--- Log opened Wed Aug 07 00:00:04 2013 | ||
@wiking | sonney2k: btw: if we would use cdash we could keep track on our buildbots how long does it take to run an example. this could be a good performance indicator as well ;) | 00:02 |
---|---|---|
@wiking | sonney2k: http://open.cdash.org/testDetails.php?test=202407706&build=2990014 | 00:05 |
@wiking | and click on Show Test Time Graph | 00:05 |
@sonney2k | heh | 00:06 |
@wiking | we could of course generate this somehow on our own | 00:06 |
@sonney2k | wiking, I am fine with gtest' ascii | 00:06 |
@wiking | just needs some postprocessing on buildbot | 00:06 |
@sonney2k | wiking, yeah true probably very easy | 00:07 |
@wiking | as we can generate | 00:07 |
@wiking | everything in xml | 00:07 |
@wiking | which is pretty easy to parse | 00:07 |
@sonney2k | wiking, well lets land on planet earth | 00:07 |
@sonney2k | and get things working first | 00:07 |
@sonney2k | alrighty | 00:10 |
@sonney2k | bed time! | 00:10 |
@iglesiasg | good night sonney2k | 00:13 |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun | 00:15 | |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 00:20 | |
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/9918747 | 00:20 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 00:20 | |
shogun-buildbot | build #1674 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1674 blamelist: Soeren Sonnenburg <sonne@debian.org> | 00:21 |
@iglesiasg | ok, good night to all | 00:39 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 00:40 | |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 00:49 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9919124 | 00:49 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 00:49 | |
-!- HeikoS [~heiko@05419c83.skybroadband.com] has joined #shogun | 00:50 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:50 | |
-!- HeikoS [~heiko@05419c83.skybroadband.com] has quit [Remote host closed the connection] | 00:53 | |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 78a5a6b / / (4 files): https://github.com/shogun-toolbox/shogun/commit/78a5a6bfa48a7163da7e2ee6820434bab581a5b8 | 02:27 |
shogun-notifier- | shogun: Switch octave build to gcc | 02:27 |
shogun-notifier- | shogun: upgrade gmock to HEAD of trunk to support c++11 | 02:27 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 02:35 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9923751 | 02:35 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 02:35 | |
gsomix_ | sonney2k, sonne|work ok, I'll check it. | 03:41 |
gsomix_ | sonney2k, sonne|work can you give me example file with data? | 03:53 |
shogun-buildbot | build #416 of nightly_all is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_all/builds/416 | 03:53 |
gsomix_ | sonney2k, sonne|work cu at evening in chat | 03:54 |
-!- gsomix_ [~gsomix@95.67.140.176] has quit [Quit: Leaving] | 03:55 | |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has quit [Ping timeout: 276 seconds] | 04:45 | |
shogun-buildbot | build #481 of nightly_default is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/481 | 04:46 |
shogun-buildbot | build #426 of nightly_none is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_none/builds/426 | 04:47 |
-!- shogun-notifier- [~irker@94.23.237.10] has quit [Quit: transmission timeout] | 05:27 | |
-!- gsomix [~gsomix@185.6.239.105] has joined #shogun | 07:36 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 08:42 | |
shogun-notifier- | shogun: Soeren Sonnenburg :develop * 44e3c2e / examples/ (3 files): https://github.com/shogun-toolbox/shogun/commit/44e3c2e2b6de9e8c8cdcdc9bff74d32b2d03caf2 | 08:42 |
shogun-notifier- | shogun: convert cauchy/chi2 kernel to use CSVFile | 08:42 |
-!- van51 [~van51@ppp-94-66-20-32.home.otenet.gr] has joined #shogun | 09:09 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has joined #shogun | 09:10 | |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 09:14 | |
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/9931207 | 09:14 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 09:14 | |
@wiking | btw: anybody thought about adding support for std::complex? | 09:27 |
