--- Log opened Mon Feb 10 00:00:52 2014 | ||
shogun-buildbot | build #195 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/195 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:08 |
---|---|---|
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 00:32 | |
shogun-buildbot | build #190 of debian wheezy - memcheck is complete: Failure [failed memory check generate html] Build details are at http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/190 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:37 |
-!- pranet [73f91219@gateway/web/freenode/ip.115.249.18.25] has joined #shogun | 02:46 | |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has quit [Read error: Connection reset by peer] | 03:00 | |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has joined #shogun | 03:17 | |
-!- lisitsyn [~lisitsyn@109-226-71-150.clients.tlt.100megabit.ru] has quit [Read error: Operation timed out] | 03:26 | |
shogun-buildbot | build #610 of nightly_all is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/610 | 03:34 |
-!- sonne|osx_ [~sonne@e179071173.adsl.alicedsl.de] has joined #shogun | 03:47 | |
-!- sonne|osx [~sonne@f052204221.adsl.alicedsl.de] has quit [Ping timeout: 246 seconds] | 03:48 | |
-!- sonne|osx_ is now known as sonne|osx | 03:48 | |
pranet | Hi. How do I set LD_LIBRARY_PATH ? Getting missing libshogun.so.15 error during runtime. | 04:17 |
shogun-buildbot | build #708 of nightly_default is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/708 | 04:25 |
-!- Saurabh7 [~Saurabh7@115.248.130.148] has quit [Quit: Leaving] | 05:43 | |
-!- pranet [73f91219@gateway/web/freenode/ip.115.249.18.25] has quit [Ping timeout: 245 seconds] | 05:45 | |
@sonney2k | shogun-buildbot, force build --branch=develop nightly_all | 06:23 |
shogun-buildbot | build forced [ETA 9m48s] | 06:23 |
shogun-buildbot | I'll give a shout when the build finishes | 06:23 |
@sonney2k | shogun-buildbot, force build --branch=develop 'osx2 - python' | 06:23 |
shogun-buildbot | build forced [ETA 27m36s] | 06:23 |
shogun-buildbot | I'll give a shout when the build finishes | 06:23 |
shogun-buildbot | build #611 of nightly_all is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/611 | 06:28 |
shogun-buildbot | build #52 of osx2 - python is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/52 | 06:39 |
-!- sonne|osx [~sonne@e179071173.adsl.alicedsl.de] has quit [Quit: sonne|osx] | 07:12 | |
-!- sonne|osx [~sonne@82.113.121.2] has joined #shogun | 08:19 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 08:35 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 9dabfca / / (9 files): https://github.com/shogun-toolbox/shogun/commit/9dabfca2c9c14800a63ee7711c5d1164eaed46f1 | 08:35 |
shogun-notifier- | shogun: Add tmp directory inside python modular examples and direct all temporary | 08:35 |
shogun-notifier- | shogun: output to that dir. This fixes tests running concurrently that read the | 08:35 |
shogun-notifier- | shogun: files in the examples directory. | 08:35 |
-!- sonne|osx [~sonne@82.113.121.2] has quit [Quit: sonne|osx] | 08:35 | |
shogun-buildbot | build #1829 of cyg1 - libshogun is complete: Failure [failed compile test] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1829 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:00 |
@wiking | fyi: There was quite a lot of interest in a 2.0.12 release, so I've spent the weekend creating, testing and releasing it. It should contain the fixes you need. | 09:25 |
@wiking | I'll do a tag in a day or two for 3.0.0 | 09:25 |
@wiking | (libc++ fix) | 09:25 |
-!- sonne|work [~sonnenbu@24-134-74-216-dynip.superkabel.de] has joined #shogun | 09:34 | |
sonne|work | wiking: http://buildbot.shogun-toolbox.org/builders/debian%20wheezy%20-%20memcheck/builds/190/steps/generate%20html/logs/stdio | 09:34 |
sonne|work | wiking: no more memory logs | 09:34 |
-!- travis-ci [~travis-ci@ec2-107-20-126-221.compute-1.amazonaws.com] has joined #shogun | 09:35 | |
travis-ci | [travis-ci] it's Soeren Sonnenburg'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/18563953 | 09:35 |
-!- travis-ci [~travis-ci@ec2-107-20-126-221.compute-1.amazonaws.com] has left #shogun [] | 09:35 | |
sonne|work | wiking: any idea why travis claiming some massacre ^^^ ? It built just fine https://travis-ci.org/shogun-toolbox/shogun/builds/18563953 | 09:45 |
shogun-buildbot | build #186 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/186 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:58 |
shogun-buildbot | build #181 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/181 blamelist: Soeren Sonnenburg <sonne@debian.org> | 10:14 |
shogun-buildbot | build #196 of clang34 - static analysis is complete: Failure [failed analyse] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20static%20analysis/builds/196 blamelist: Soeren Sonnenburg <sonne@debian.org> | 10:55 |
-!- lambday [67157d4f@gateway/web/freenode/ip.103.21.125.79] has joined #shogun | 11:11 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has quit [Read error: Connection timed out] | 11:15 | |
-!- zxtx [~zv@c-98-223-196-32.hsd1.in.comcast.net] has joined #shogun | 11:16 | |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 11:35 | |
shogun-buildbot | build #191 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/191 blamelist: Soeren Sonnenburg <sonne@debian.org> | 11:39 |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 11:54 | |
lambday | sonne|work: lisitsyn: fixed the bug that roman reported - regarding clone()... sending PR | 12:53 |
lambday | an uninitialized memory was causing this - in copy_ptype.. | 12:54 |
lisitsyn | lambday: cool, although no idea what is it :D | 12:54 |
lambday | lisitsyn: replacing SG_MALLOC with SG_CALLOC in one place fixed this :) have a look.. attaching valgrind report as well | 12:55 |
lisitsyn | lambday: could you please send me a link when it is here | 12:56 |
lambday | lisitsyn: yeah just a sec | 12:56 |
lambday | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/base/Parameter.cpp#L4019 here we are initializing the memory - and here we're trying to dereference - https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/base/Parameter.cpp#L3646 | 12:58 |
