 seconds]15:44
-!- lisitsyn [] has quit [Quit: Leaving.]16:08
-!- thoralf [] has quit [Quit: Konversation terminated!]16:10
wikingsonney2k: gr816:22
-!- shogun-notifier- [] has joined #shogun16:29
shogun-notifier-shogun: Viktor Gal :develop * 780a017 / tests/unit/discover_gtest_tests.cpp:
shogun-notifier-shogun: Ignore memory trace output in unit test discovery16:29
-!- besser82 [~besser82@fedora/besser82] has quit [Quit: freedom, friends, features, first --->]17:11
wikingshogun-buildbot: force build --branch=develop 'nightly_all'17:13
shogun-buildbotThe build has been queued, I'll give a shout when it starts17:13
-!- shogun-buildbot [] has quit [Quit: buildmaster reconfigured: bot disconnecting]17:15
-!- shogun-buildbot [] has joined #shogun17:15
shogun-buildbotbuild #63 of clang34 - undefined behaviour analysis is complete: Failure [failed test]  Build details are at  blamelist: Viktor Gal <>17:25
shogun-buildbotbuild #65 of clang34 - thread analysis is complete: Failure [failed test]  Build details are at  blamelist: Viktor Gal <>17:40
-!- travis-ci [] has joined #shogun17:47
travis-ci[travis-ci] it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun:
-!- travis-ci [] has left #shogun []17:47
shogun-buildbotbuild #479 of nightly_all is complete: Success [build successful]  Build details are at
shogun-buildbotbuild #83 of clang34 - static analysis is complete: Failure [failed analyse]  Build details are at  blamelist: Viktor Gal <>18:23
@sonney2kshogun-buildbot, force build --branch=develop nightly_default18:26
shogun-buildbotbuild forced [ETA 36m00s]18:26
shogun-buildbotI'll give a shout when the build finishes18:26
shogun-buildbotbuild #563 of nightly_default is complete: Success [build successful]  Build details are at
-!- shogun-notifier- [] has quit [Quit: transmission timeout]19:30
-!- iglesiasg [] has joined #shogun21:44
-!- mode/#shogun [+o iglesiasg] by ChanServ21:44
-!- hushell [] has joined #shogun22:01
-!- iglesiasg [] has quit [Ping timeout: 264 seconds]22:02
-!- iglesiasg [] has joined #shogun22:08
-!- mode/#shogun [+o iglesiasg] by ChanServ22:08
-!- lisitsyn [~lisitsyn@] has joined #shogun22:19
-!- iglesiasg [] has quit [Quit: Ex-Chat]23:16
--- Log closed Sat Oct 05 00:00:28 2013