sonne|work | wiking: errm we have complex now - lambday did this | 09:28 |
@wiking | yes but it's std:: ;) | 09:28 |
-!- Yanglittle [b74040fc@gateway/web/freenode/ip.183.64.64.252] has joined #shogun | 09:29 | |
sonne|work | ? | 09:30 |
Yanglittle | excuse me, does the svm can return the score of classifying now? | 09:36 |
sonne|work | Yanglittle: sure it always did this | 09:37 |
sonne|work | Yanglittle: just use labels.get_values() IIRC | 09:37 |
shogun-buildbot | build #1675 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1675 blamelist: Soeren Sonnenburg <sonne@debian.org> | 09:37 |
-!- foulwall [~user@106.186.27.65] has joined #shogun | 09:38 | |
Yanglittle | labels.get_values(), here labels is ? | 09:41 |
-!- foulwall [~user@106.186.27.65] has quit [Remote host closed the connection] | 09:41 | |
-!- foulwall [~user@106.186.27.65] has joined #shogun | 09:41 | |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 2a7f045 / CMakeLists.txt,cmake/CheckCXX11Features.cmake: https://github.com/shogun-toolbox/shogun/commit/2a7f045280ea228dbab78cf9dff1e8924f966e9d | 09:41 |
shogun-notifier- | shogun: Use -stdlib=libc++ only on Darwin systems for clang c++11 support | 09:41 |
thoralf | Hey all | 09:42 |
sonne|work | Yanglittle: the labels you get from an svm | 09:42 |
sonne|work | thoralf: goard morning! | 09:42 |
thoralf | sonne|work: Nice accent ;) | 09:42 |
gsomix | sonney2k, hey | 09:43 |
gsomix | sonne|work, ^ | 09:43 |
Yanglittle | like this: svm.apply().get_labels().get_values()? it doesn't work. here it's a multiclasssvm | 09:44 |
lisitsyn | Yanglittle: you need get_multiclass_confidences(int) method | 09:54 |
lisitsyn | the parameter in parens is the index of the feature vector you need | 09:54 |
lisitsyn | so that if you need confidences for 0th feature vector | 09:54 |
lisitsyn | you do | 09:54 |
lisitsyn | labels.get_multiclass_confidences(0) | 09:55 |
lisitsyn | it will return a vector of size #(number of classes) | 09:55 |
lisitsyn | with corresponding scores | 09:55 |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has joined #shogun | 09:58 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9932411 | 09:58 |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has left #shogun [] | 09:58 | |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 85677ec / .travis.yml: https://github.com/shogun-toolbox/shogun/commit/85677ecc82d8891a00ac5e41f9e2840e9f908134 | 10:05 |
shogun-notifier- | shogun: Travis: Force to use clang/clang++ when compiler is clang | 10:05 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 10:09 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 10:09 | |
@iglesiasg | good morning | 10:10 |
thoralf | Hey iglesiasg | 10:10 |
@wiking | iglesiasg: mornin' .. i've made your eigen fetcher/compiler | 10:10 |
@wiking | sonne|work: have u seen this before: https://travis-ci.org/shogun-toolbox/shogun/jobs/9933019#L2086 | 10:10 |
@iglesiasg | wiking, all good with feature/CMake? | 10:10 |
@wiking | iglesiasg: yep... there's some problem with gcc + c++11 it seems though | 10:11 |
@wiking | oooh fuck | 10:12 |
@wiking | c++11 vs python modular :( | 10:12 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/jobs/9933021#L2760 | 10:13 |
@wiking | is_array vs std:is_array | 10:14 |
@iglesiasg | wtf | 10:14 |
@wiking | well there's is std::is_array in c++11 | 10:14 |
lisitsyn | haha | 10:15 |
lisitsyn | ::is_array should help? | 10:15 |
@wiking | but there's a static function as well in src/interfaces/python_modular/swig_typemaps.i | 10:15 |
@wiking | lisitsyn: let's see ;) | 10:15 |
@wiking | ::is_array would refer to the static function or? | 10:15 |
lisitsyn | wiking: would refer to the name in the current namespace | 10:16 |
lisitsyn | so he could recognize it is not std | 10:16 |
lisitsyn | I hope so! | 10:16 |
@iglesiasg | has using namespace being used before? | 10:16 |
@iglesiasg | namespace std | 10:16 |
lisitsyn | it won't be ambiguous anymore | 10:16 |
lisitsyn | at least it compiles in my head | 10:16 |
@wiking | lisitsyn: what about the other ? :) | 10:19 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/jobs/9933019#L2086 | 10:19 |
@wiking | ideas? | 10:19 |