lambday | valgrind was complaining regarding conditional jump base don uninitialized memory | 12:58 |
lambday | lisitsyn: log before fix - https://gist.github.com/votjakovr/8834064 | 13:03 |
lambday | lisitsyn: I gotta go.. will be back in an hour! please have a look and let me know! | 13:08 |
-!- lambday [67157d4f@gateway/web/freenode/ip.103.21.125.79] has quit [] | 13:08 | |
sonne|work | lisitsyn: not sure if this was really intended... | 13:43 |
lisitsyn | sonne|work: what? | 13:43 |
sonne|work | the memory access there - I don't know if an init with NULL is what we want... | 13:44 |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has joined #shogun | 14:01 | |
lambday | sonne|work: hi | 14:03 |
sonne|work | lambday: hi | 14:03 |
sonne|work | about your patch - I am not sure if the bug is not sth else... you sure that the comparison at this stage there is intended? | 14:04 |
sonne|work | lambday: ^ | 14:04 |
sonne|work | lambday: I mean sure it fixes the valgrind error but will there be just some other bug? | 14:04 |
lambday | sonne|work: well... I need to ask Heiko regarding what its meant to be... do we need to check both here? I mean, if source is NULL, we practically have nothing to do, right? | 14:06 |
lambday | so, wonder if we can replace if (!casted1 & !casted2) with just if (!casted1) | 14:06 |
lambday | in that case, it would also serve the purpose | 14:07 |
lambday | sonne|work: will there be problems if we use calloc? | 14:09 |
sonne|work | lambday: no it is just malloc + memset with 0 value | 14:10 |
sonne|work | lambday: question is just if that was the bug or not | 14:10 |
lambday | sonne|work: well, I can't think of a scenario when both casted1 and casted2 checks are required - https://github.com/lambday/shogun/blob/develop/src/shogun/base/Parameter.cpp#L3646 | 14:11 |
lambday | if we don't need to check casted2 here at all, then that is the bug | 14:12 |
lambday | we should just check casted1 instead | 14:12 |
lisitsyn | sonne|work: https://twitter.com/ChengSoonOng/status/432863700500762624 | 14:13 |
sonne|work | lambday: how can any of them be null? | 14:16 |
sonne|work | lambday: I mean that should never happen or? | 14:16 |
sonne|work | lisitsyn: ohh sh.... | 14:17 |
lambday | checking... | 14:18 |
lisitsyn | sonne|work: this sucks I guess? | 14:22 |
lambday | sonne|work: you're right - the error is something else.. didn't find it yet | 14:44 |
lambday | sonne|work: regarding being null, casted2 can be null.. casted1 shouldn't be as per I think | 14:49 |
sonne|work | lambday: I am not even sure where this code is in - some procedure where we do b=a.clone() or what? | 14:53 |
sonne|work | lambday: if yes then of course things in b can be NULL | 14:53 |
lambday | sonne|work: yeah... while copying all parameters hierarchically, we do this clone... and things in b can be null.. | 14:55 |
lambday | but I just checked that b was null before in another level of this hierarchy, there it didn't give this error... | 14:55 |
sonne|work | lambday: so it should do depth first wise copy | 14:55 |
lambday | yep.. that's what it does | 14:56 |
sonne|work | lambday: but then the check for casted2 != doesn't make sense to me | 14:58 |
lambday | for this particular example that roman gave, DenseFeatures (b was 0) --> DynamicObjectArray --> PruneVarSubMean (b is 0 - here it gives error upon unref-ing) | 14:58 |
lambday | sonne|work: yeah I too am confused if its at all needed | 14:59 |
sonne|work | ahh he is returning true when both are null | 14:59 |
sonne|work | hmmhh but that would be done anyways | 14:59 |
lambday | sonne|work: is there any probably scenario when casted1 is null and casted2 is not? shouldn't be! | 15:00 |
lambday | its all newly allocated memory | 15:00 |
sonne|work | lambday: yeah it should just be set to null there | 15:01 |
lambday | yep! | 15:01 |
sonne|work | lambday: I mean if (!casted1) { *((CSGObject**)target) = null; ) | 15:01 |
sonne|work | and all good | 15:01 |
lambday | yeah | 15:01 |
lambday | let me try this but I still think the error persists | 15:02 |
sonne|work | lambday: or is it possible to clone into some *existing* object | 15:02 |
lambday | checking | 15:02 |
lambday | oh! | 15:02 |
sonne|work | yeah look | 15:02 |
sonne|work | he does an SG_UNREF there | 15:02 |
lambday | yeah! | 15:02 |
sonne|work | so maybe that is why | 15:02 |
sonne|work | lambday: so indeed the newly allocated memory should be NULL by default | 15:03 |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has quit [Ping timeout: 245 seconds] | 15:19 | |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has joined #shogun | 15:55 | |
lambday | sonne|work: even when we keep only one check for if (!casted1) and remote the check for casted2, its checking again in the SG_UNREF(...)... if(x) thing... | 15:58 |
lambday | remove* | 15:58 |
lambday | sonne|work: so getting the same error again... conditional jump or move depends on uninitialized value... | 15:59 |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has quit [Quit: Page closed] | 16:26 | |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 17:56 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 17:56 | |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has joined #shogun | 18:29 | |
-!- pranet [ca4eafc7@gateway/web/freenode/ip.202.78.175.199] has joined #shogun | 18:40 | |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving] | 18:55 | |
-!- lisitsyn1 [~lisitsyn@80.252.20.67] has joined #shogun | 18:56 | |
@sonney2k | lambday, yes of course we need to init things with NULL - I guess your CALLOC patch would do it | 19:24 |
lambday | sonney2k: this is just one possible scenario then - we *have to* test the parameter framework and write testcases carefully | 19:25 |
@sonney2k | lambday, yeah ... it might need some split up into smaller functions | 19:26 |
lambday | sonney2k: heiko must have some ideas regarding this - I am always up for manpower wise :) | 19:29 |
lambday | sonney2k: I think I will be applying for gsoc as a student this year - this will be my last year as a student | 19:29 |
-!- pickle27 [d8bfe7da@gateway/web/freenode/ip.216.191.231.218] has joined #shogun | 19:40 | |
@sonney2k | pickle27, hey publicity chair ;) | 19:42 |
pickle27 | hey! | 19:43 |
-!- gsomix [~gsomix@188.168.2.19] has joined #shogun | 19:43 | |
pickle27 | btw sorry I didn't have a chance to setup the screen session before I left for work today | 19:43 |
pickle27 | I'll get it up this week then we can discuss next stammtisch | 19:44 |
-!- votjakovr [~votjakovr@188.134.46.30] has joined #shogun | 19:55 | |
@sonney2k | pickle27, can't you just do it now? | 20:07 |
@sonney2k | pickle27, I mean it takes like 20 seconds... | 20:08 |
pickle27 | sonney2k: the ssh key isn't on my work pc | 20:08 |
pickle27 | so if wiking can add another key for me then I can make one here | 20:09 |
pickle27 | and then do it | 20:09 |
@sonney2k | pickle27, did you twitter btw? | 20:16 |
pickle27 | I tweeted about the debian package and python3 | 20:16 |
pickle27 | Im waiting on the workshop til we have the dats | 20:16 |
pickle27 | and I'll tweet about gsoc soon too | 20:16 |
@sonney2k | pickle27, we have sth else now too | 20:19 |
pickle27 | yeah? | 20:19 |
@sonney2k | we got the osx buildbot to work with no tests failing http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python | 20:20 |
@sonney2k | so python etc | 20:20 |
@sonney2k | maybe that one http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/52 | 20:20 |
@sonney2k | it is even creating a .dmg | 20:20 |
@sonney2k | lisitsyn, lisitsyn1 did you fix the uninited memory reads? | 20:25 |
@sonney2k | lisitsyn, yandex is coming to germany - they open up sth in berlin | 20:26 |
-!- pradeepkr [0e8b7bb4@gateway/web/freenode/ip.14.139.123.180] has joined #shogun | 20:27 | |
pickle27 | sonney2k: so that builds a mgd file for osx drag and drop install for python interface? | 20:28 |
lisitsyn1 | sonney2k: yeah heard about it | 20:28 |
pickle27 | awesome! | 20:28 |
@sonney2k | pickle27, and we got a google cloud grant worth 2k$ | 20:29 |
pickle27 | awesome | 20:29 |
pickle27 | what are we using that for? | 20:29 |
lisitsyn1 | sonney2k: I am not sure about uninited stuff | 20:29 |
pickle27 | I can tweet thanking them | 20:29 |
lisitsyn1 | sonney2k: can I check somewhere? | 20:29 |
lisitsyn1 | sonney2k: would you apply? ;) | 20:34 |
@sonney2k | lisitsyn, http://buildbot.shogun-toolbox.org/memcheck/20140210-0855.html | 20:37 |
@sonney2k | pickle27, wiking and besser and lamday want to do cloud stuff with shogun, i.e. parallel comp. framework like for x-val etc | 20:38 |
pickle27 | very cool | 20:39 |
pickle27 | I'll just tweet something genric about excting cloud r & d | 20:39 |
-!- pradeepkr [0e8b7bb4@gateway/web/freenode/ip.14.139.123.180] has quit [Quit: Page closed] | 20:41 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 20:48 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:48 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 20:50 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 20:50 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has quit [Client Quit] | 20:53 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:53 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Remote host closed the connection] | 20:53 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:53 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Remote host closed the connection] | 20:53 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 20:57 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Remote host closed the connection] | 20:58 | |
lisitsyn1 | sonney2k: thanks | 20:59 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has joined #shogun | 20:59 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 21:00 | |
@iglesiasg | hey people! | 21:00 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 21:00 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Remote host closed the connection] | 21:00 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 21:00 | |
-!- Netsplit *.net <-> *.split quits: besser82, tricksy___ | 21:02 | |
@sonney2k | lisitsyn, and ? any insights? | 21:02 |
-!- sonne|osx [~sonne@e179071173.adsl.alicedsl.de] has joined #shogun | 21:02 | |
@sonney2k | shogun-buildbot, force build --branch=develop 'osx2 - python' | 21:02 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 21:02 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Client Quit] | 21:03 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 21:03 | |
-!- tricksy___ [uid12458@gateway/web/irccloud.com/x-bnbwjtszpxgvbgbc] has joined #shogun | 21:04 | |
-!- Netsplit over, joins: besser82 | 21:05 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 21:08 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 332be4b / src/shogun/mathematics/Math.cpp: https://github.com/shogun-toolbox/shogun/commit/332be4b5751978760c5e1be4c4c09db803031a6a | 21:08 |
shogun-notifier- | shogun: attempt to fix cygwin build | 21:08 |
@sonney2k | iglesiasg, hey welcome! | 21:09 |
@sonney2k | iglesiasg, could you start polishing ideas - I mean bring them in gsoc 2013 shape? | 21:09 |
@iglesiasg | sonney2k, sure | 21:09 |
@iglesiasg | let me see.. | 21:10 |
@iglesiasg | sonney2k, what about the other document we have to send for the application? | 21:11 |
@iglesiasg | I remember another one from last year we had in drive too | 21:11 |
-!- HeikoS [~heiko@nat-240-55.internal.eduroam.ucl.ac.uk] has joined #shogun | 21:14 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 21:14 | |
@HeikoS | heya | 21:14 |
@sonney2k | HeikoS, hey... | 21:14 |
@HeikoS | sonney2k: hi! how are things? | 21:15 |
@sonney2k | HeikoS, can you take care of writing the gsoc applicaton? | 21:15 |
@iglesiasg | hey man | 21:15 |
@HeikoS | sonney2k: yep can do, deadline is 14th right? | 21:15 |
@sonney2k | HeikoS, we finally have python3 tests running | 21:15 |
@HeikoS | iglesiasg: hi! | 21:15 |
@HeikoS | sonney2k: nice! | 21:16 |
@sonney2k | HeikoS, yes but lets finish of the document now and then fully focus on the ideas list | 21:16 |