lisitsyn | wiking: that 'new' thing? | 10:20 |
@wiking | yep | 10:20 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 8425ef4 / .travis.yml,src/interfaces/python_modular/swig_typemaps.i: https://github.com/shogun-toolbox/shogun/commit/8425ef4ceb8d5ec56fb04de164d7767d17c53a7f | 10:20 |
shogun-notifier- | shogun: Fix is_array ambiguity with std::is_array in c++11 mode | 10:20 |
shogun-notifier- | shogun: Travis: Fix forcing clang as a compiler in all clang target | 10:20 |
lisitsyn | wiking: sonney2k should know about that | 10:20 |
lisitsyn | he fixed it once | 10:20 |
-!- Yanglittle [b74040fc@gateway/web/freenode/ip.183.64.64.252] has quit [Quit: Page closed] | 10:20 | |
@wiking | lately? | 10:21 |
@wiking | since i rebased yesterday some time | 10:21 |
lisitsyn | like 2 weeks ago | 10:21 |
@iglesiasg | why is it that we have CFeatures::duplicate that needs to be implemented for every subclass? I think it is the same that CSGObject::clone does. | 10:22 |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has joined #shogun | 10:23 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9933018 | 10:23 |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has left #shogun [] | 10:23 | |
@wiking | iglesiasg: mmm i gues it hsould be marked as deprecated...? | 10:23 |
@wiking | we anyways should start deprecating some stuf... | 10:24 |
@iglesiasg | wiking, yeah maybe. I was not sure if I was missing something special about dupplicate | 10:24 |
@iglesiasg | yeah, agree | 10:24 |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has joined #shogun | 10:26 | |
@wiking | lisitsyn: https://travis-ci.org/shogun-toolbox/shogun/jobs/9933395#L2781 | 10:28 |
@wiking | :D | 10:28 |
lisitsyn | wiking: wtf | 10:34 |
lisitsyn | wiking: what happens here | 10:34 |
hushell | iglesiasg: you are talking about FactorGraph::duplicate :) Anyway, it never be used seems | 10:35 |
-!- foulwall` [~user@106.186.27.65] has joined #shogun | 10:35 | |
@wiking | lisitsyn: afaik it's something with -Wc++11-narrowing | 10:35 |
-!- foulwall [~user@106.186.27.65] has quit [Remote host closed the connection] | 10:35 | |
@iglesiasg | hushell, not just about the CFactorGraph one but in general why CFeatures even has that method since to me it seems that it does the same functionality of clone in CSGObject | 10:35 |
hushell | iglesiasg: indeed it is, but I think it was designed that more things could be done together with the duplication of features | 10:38 |
@iglesiasg | hushell, mm sorry, I don't understand that. Designed so more things could be done together? What things? | 10:38 |
hushell | iglesiasg: because I never see this function being called, so I have no idea, I just guess | 10:39 |
hushell | iglesiasg: how is my PR? | 10:41 |
@iglesiasg | hushell, pretty much ready I think. I wrote a couple of comments, did you see them? | 10:41 |
@iglesiasg | hushell, the most important thing is about find_set. You said you have tested it, are these tests you mentioned already in the unit test? | 10:41 |
@iglesiasg | I can see in the unit test something about loopy structure analysis, but neither tree nor forest | 10:42 |
@iglesiasg | that you mentioned in the comment | 10:42 |
hushell | iglesiasg: yeah, I added 2 more tests | 10:42 |
-!- foulwall` [~user@106.186.27.65] has quit [Read error: Connection reset by peer] | 10:42 | |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 10:46 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9933391 | 10:46 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 10:46 | |
hushell | iglesiasg: the structure_analysis_loopy got a loopy graph, 3x3 grid graph indeed | 10:47 |
@iglesiasg | hushell, good | 10:48 |
hushell | and the structure_analysis_disconnected got 3 trees in the graph, which is actually 3 lines | 10:48 |
hushell | it works better than my expectation :) | 10:48 |
@iglesiasg | cool! | 10:49 |
shogun-notifier- | shogun: Fernando Iglesias :develop * d4e02ed / src/shogun/base/SGObject.h: https://github.com/shogun-toolbox/shogun/commit/d4e02ed01f497e16832fd5f9bdcb01520b198ff9 | 10:49 |