@HeikoS | sonney2k: I have to redo one of my exams until the end of the week, thats why it has been quiet from my side, but at least two project descriptions are already in | 21:16 |
@HeikoS | sonney2k: EMT khan will write his tomorrow (I pushed him to that at least) | 21:16 |
@sonney2k | HeikoS, well the proposal is most important | 21:17 |
@sonney2k | HeikoS, I think we have enough ideas that we can also polish on the weekend | 21:17 |
@HeikoS | sonney2k: ok, cool | 21:17 |
@sonney2k | HeikoS, ahh and we finally got the google cloud grant | 21:17 |
@HeikoS | so application now | 21:17 |
@sonney2k | HeikoS, and I guess we will get some amazon one too | 21:17 |
@HeikoS | sonney2k: ah thats cool, whats that for? | 21:18 |
@HeikoS | sonney2k: what about templates from last years application | 21:19 |
@sonney2k | HeikoS, well we can use it a) for shogun cloud (notebooks) and b) for demos and c) for gsoc trying to run stuff in the cloud | 21:19 |
@sonney2k | HeikoS, all in publicity | 21:19 |
@HeikoS | sonney2k: ok great | 21:20 |
shogun-notifier- | shogun: Roman Votyakov :develop * 9fae7d7 / src/shogun/distributions/classical/GaussianDistribution.cpp: https://github.com/shogun-toolbox/shogun/commit/9fae7d733fda68043f9901e928a41add71305682 | 21:23 |
shogun-notifier- | shogun: fix check of the dimensions in log_pdf_multiple() method of the CGaussianDistribution class | 21:23 |
shogun-notifier- | shogun: Heiko Strathmann :develop * 44ba9aa / src/shogun/distributions/classical/GaussianDistribution.cpp: https://github.com/shogun-toolbox/shogun/commit/44ba9aa6474e975624de357dbaa2d17a19d77d72 | 21:23 |
shogun-notifier- | shogun: Merge pull request #1859 from votjakovr/develop | 21:23 |
shogun-notifier- | shogun: | 21:23 |
shogun-notifier- | shogun: fix check of the dimensions in log_pdf_multiple() method of the CGaussianDistribution class | 21:23 |
@sonney2k | HeikoS, so just to summarize what is left to do for gsoc | 21:25 |
@sonney2k | HeikoS, a) we need the application in (please go to the melange page and register again) | 21:26 |
@sonney2k | HeikoS, b) we need the ideas list and done | 21:26 |
@sonney2k | a) is urgent | 21:26 |
@sonney2k | b) is sth that we should already now start to make public in some polished form but I hope iglesiasg you can help here! | 21:27 |
@sonney2k | at the same time we should do one more release with python3 etc support | 21:27 |
@iglesiasg | sonney2k, I am working on the fundamental ML algorithms proposal now, and during the next days I will put the ideas on the website | 21:27 |
@sonney2k | and there are a couple of buildbot fixes necessary | 21:27 |
lisitsyn1 | sonney2k: need advice on possible idea | 21:28 |
@sonney2k | HeikoS, and e.g. the doc page is still obsolete | 21:28 |
@sonney2k | so the buildbot needs fixing | 21:28 |
lisitsyn1 | sonney2k: I'd like to see dictionary learning here in shogun but you know I am not a dictionary learning guy | 21:28 |
lisitsyn1 | want to know what you think | 21:28 |
@sonney2k | HeikoS, not sure if you have seen but we also now have this nice valgrind page http://buildbot.shogun-toolbox.org/memcheck/20140210-0855.html | 21:29 |
@HeikoS | sonney2k: just registered | 21:29 |
@sonney2k | lisitsyn, no idea what that means? | 21:29 |
-!- mattnorton [80e30340@gateway/web/freenode/ip.128.227.3.64] has joined #shogun | 21:29 | |
@sonney2k | HeikoS, cool thx | 21:29 |
lisitsyn1 | sonney2k: do you consider a project with no real expert mentor as a good idea? | 21:30 |
@sonney2k | lisitsyn, what is it first of all? | 21:30 |
mattnorton | Hi. I am using Enthought Canopy on Mac. Does anyone know how install shogun so it works with Canopy? | 21:30 |
@HeikoS | mattnorton: yes | 21:31 |
@sonney2k | mattnorton, no idea what canopy is... | 21:31 |
@HeikoS | just normally and add it to the python path | 21:31 |
@HeikoS | sonney2k: its just another python packaging system | 21:31 |
@HeikoS | mattnorton: just compile/install or install binary package (if available, like in linux) then add the python files to your pythonpath | 21:32 |
@HeikoS | sonney2k: ok, application is next, I will do that tomorrow and put it to discussion then | 21:32 |
lisitsyn1 | sonney2k: e.g. learning patches of images to represent image as a sparse vector | 21:33 |
lisitsyn1 | sonney2k: http://scikit-learn.org/stable/modules/decomposition.html | 21:33 |
mattnorton | ok. I used the macports method that it talks about on the website. where did it install the python files? i cannot find them? | 21:33 |
@HeikoS | mattnorton: did you compile? | 21:34 |
@HeikoS | then its in the source folder under lib/python or something | 21:34 |
-!- pranet [ca4eafc7@gateway/web/freenode/ip.202.78.175.199] has quit [Quit: Page closed] | 21:36 | |
mattnorton | Sorry. I'm a math major who is very new to programming. | 21:36 |
mattnorton | i thought that you did not need to compile if you used mac ports? | 21:36 |
mattnorton | do I type the suggested compile commands from the terminal? | 21:37 |
@HeikoS | mattnorton: dont worry, I actually am the wrong person to ask because I never used a mac ;) sonney2k maybe? | 21:37 |
mattnorton | I think it might be a better idea to email my question haha. I need specific instructions like you are directing a child haha | 21:38 |
@sonney2k | mattnorton, I am no longer using macports but brew so I cannot give 100% instructions either | 21:39 |
@sonney2k | mattnorton, just to understand - if you open up a terminal and type python | 21:39 |
@sonney2k | mattnorton, then import modshogun | 21:39 |
@sonney2k | will that work | 21:39 |
@sonney2k | ? | 21:39 |
mattnorton | "no module named modshogun" | 21:40 |
mattnorton | i downloaded the .tar file. How do you compile it? maybe that would solve the problem | 21:41 |
@HeikoS | sonney2k: its just a set of python files just like the packages under debian | 21:41 |
@HeikoS | just need to put the shogun python files in the path | 21:41 |
shogun-buildbot_ | build #53 of osx2 - python is complete: Failure [failed] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/53 | 21:41 |
@iglesiasg | sonney2k, wiking, lisitsyn, HeikoS, who had the idea of this fundamental ML algorithms project originally ? | 21:41 |
@iglesiasg | There are some things I need to know more in particular | 21:42 |
@iglesiasg | like what is it to do with kNN, PCA, and some others | 21:42 |
@HeikoS | iglesiasg: me and sonney in the desert | 21:43 |
-!- mattnorton [80e30340@gateway/web/freenode/ip.128.227.3.64] has quit [Quit: Page closed] | 21:44 | |
@HeikoS | polish all of those implementations, make them work (some dont), make them flexible, and most important, write nice examples and notebooks | 21:44 |
@HeikoS | should be fun since one learns a lot about all this fundamental stuff | 21:44 |
@iglesiasg | yeah, I agree with the fun part | 21:44 |
@iglesiasg | I think it can be a nice project | 21:44 |
@iglesiasg | let me try to get a proposal in shape | 21:45 |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 21:45 | |
@HeikoS | iglesiasg: would be great if you pushed that, I think it would be one of the most important projects for us | 21:45 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 21:45 | |
@HeikoS | since it attracts people | 21:45 |
@sonney2k | HeikoS, yeah it might be that we have to split this project into multiple | 21:50 |
@HeikoS | lambday: sorry about the PR, but that might be a bit dangerous to do since its not really solving the problem | 21:50 |
@HeikoS | sonney2k: yeah why not | 21:50 |
@HeikoS | sonney2k: I am quite excited about the projects that I added btw, good stuff | 21:50 |
@sonney2k | shogun-buildbot, force build --branch=develop 'osx2 - python' | 21:50 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 21:50 |
@HeikoS | and the other mentors are excited/motivated too | 21:50 |
@sonney2k | lisitsyn, you sure could do the project 'just' converting what is in scikits right? | 21:51 |
@sonney2k | HeikoS, lisitsyn can we remove the shogun-tutorial folder? | 21:52 |
@HeikoS | sonney2k: I dont know whether sergey translated all his stuff to notebvooks yet, lisitsyn? | 21:54 |
@sonney2k | lisitsyn, did you at least copy all text over? I want to get rid of that legacy repository | 21:56 |
-!- gxr [8ca3fe9e@gateway/web/freenode/ip.140.163.254.158] has joined #shogun | 21:57 | |
@HeikoS | leaving now guys, see you tomorrow! :) | 21:58 |
-!- HeikoS [~heiko@nat-240-55.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 21:58 | |
@sonney2k | gxr, hey gunnar! | 21:59 |
shogun-buildbot | build #1830 of cyg1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/cyg1%20-%20libshogun/builds/1830 | 21:59 |
lisitsyn1 | sonney2k: yeah sure remove it | 21:59 |
@sonney2k | lisitsyn, then please upload the notebook | 22:02 |
gxr | Hi shogun-team! | 22:03 |
gxr | (somebody came to my office and distracted me from my IRC experience...) | 22:04 |
gxr | Looks like I just missed Heiko. | 22:04 |
gxr | Well, I'm new here and I'm late as well. But I have a few questions. | 22:06 |
@iglesiasg | shoot | 22:06 |
gxr | Maybe you covered them before... | 22:06 |
shogun-buildbot | build #2190 of deb3 - modular_interfaces is complete: Failure [failed test ruby modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2190 blamelist: Soeren Sonnenburg <sonne@debian.org> | 22:06 |
gxr | 1. I suggested to have the project on "easy-install shogun". Thanks for all your suggestions and comments. I wanted to have something even more dummy proof. | 22:07 |
gxr | I think what's needed is a downloadable package that one can "activate" without any root access or extended linux knowledge. | 22:08 |
gxr | At least one suggestion included installing a specific package. This implies that one has rights to do so. Most users don't have these rights. We miss out on those. | 22:08 |
@sonney2k | gxr, yes sure halfway done already though | 22:09 |
gxr | So, can we have a tar.gz file that contains binaries including all dependencies to run shogun on a) linux (multiple distributions) b) mac (without macports & developer tools) and c) windows. | 22:10 |
@sonney2k | gxr, so the real gsoc project would be to get shogun working on windows | 22:10 |
gxr | I understood that all the solutions implied that you need to be able to install specific packages. | 22:10 |
pickle27 | bah windows | 22:10 |
pickle27 | not worth the effort | 22:11 |
gxr | Also, for mac, it would be required to have macports and developer tools. | 22:11 |
@sonney2k | gxr, well a .dmg for osx right http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/54 | 22:11 |
pickle27 | just link a tutorial on running a linux vm or something | 22:11 |
@sonney2k | gxr, we create the .dmg already but I am just fighting with getting it uploaded to the right directory | 22:11 |
gxr | (I'm fine if Windows is not supported, but linux and mac would be great, and I mean not only on Debian, but also on Centos and other distributions that are used a lot in the US). | 22:12 |
@sonney2k | gxr, also we do a nightly build on linux - we can just tar up the install and fine | 22:12 |
@sonney2k | gxr, same with fedora | 22:12 |
@sonney2k | gxr, finally I am doing real debian packages again and besser82 does rpms | 22:12 |
@sonney2k | so we can create them in the very same way too | 22:13 |
@sonney2k | but windows really is tough | 22:13 |
gxr | Can you install the rpms without root rights? What about dependencies? | 22:13 |
gxr | re vm: this won't be used if people can use other tools (that are worse, but easier to install). | 22:14 |
@sonney2k | gxr, you can extract rpms w/o root rights | 22:14 |
gxr | sonney2k, What about dependencies? | 22:15 |
@sonney2k | no chance | 22:15 |
@sonney2k | you would need to have them installed | 22:15 |
gxr | The guys in a neighboring group needed two months and lots of help to get shogun compiled on centos. | 22:15 |
gxr | Mostly because of the dependencies. | 22:15 |
@sonney2k | gxr, well it is a lot easier now with besser82's work - he packaged a couple of deps | 22:16 |
gxr | What I did a while back is to create a package that includes shogun and the dependencies. Maybe that's what besser82 did? | 22:17 |
gxr | In my case we had a build script that builds all the software including the specific atlas options that are needed etc | 22:18 |