shogun-notifier- | shogun: Fix a couple of doc typos in CSGObject | 10:49 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 41654d9 / src/shogun/base/SGObject.h: https://github.com/shogun-toolbox/shogun/commit/41654d9769e982376406b15928b5f3706e262279 | 10:49 |
shogun-notifier- | shogun: Merge pull request #1375 from iglesias/develop | 10:49 |
shogun-notifier- | shogun: | 10:49 |
shogun-notifier- | shogun: Fix a couple of doc typos in CSGObject | 10:49 |
hushell | iglesiasg: yeah, some progresses finally | 10:50 |
@iglesiasg | hushell, let's merge it then? | 10:52 |
shogun-buildbot | build #1676 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1676 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 10:53 |
-!- foulwall` [~user@106.186.27.65] has joined #shogun | 10:54 | |
hushell | iglesiasg: if you think is good to merge, I'll send another PR before sleeping | 10:54 |
shogun-notifier- | shogun: hushell :develop * 8adb4f0 / / (8 files): https://github.com/shogun-toolbox/shogun/commit/8adb4f01d7f4d6d4b66e1f1f59286bb357990d5c | 10:55 |
shogun-notifier- | shogun: FactorGraph structure analysis | 10:55 |
shogun-notifier- | shogun: Fernando Iglesias :develop * 5a8268d / / (8 files): https://github.com/shogun-toolbox/shogun/commit/5a8268d7a20944eb1411bdaf911a4f56c8086da0 | 10:55 |
shogun-notifier- | shogun: Merge pull request #1369 from hushell/fg_dev | 10:55 |
shogun-notifier- | shogun: | 10:55 |
shogun-notifier- | shogun: FactorGraph structure analysis | 10:55 |
hushell | iglesiasg: for the connect_components(), any suggestion to make it safer? | 10:57 |
@iglesiasg | hushell, mmmm is there any way to know if it has been called? | 10:58 |
@iglesiasg | hushell, does the state of the CFactorGraph object change in particular when this function is called? | 10:59 |
shogun-buildbot | build #1678 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1678 blamelist: hushell <hushell@hushell-U510.(none)> | 10:59 |
shogun-buildbot | build #1679 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1679 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 11:02 |
@wiking | /usr/include/python2.7/numpy/npy_deprecated_api.h:11:2: warning: #warning "Using deprecated NumPy API, disable it by #defining NPY_NO_DEPRECATED_API NPY_1_7_API_VERSION" [-Wcpp] | 11:02 |
hushell | iglesiasg: I forgot I did something for this | 11:03 |
hushell | iglesiasg: I made a flag is_connected in CDisjointSet | 11:03 |
hushell | once the graph is updated, we will know | 11:03 |
sonne|work | wiking: yes I know that. it is easy to get rid of with that define but we don't know if we use the intended numpy api so maybe we don't want to do that just yet | 11:03 |
hushell | and I add a note to the documentation | 11:04 |
@iglesiasg | hushell, good! | 11:04 |
@wiking | sonne|work: and what about the other error (gcc) | 11:05 |
@wiking | sonne|work: i've just tested the same with gcc 4.7 and it worked fine... | 11:05 |
@wiking | maybe we should just drop c++11 for gcc that is < 4.7 | 11:05 |
sonne|work | wiking: which other error? | 11:05 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/jobs/9933401#L2141 | 11:06 |
@wiking | and i think maybe we should fix this https://travis-ci.org/shogun-toolbox/shogun/jobs/9933395#L2781 | 11:06 |
sonne|work | wiking: but that is fixed in develop already | 11:07 |
@wiking | sonne|work: this is a rebase of yesterday | 11:08 |
@wiking | but yeah that python error could be fixed with -Wnoc++11-narrowing | 11:09 |
-!- foulwall` [~user@106.186.27.65] has quit [Remote host closed the connection] | 11:09 | |
@wiking | but actually really it'd be better to fix it in a way that it doesn't do an unsigned long -> long conversion | 11:10 |
@wiking | or in worst case we could do a static_cast<Py_ssize_t>( as it's said there in the error | 11:11 |
@wiking | sonne|work: i dont get that memory problem with gcc 4.7 | 11:11 |
@wiking | sonne|work: this is only travis' gcc 4.6 + c++11 enabled | 11:11 |
sonne|work | wiking: then I guess HAVE_CXX11 is not used? | 11:13 |
sonne|work | defined I mean | 11:13 |
-!- sonne|work [~sonnenbu@91-64-72-127-dynip.superkabel.de] has quit [Quit: Leaving.] | 11:20 | |
-!- sonne|work [~sonnenbu@91-64-72-127-dynip.superkabel.de] has joined #shogun | 11:20 | |