@sonney2k | gxr, I am afraid there are very few students that can handle osx/windows/linux well enough for this task... | 22:18 |
pickle27 | Im a big fan of what our readme says currently | 22:18 |
@sonney2k | gxr, no it is a clean package now in fedora core | 22:18 |
@sonney2k | pickle27, what does it say? | 22:18 |
pickle27 | something along the lines of if you are doing any serious scientific computing you should be using linux | 22:18 |
@sonney2k | pickle27, heros of the universe? | 22:18 |
@sonney2k | pickle27, ahh that yeah that was me :) | 22:18 |
pickle27 | its great we should keep it that way | 22:19 |
pickle27 | fun fact | 22:19 |
@sonney2k | gxr, but it is easy to hack sth (not so elegant) up | 22:19 |
pickle27 | opencv dropped about half their windows pre-built binaries in their latest releases | 22:19 |
@sonney2k | pickle27, hehe | 22:19 |
gxr | I'm fine if we only support linux, but from my experience it isn't much harder to do the same on a mac. | 22:20 |
gxr | (if you have a operating system specific binary release.) | 22:20 |
pickle27 | we have mac though don't we | 22:21 |
pickle27 | I thought sonney2k announced earlier that our build is fixed | 22:21 |
gxr | Somebody with serious scripting abilities and good linux knowledge to should be able to do this. | 22:21 |
gxr | where do I find the dmg? | 22:22 |
pickle27 | ask sonney2k about it | 22:22 |
pickle27 | I think he just got it working | 22:22 |
pickle27 | I'd like to know where too cause I was gonna tweet about it | 22:22 |
@sonney2k | gxr, http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/54 | 22:23 |
@sonney2k | gxr, no idea why the upload takes so long... | 22:23 |
gxr | sonney2k, can you please send me a dmg, such I can test it on a vanilla machine? | 22:23 |
@sonney2k | gxr, but please test the dmg | 22:23 |
gxr | The link does not seem to contain a page that links to a dmg. | 22:23 |
@sonney2k | gxr, see the yellow uploading dmg thing? | 22:23 |
@sonney2k | it contains a link | 22:24 |
shogun-buildbot | build #54 of osx2 - python is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/54 | 22:24 |
shogun-buildbot | build forced [ETA 26m37s] | 22:24 |
shogun-buildbot | I'll give a shout when the build finishes | 22:24 |
gxr | sonney2k, I'll test the dmg when I have it. | 22:24 |
@sonney2k | click it | 22:24 |
@sonney2k | shogun-buildbot, is just done uploading ^ | 22:24 |
gxr | No such file 'shogun-3.1.2-Darwin.dmg'. | 22:24 |
@sonney2k | gxr, try again | 22:25 |
gxr | sonney2k, works | 22:25 |
pickle27 | awesome! | 22:26 |
gxr | looks great. But, I need admin rights to install the package... | 22:26 |
gxr | reduces the number of users by a large factor. | 22:26 |
gxr | sonney2k, it is still much better than having to install developer tools and macports first. Hence, this is a huge improvement. | 22:27 |
gxr | great work! | 22:27 |
pickle27 | sonney2k: I'll look into adding builds to the website when Im done with other things | 22:28 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 22:28 | |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has joined #shogun | 22:28 | |
gxr | Btw, I have a intel mac (core duo) in a drawer unused. | 22:29 |
@sonney2k | gxr, well we could also upload a .tar.gz for osx that one can extract | 22:29 |
gxr | However, one can only update the OS to 10.6 or so. | 22:29 |
@sonney2k | gxr, did the dmg install sth? I mean could you run shogun & python stuff? | 22:30 |
@sonney2k | I never tried... | 22:30 |
gxr | sonney2k, tar.bz ball would be a useful addition. | 22:30 |
gxr | The package installed something. Just don't know where to. | 22:30 |
gxr | [MSKI1327 4] /usr/include/shogun > ls base/ distributions/ kernel/ machine/ optimization/ transfer/ classifier/ ensemble/ labels/ mathematics/ preprocessor/ ui/ clustering/ evaluation/ latent/ metric/ regression/ converter/ features/ lib/ modelselection/ statistics/ distance/ io/ loss/ multiclass/ structure/ | 22:32 |
@sonney2k | gxr, I guess /usr/local/* | 22:32 |
@sonney2k | gxr, looks good and does sth like python | 22:32 |
@sonney2k | gxr, import modshogun work? | 22:32 |
-!- lambday [67157f4f@gateway/web/freenode/ip.103.21.127.79] has quit [Ping timeout: 245 seconds] | 22:33 | |
gxr | sonney2k, no. Installed headers and libs, but I cannot find python packages | 22:34 |
@sonney2k | gxr, does anything like /usr/local/Cellar* exist? | 22:35 |
gxr | sonney2k, no. | 22:35 |
gxr | Ok, maybe this goes into too much detail here... | 22:35 |
gxr | I understand there is lot's of effort already done, but that last bit to pull it all together to something that people can just "click" or run that basic script and have shogun running, that is missing and that is something that needs somebody to spend time on. | 22:36 |
gxr | I've done this before and it takes time to do this right. | 22:36 |
gxr | That's the idea of the project that I had in mind. | 22:37 |
gxr | The solution for the "standard university student" with limited access and in a university determined environment. | 22:38 |
-!- votjakovr [~votjakovr@188.134.46.30] has quit [Quit: WeeChat 0.4.0] | 22:38 | |
gxr | Any more comments? If not, question two. | 22:39 |
shogun-buildbot_ | build #55 of osx2 - python is complete: Failure [failed test python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/55 | 22:39 |
@sonney2k | gxr sure all clear | 22:39 |
gxr | I got somebody from my group to commit to supervise clustering algorithms. | 22:40 |
gxr | But this needs a person to co-supervise who knows shogun better. | 22:41 |
gxr | (sorry, this is about GSoC). | 22:41 |
gxr | Are there any takers? | 22:41 |
@sonney2k | gxr, depends on how involved these are... | 22:42 |
@sonney2k | gxr, and that person certainly needs to be able to run shogun and check if some example does what is intended | 22:43 |
shogun-buildbot_ | build #2191 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2191 | 22:44 |
gxr | sonney2k, well, they have a c++ reference implementation, they can learn how to run shogun, they can supervise the ML side and how to check whether the implementation works and they can work with the gsoc student to get the algorithm worked out better. | 22:44 |