@wiking | sonne|work: it is defined... that causes the problem | 11:22 |
@wiking | sonne|work: gcc 4.6 + HAVE_CXX11 breaks hell | 11:22 |
sonne|work | ohh man and with gcc 4.4 it causes trouble when HAVE_CXX11 is not there | 11:23 |
sonne|work | excellent | 11:23 |
@wiking | mmm | 11:23 |
@wiking | asdf | 11:24 |
@wiking | - Performing Test _HAS_CXX11_FLAG | 11:24 |
@wiking | -- Performing Test _HAS_CXX11_FLAG - Failed | 11:24 |
@wiking | -- Performing Test _HAS_CXX0X_FLAG | 11:24 |
@wiking | -- Performing Test _HAS_CXX0X_FLAG - Success | 11:24 |
@wiking | -- Checking C++11 support for "__func__" | 11:24 |
@wiking | -- Checking C++11 support for "__func__": works | 11:24 |
@wiking | -- Checking C++11 support for "atomic" | 11:24 |
@wiking | -- Checking C++11 support for "atomic": works | 11:24 |
@wiking | i dont even know how it is possible to not TO have _HAS_CXX11_FLAG but have aotmic | 11:24 |
@wiking | *atomic | 11:24 |
@wiking | gotta debug this | 11:25 |
@wiking | sonne|work: ah ok i remember... as i wanted to tell you before HAVE_CXX11 != HAVE_ATOMIC... | 11:28 |
sonne|work | yeah sounds impossible | 11:29 |
@wiking | sonne|work: what? | 11:29 |
sonne|work | I mean well ok maybe yes | 11:29 |
@wiking | i mean not maybe | 11:30 |
sonne|work | I know that clang people gradually implemented c++11 support | 11:30 |
@wiking | it's for sure | 11:30 |
@wiking | even in gcc | 11:30 |
sonne|work | so they had atomic working long before the whole standard got implemented | 11:30 |
sonne|work | but this here is still strange I mean we use -std=c++11 for the atomic test right? | 11:31 |
sonne|work | and the _HAS_CXX* is sth compiler specific? | 11:31 |
@wiking | no | 11:32 |
@wiking | it's created by cmake test | 11:32 |
sonne|work | ahh ok then makes sense | 11:34 |
sonne|work | so it needs a different treatment in memory.cpp | 11:34 |
gsomix | sonne|work, please, say me what example you use for test? | 11:34 |
@wiking | sonne|work: can we have atomic w/o -std=c++11 ? | 11:35 |
gsomix | I mean 92x92 data | 11:35 |
@wiking | sonne|work: i dont see it happening | 11:35 |
sonne|work | gsomix: well the unit tests fail | 11:35 |
gsomix | sonne|work, of course, we changed order. | 11:36 |
sonne|work | gsomix: it is very confusing btw with the ascii / sgvector etc stuff | 11:36 |
sonne|work | gsomix: no | 11:36 |
sonne|work | gsomix: yes | 11:36 |
sonne|work | o_O | 11:36 |
@wiking | ah oookeeey | 11:36 |
@wiking | i works with -std=c++0x ;) | 11:36 |
sonne|work | gsomix: so basically we have a clash with everything | 11:36 |
gsomix | sonne|work, lol | 11:36 |
sonne|work | wiking: asdfasdfasdfasdf | 11:37 |
@wiking | sonne|work: ? :) | 11:37 |
sonne|work | gsomix: so basically we have | 11:37 |
gsomix | sonne|work, I don't realyy figure out why data in examples is in c_order. | 11:37 |
@wiking | sonne|work: i think i solved it :P | 11:37 |
sonne|work | a) an SGVector and want to convert it into a SGMatrix | 11:37 |
sonne|work | b) an ascii file that you convert into an SGVector and then a SGMatrix | 11:38 |
sonne|work | and now the fun begins: | 11:38 |
sonne|work | when you have an sgvector -> sgmatrix | 11:39 |
sonne|work | we have C and FORTRAN order | 11:39 |
gsomix | sonne|work, don't stop writing. I need switch to smartphone. | 11:39 |
sonne|work | with fortran order (which is what shogun uses internally everywhere) | 11:39 |
-!- gsomix [~gsomix@185.6.239.105] has quit [Quit: Leaving] | 11:39 | |
sonne|work | errm | 11:39 |
lisitsyn | sonne|work: keep writing! | 11:40 |
lisitsyn | :D | 11:40 |
sonne|work | we want that column vectors are examples | 11:40 |
sonne|work | so one column == one example | 11:40 |
sonne|work | and we concatenate them in memory since that is how we store the matrix | 11:40 |
sonne|work | so basically we don't have to do anything with sgvector | 11:40 |
sonne|work | just copy the data over for *FORTRAN* order | 11:41 |
sonne|work | for C order we need to transpose | 11:41 |
sonne|work | gsomix that is how you wrote it and it is ok and what we want | 11:41 |
sonne|work | but now comes the ascii file | 11:41 |
sonne|work | and what you do is create a huge SGVector that you then convert into a SGMatrix | 11:42 |
sonne|work | since you read line-by-line | 11:43 |
shogun-buildbot | build #1677 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1677 blamelist: Fernando Iglesias <fernando.iglesiasg@gmail.com> | 11:43 |
thoralf | sonne|work: Just curious: Any specific reason why you decided for fortran order? | 11:43 |
sonne|work | thoralf: it is what R, matlab, octave, lapack, blas use | 11:44 |
lisitsyn | it is wise SVM-wise! | 11:44 |
thoralf | sonne|work: Okay, makes sense. Just seems "not natural", since when you do matrix[i] in matlab, you get a row vector. Even with continuously streaming data it does make thinks complicated, since you have to wait for the last example, until you have a consistent column. | 11:46 |
sonne|work | gsomix anyway since you read line by line you now have something that with a SGVector -> SGMatrix conversion and FORTRAN order will give you a transposed (C_ORDER) matrix %-) | 11:47 |
lisitsyn | thoralf: colwise is better to compute dot products :) | 11:47 |
lisitsyn | thoralf: and rowwise is better for lasso/etc | 11:47 |
lisitsyn | actually I believe it should transpose all the time automagically to achieve better performance | 11:48 |
thoralf | lisitsyn: Now it's getting interesting. :) Why is colwise better for dot products? | 11:48 |
lisitsyn | thoralf: I'd say caching | 11:48 |
sonne|work | gsomix so that is why it is confusing | 11:48 |
@iglesiasg | hushell, I just updated to latest code after your PR. SGObject.clone_equals_FactorGraph unit test fails here, what about in your machine? | 11:49 |
sonne|work | thoralf: matrix[i] in matlab is not defined :D | 11:49 |
thoralf | lisitsyn: If I have a dot-product of a row-vector with something else, it's just perfect. | 11:49 |
thoralf | sonne|work: matrix[i,:], sorry. | 11:49 |
thoralf | :D | 11:49 |
lisitsyn | thoralf: when you have your vector contiguous it is surely better | 11:49 |
sonne|work | thoralf: but you don't have to wait for all data no no it is one column vector is stored in memory after another | 11:50 |
lisitsyn | thoralf: to understand how it affects the performance | 11:50 |
lisitsyn | try fill some matrix with (i,j) and (j,i) | 11:50 |
sonne|work | thoralf: so with streaming it is contigous and everything | 11:50 |
lisitsyn | like symmetric matrix | 11:50 |
lisitsyn | such jumps make it like 2x-3x slower | 11:51 |
sonne|work | thoralf, lisitsyn: it is faster not due to caching but 'prefetching' memory can be read linearly | 11:51 |
thoralf | sonne|work: Oooh-kay. | 11:51 |
sonne|work | so you can read big blocks | 11:51 |
lisitsyn | sonne|work: how would you call such prefetching? | 11:51 |
thoralf | sonne|work: So you just store examples as columns? | 11:51 |
thoralf | sonne|work: Then it doesn't make any difference. | 11:52 |
lisitsyn | I am used to think of it like caching | 11:52 |
thoralf | :) | 11:52 |
-!- gsomix [~gsomix@85.26.165.90] has joined #shogun | 11:52 | |
sonne|work | thoralf: for GPUs it is even more extreme - they have commands to fetch contiguous data and then use the many GPU cores to do work | 11:52 |
gsomix | Hey there | 11:52 |
sonne|work | gsomix: clear? | 11:52 |
thoralf | sonne|work: I think it's just a matter of definition. If I concattenate row-vectors or col-vectors, it's all the same but the indices. | 11:53 |
sonne|work | thoralf: yes 1 example == 1 column | 11:53 |
gsomix | Wait a moment. Need read log. | 11:53 |
sonne|work | thoralf: sure | 11:53 |
sonne|work | thoralf: with python you can do that | 11:53 |
sonne|work | thoralf: just specify fortran or c order | 11:53 |
thoralf | sonne|work: Okay, so fortran order just swaps indices? Okay, then it's trivial. Memory layout will be the same... :) | 11:55 |
sonne|work | thoralf: but fortran order certainly is a brain twist for (C) programmers | 11:55 |
lisitsyn | ?? | 11:55 |
sonne|work | thoralf: if you want it that way yes | 11:55 |
thoralf | sonne|work: Thanks. | 11:56 |
lisitsyn | it matters whether you use contiguous (leading dimension = 1) | 11:56 |
lisitsyn | or non-contiguous (leading dimension >1) | 11:56 |
lisitsyn | that's it ;) | 11:56 |