@iglesiasg | I would not mind being the "Shogun mentor" for a project like that one if we don't get anyone for the project I am proposing to mentor | 22:44 |
gxr | sonney2k, that's not the case now, but the person can easily learn that. | 22:44 |
@sonney2k | gxr, then sure | 22:46 |
gxr | iglesiasg, ok, great. This is what's needed here. (Assuming it is ok to offer more projects that we can do in parallel at the same time). | 22:46 |
@sonney2k | gxr, what we did last year is always have a domain expert and a shogun mentor | 22:46 |
gxr | sonney2k, yes, I think that's a good idea. | 22:46 |
@sonney2k | in the end result one made sure the algorithm does what it is supposed to do and came up with some nice examples the student could describe in a ipython notebook | 22:46 |
@sonney2k | and the other did ensure that it fit into the framework coding style, basic unit tests etc | 22:47 |
gxr | sonney2k, just wasn't sure how to proceed if there are no takers for the proposed project (I did not get a response to two of the three projects that I have ML mentors for). | 22:47 |
@sonney2k | well everyone is busy.... | 22:47 |
gxr | sonney2k, ok, understand. When is the proposal due? | 22:48 |
gxr | sonney2k, can we then agree to add two of the three projects to the GSoC idea list? | 22:49 |
gxr | (actually, I got responses for two out of the four projects I proposed). | 22:49 |
-!- besser82 [quassel@fedora/besser82] has quit [Remote host closed the connection] | 22:50 | |
@sonney2k | gxr, well the proposal is due friday | 22:51 |
@sonney2k | gxr, the ideas list is sth we can still polish but we should have some version already even then | 22:51 |
@iglesiasg | gxr, it feels like this year we have many proposals supported by domain expert mentors, but not that many mentors that know a little bit about Shogun | 22:51 |
@iglesiasg | gxr, but in my opinion we can add all those proposals, and then see what students are more interested in and decide which projects should be preferred/are feasible to do taking into account mentor power | 22:52 |
gxr | iglesiasg, sounds like a good plan. | 22:54 |
@sonney2k | iglesiasg, indeed... simply because lisitsyn is no longer a student and wiking is finishing is phd etc etc | 22:55 |
gxr | iglesiasg, I'd think that if the student isn't adding new infrastructure to shogun, but just a new algorithm, then this should be doable without deep knowledge of shogun. | 22:55 |
shogun-buildbot_ | build #187 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/187 blamelist: Roman Votyakov <votjakovr@gmail.com> | 22:56 |
shogun-buildbot_ | build #56 of osx2 - python is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/osx2%20-%20python/builds/56 | 22:56 |
gxr | btw, I'll ask for money in the NCI cloud initiative for a programmer (half a year or so). | 22:56 |
@iglesiasg | gxr, it would also be great to get the ideas in a format similar to the ones in http://shogun-toolbox.org/page/Events/gsoc2013_ideas | 22:56 |
@sonney2k | iglesiasg, hurray http://www.shogun-toolbox.org/doc/en/latest/ | 22:57 |
@sonney2k | I finally fixed the doc upload - I hope it will work again at least ;) | 22:57 |
@iglesiasg | great! :) | 22:57 |
gxr | :-) | 22:57 |
gxr | sonney2k, what's the status regarding the BSD license stuff? | 22:58 |
-!- shogun-buildbot_ [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 22:58 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 22:58 | |
@sonney2k | gxr, wiking has some page I don't remember | 22:59 |
@iglesiasg | gxr, sonney2k there you go http://maeth.com/shogun_licensing_stats.php | 22:59 |
@sonney2k | gxr, maybe 40% green | 22:59 |
gxr | Soeren Sonnenburg (sonne@debian.org)NOT SIGNED | 22:59 |
gxr | Are you in or not? | 22:59 |
@iglesiasg | :DD | 22:59 |
@iglesiasg | I am more worried about the rest though, we can always ping sonney2k here :) | 23:01 |
gxr | Is there a way to sort by number of contributed lines? | 23:01 |
@iglesiasg | gxr, there is something in github for that, but there we don't have the signed/not signed thing | 23:01 |
gxr | iglesiasg, yes, but we have to make sure that the people who contributed most code lines do agree. The bottom 50% probably contributed only a few %. | 23:02 |
gxr | this one is also problematic: Sebastian Henschel (shogun@kodeaffe.de)NOT SIGNED | 23:03 |
@sonney2k | gxr, well he did his work under contract of mpg etc so I don't see a problem with sebastian anyways | 23:04 |
@sonney2k | gxr, I am more concerned about what we do with fhg | 23:05 |
gxr | sonney2k, ok, good re sebastian. FhG FIRST was closed. | 23:05 |
gxr | So, nobody will care or know. | 23:06 |
shogun-buildbot | build #188 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/188 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 23:07 |
gxr | To play safe you can ask a lawyer about it... but I would suggest not to ask the FhG lawyers. | 23:07 |
gxr | We could have released the software as BSD software in the days. Nobody would have minded then. | 23:08 |
gxr | For some reason we choose GPL. But BSD would have been a option as well. | 23:08 |
gxr | Ok, looks like we should talk about this aspect some more... maybe a call would be better for this. | 23:09 |
gxr | for the NCI cloud. do you think the shogun foundation can sign me a letter that there is an effort to shift the license to a non-viral one? That's what's needed to get funding for work on shogun by a programmer. | 23:11 |
@sonney2k | gxr, just ask HeikoS - he is all in and we are obviously making progress towards this. | 23:12 |
gxr | Ok, will talk to HeikoS then. | 23:12 |
-!- besser82 [quassel@fedora/besser82] has joined #shogun | 23:12 | |
gxr | (sonney2k, btw, will be in Berlin around easter. Will follow-up by email re meeting on easter monday) | 23:13 |
gxr | What's needed to get the proposal done? | 23:14 |
@sonney2k | gxr, I hope there is some more incentive though like e.g. something joint with scikits | 23:14 |