thoralf | lisitsyn: Sure. :) | 11:57 |
lisitsyn | I wouldn't say it is crucial though | 11:57 |
lisitsyn | probably one can gain performance somewhere else | 11:57 |
thoralf | I used to implement n-dimensional arrays in C to have n-dimensional probability tables. | 11:57 |
lisitsyn | in the end such thing matters only for 'scientific' papers :D | 11:58 |
thoralf | So I'm away of that, but was just confused about what fortran order does. | 11:58 |
-!- van511 [~van51@athedsl-380772.home.otenet.gr] has joined #shogun | 11:58 | |
thoralf | s/away/aware/ | 11:58 |
lisitsyn | thoralf: how did you implement it? | 11:58 |
lisitsyn | like C[i][j][k]? | 11:58 |
thoralf | lisitsyn: No. One continuous array. | 11:59 |
sonne|work | thoralf: like shogun | 11:59 |
thoralf | lisitsyn: i*base^2 + j^base + k | 11:59 |
thoralf | (or the other way round, doesn't even matter) | 11:59 |
-!- van51 [~van51@ppp-94-66-20-32.home.otenet.gr] has quit [Ping timeout: 264 seconds] | 11:59 | |
lisitsyn | thoralf: it is better to use that right? | 11:59 |
thoralf | lisitsyn: Yes. | 12:00 |
gsomix | sonne|work clear. I wrote something confusint | 12:00 |
lisitsyn | than pointers to pointers | 12:00 |
thoralf | lisitsyn: Pointers will kill you. It's slow when you need to access all elements at a time. | 12:00 |
thoralf | lisitsyn: for example when finding the max. | 12:00 |
thoralf | lisitsyn: or projections | 12:01 |
lisitsyn | yeah that's what I wanted to say | 12:01 |
sonne|work | gsomix: no not really but it still needs 'fixing' of unit tests and maybe a doc overhaul | 12:01 |
gsomix | * confusing | 12:02 |
gsomix | OK, I'll check it twice. | 12:03 |
sonne|work | gsomix: the good news is that I converted a few examples already and they seemed to work | 12:05 |
gsomix | Ok, I'll fix it asap. But I'm now in bus. | 12:05 |
gsomix | BTW you said that there is one more bug. | 12:06 |
gsomix | With 92x92 matrix. | 12:06 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * b485ac9 / / (4 files): https://github.com/shogun-toolbox/shogun/commit/b485ac984a02784ad2edb912ab054235973992a1 | 12:06 |
shogun-notifier- | shogun: Fix HAVE_CXX11 doesn't imply HAVE_CXX11_ATOMIC and vice versa | 12:06 |
shogun-notifier- | shogun: Travis: fix allow_failures keys | 12:06 |
@wiking | gsomix: [ FAILED ] LineReaderTest.constructor | 12:07 |
@wiking | [ FAILED ] LineReaderTest.read_yourself | 12:07 |
@wiking | is this you? | 12:07 |
gsomix | Yep. | 12:07 |
gsomix | What's wrong? | 12:08 |
gsomix | Bad day for me. :) | 12:08 |
gsomix | wiking: ^ | 12:10 |
gsomix | sonne|work: ^^ | 12:10 |
sonne|work | gsomix: no the rest works | 12:11 |
sonne|work | wiking: what does this test do? | 12:11 |
@wiking | gsomix: maybe you fixed it yesterday | 12:12 |
sonne|work | wiking: no he didn't submit yesterday | 12:12 |
gsomix | Nope. | 12:12 |
@wiking | gsomix: https://travis-ci.org/shogun-toolbox/shogun/jobs/9933394#L3956 | 12:13 |
gsomix | Thnx | 12:13 |
@wiking | btw: we should consider adding time restriction for a given tests | 12:16 |
@wiking | anybody knows how to mess with the /* tp_dictoffset */ variable in SwigPyBuiltin__shogun__CStreamingFileFromSparseFeaturesT_char_t_type generated PyHeapTypeObject ? | 12:20 |
gsomix | wiking: fail, mobile browser can't load travis log. | 12:21 |
@wiking | gsomix: yeah it's a long one | 12:21 |
@wiking | gsomix: chrome should be able to do it ... mobile chrom | 12:22 |
gsomix | I'll see at evening, please disable this test. | 12:22 |
gsomix | wiking: nope. I have mobile chrome. | 12:23 |
@wiking | gsomix: basically its this: [ERROR] CLineReader::has_next():: Class is not initialized | 12:23 |
@wiking | unknown file: Failure | 12:23 |
@wiking | Unknown C++ exception thrown in the test body. | 12:23 |
@wiking | same for | 12:23 |
@wiking | LineReaderTest.read_yourself | 12:23 |
gsomix | Him, looks like test can't find yourself file | 12:24 |
gsomix | *hm | 12:24 |
@wiking | ? | 12:24 |
gsomix | Test just reads yourself. | 12:26 |
gsomix | And opens file io/FileReader_unit.cc | 12:27 |
gsomix | Or so | 12:27 |
gsomix | Doesn't really remember filename | 12:28 |
gsomix | wiking: ^ | 12:28 |
@wiking | yourself? meaning itself? :) | 12:30 |