gxr | I have limited time, but if there is some urgencies or something specific I can do, then I'll try to help. | 23:14 |
@sonney2k | gxr, well Heiko will do the proposal | 23:14 |
@sonney2k | hopefully send around a draft | 23:15 |
@sonney2k | tomorrow | 23:15 |
@sonney2k | so doing an iteration would help | 23:15 |
gxr | sonney2k, don't understand the scikits reference. | 23:15 |
@sonney2k | the rest is on the way I would say | 23:15 |
@sonney2k | gxr, ahh we want to organize another workshop just after europython in berlin | 23:15 |
gxr | Ok. not sure I can make it. | 23:16 |
@sonney2k | gxr, well some kind of collaboration with them would be nice. they have all the basic algorithms but miss some advanced stuff | 23:16 |
gxr | (will try though). | 23:16 |
gxr | ok, re joint meeting. | 23:17 |
@sonney2k | gxr, july 28th would be the plan | 23:17 |
gxr | sonney2k, but I have some concerns regarding too much python in shogun. My take on this is that shogun is great and one important reason is that it can be used from many different languages and hence can reach much more people. | 23:18 |
@sonney2k | gxr, well python is *the* language of scientific computing nowadays | 23:19 |
@sonney2k | all the rest has become pretty insignificant | 23:19 |
gxr | Well, that's an opinion that not everybody will agree on. | 23:20 |
@sonney2k | but yes it is nice to have pure c++ programs and java | 23:20 |
@sonney2k | and there is little to nothing for octave except shogun. same holds for ruby etc | 23:20 |
gxr | R, java, octave/matlab ... | 23:20 |
@sonney2k | matlab doesn't even work atm I think - just because no one here has it | 23:21 |
gxr | R is a case which has lot's of computation. Combio people appear to use more R and less python. | 23:21 |
@sonney2k | true R is also pretty popular | 23:22 |
@sonney2k | but python is a different league | 23:22 |
shogun-buildbot | build #182 of clang34 - thread analysis is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/clang34%20-%20thread%20analysis/builds/182 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Roman Votyakov <votjakovr@gmail.com> | 23:22 |
@sonney2k | R modular works now with shogun but sometimes the error messages are weird | 23:23 |
gxr | R support is rather limited and much less tested than python. To get people to use shogun in R is to have a bioconductor package. | 23:23 |
gxr | pubmed finds ≈500 publications on "bioconductor" and ≈1200 papers with "python". (which isn't quite a fair comparison as I cannot use the broader term "R" for the search) | 23:24 |
gxr | So, yes, python is somewhat more popular, but I wouldn't say the difference is huge. | 23:25 |
gxr | (in the field of "data scientists", not computer science or engineering in general) | 23:26 |
@sonney2k | gxr, well R works the way it can with swig so we cannot really do much | 23:26 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 23:27 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 23:27 | |
gxr | Not sure what the state is here, but from what I heard is that the swig/R interface isn't stable enough for serious use. | 23:27 |
gxr | Am I wrong her? | 23:28 |
@sonney2k | gxr, for sure there is no way around shogun if you want to use it from multiple languages | 23:28 |
@sonney2k | gxr, well all the shogun examples just run fine... | 23:28 |
gxr | I understand that there are memory management issues. | 23:28 |
@sonney2k | gxr, http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2191/steps/test%20r%20modular/logs/stdio | 23:29 |
@sonney2k | gxr, not that I know of | 23:29 |
@sonney2k | we only have 54 examples but still | 23:29 |
gxr | ok, maybe then this is working ok now and I'm out-of-date (quite possibly). | 23:29 |
@sonney2k | gxr, yeah we had these problems maybe until 1 year back | 23:30 |
@sonney2k | but it is ok now | 23:30 |
gxr | getting a bioconductor package would increase usage of shogun a lot. | 23:30 |
gxr | but this requires to have the package available on multiple platforms. I believe one is windows. | 23:30 |
@sonney2k | gxr, very tough | 23:31 |
gxr | And the acceptance criteria are tough | 23:31 |
gxr | Is this something a GSoC student could do? | 23:31 |
@sonney2k | gxr, I think already a win32 port is very tough | 23:32 |
gxr | (tough here mostly means, to do all the leg work, fix warnings, getting the packages to compile etc). | 23:32 |
gxr | Hm, maybe too tough. | 23:32 |
@sonney2k | yeah it might mean not to get it to work even | 23:32 |
@sonney2k | wiking tried and failed | 23:33 |
@sonney2k | but of course no one had the time to try it for 3 months straight | 23:33 |
gxr | Can we break it into pieces and come up with a plan to have it in 2-3 years? | 23:33 |
gxr | So, for instance, first only lib-shogun with R-static basic functionality. | 23:34 |
@iglesiasg | All right guys, I abandon for today. Tty again soon! | 23:34 |
@sonney2k | iglesiasg, see you around! | 23:35 |
@sonney2k | gxr, yes first libshogun | 23:35 |
-!- iglesiasg [~iglesiasg@524AE0A7.cm-4-3d.dynamic.ziggo.nl] has quit [Quit: Leaving] | 23:35 | |
gxr | see you! | 23:35 |
@sonney2k | gxr, the rest will be easy I suspect | 23:35 |
gxr | so, is this something one should try as a "hard" GSoC project? | 23:36 |
gxr | (I personally think it would be worth trying, to get all these R people to use and like shogun) | 23:37 |
@sonney2k | gxr, the windows port yes | 23:38 |
@sonney2k | gxr, alright I am too tired too | 23:39 |
gxr | ok, good. was great irc'ing. Sorry that I was late... | 23:39 |
gxr | Will try to join earlier next time. | 23:39 |
@sonney2k | gxr, thanks for joining | 23:39 |
gxr | thanks for organizing the meeting! | 23:40 |
gxr | CU | 23:40 |
-!- gxr [8ca3fe9e@gateway/web/freenode/ip.140.163.254.158] has left #shogun [] | 23:40 | |
@sonney2k | cu | 23:40 |
-!- FSCV [~FSCV@50.7.50.60] has joined #shogun | 23:56 | |
-!- FSCV [~FSCV@50.7.50.60] has quit [Remote host closed the connection] | 23:57 | |
-!- pickle27 [d8bfe7da@gateway/web/freenode/ip.216.191.231.218] has quit [Ping timeout: 245 seconds] | 23:59 | |
--- Log closed Tue Feb 11 00:00:53 2014 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!