gsomix | Hello. My name is Zhenya. I am from Russia. | 12:32 |
gsomix | Maybe itself | 12:32 |
gsomix | :) | 12:32 |
@wiking | wtf is happening here :) | 12:32 |
gsomix | I mean my English is bad and I can make errors | 12:35 |
@wiking | u can haz it | 12:35 |
gsomix | Indeed | 12:36 |
gsomix | OK, its needed just check path and filename in unit-test | 12:37 |
@wiking | doh bad news | 12:41 |
@wiking | that tp_dictoffset is generated by swig | 12:41 |
@wiking | and it's a bug :S | 12:41 |
@wiking | :) | 12:41 |
@wiking | ok let's create a PR for them | 12:41 |
@wiking | ;) | 12:41 |
@wiking | forking swig.... | 12:41 |
@wiking | if anybody's intersted: https://github.com/swig/swig/tree/joelandersson-matlab | 12:42 |
-!- van511 [~van51@athedsl-380772.home.otenet.gr] has quit [Quit: Leaving.] | 12:42 | |
gsomix | I'm interested | 12:45 |
gsomix | Off | 12:47 |
-!- gsomix [~gsomix@85.26.165.90] has quit [Quit: Bye] | 12:47 | |
lisitsyn | wiking: guys were sure it is dead and etc | 12:48 |
lisitsyn | wiking: no idea about it | 12:49 |
@wiking | heh pr is created for swig :) | 12:51 |
@wiking | ok not yet | 12:52 |
@wiking | https://github.com/swig/swig/pull/70 | 12:54 |
@wiking | yey | 12:54 |
@wiking | ;) | 12:54 |
@wiking | now let's temp fix this shit in the meanwhile | 12:54 |
lisitsyn | wiking: ha! | 12:55 |
@wiking | i think travis got stuck in infinite loop | 12:55 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/builds/9934180 is running like for 2 hours | 12:55 |
lisitsyn | wiking: but it should be killed? | 12:56 |
lisitsyn | IIRC it is killed if there is no output for 10 minutes | 12:56 |
@wiking | lisitsyn: afaik yes | 12:56 |
@wiking | but seems that watchdog script died or seomthign | 12:56 |
lisitsyn | heh | 12:56 |
lisitsyn | cool | 12:56 |
hushell | iglesiasg: unfortunately, it failed here as well, I am trying to fix it | 12:59 |
@iglesiasg | hushell, I think the best idea is to move DistjointSet to its own file and that's all. It should be easy. | 13:04 |
@iglesiasg | hushell, please submit a pull request just with that when it is ready | 13:04 |
hushell | iglesiasg: I think you are right, I got error like this: Could not create empty instance of "DisjointSet" | 13:08 |
@iglesiasg | hushell, yes | 13:09 |
@wiking | sonne|work: can u add a A record in shogun-toolbox.org domain? | 13:25 |
sonne|work | wiking: no | 13:26 |
@wiking | sonne|work: if so then please just add: cdash.shogun-toolbox.org 178.63.98.67 | 13:26 |
sonne|work | I don't own it | 13:26 |
@wiking | sonne|work: why not? | 13:26 |
@wiking | hah lol | 13:26 |
@wiking | gunnar has it i see ;) | 13:26 |
@wiking | sonne|work: i'm happy to host the DNS resolvers for the domain.... just make a transfer and that way we can do the management | 13:27 |
@wiking | just add ns.info2k1.hu and ns3.info2k1.hu as resolvers for the domain | 13:29 |
sonne|work | wiking: I have nameservers too but I don't own it :/ | 13:30 |
@wiking | sonne|work: gunnar will surely agree to transfer the resolvers ;) | 13:30 |
sonne|work | I ask him already and he said we should discuss it at WS | 13:31 |
sonne|work | but I forgot | 13:31 |
@wiking | heheh lol | 13:31 |
-!- foulwall [~user@116.114.154.68] has joined #shogun | 13:34 | |
sonne|work | wiking: ok sent him an email also asking about the osx buildbot | 13:35 |
@iglesiasg | sonne|work, BTW, Kilian answered already yesterday, very cool! | 13:38 |
sonne|work | iglesiasg: cool | 13:40 |
sonne|work | iglesiasg: did it help? | 13:40 |
-!- foulwall [~user@116.114.154.68] has quit [Remote host closed the connection] | 13:41 | |
@iglesiasg | sonne|work, yeah, quite a lot! | 13:41 |
hushell | iglesiasg: okay, I sent the PR, please give some feedback, I'll check when I got up | 13:41 |
@iglesiasg | hushell, thank you! Sure, I will do during the afternoon | 13:42 |
hushell | iglesiasg: I should thank for your reviews :) | 13:43 |
hushell | CU guys, have a nice day | 13:43 |
@iglesiasg | hushell, I am should apologize for the many comments, I think I am usually picky | 13:43 |
hushell | iglesiasg: that's very good! | 13:44 |
hushell | talk you later :) | 13:45 |
-!- hushell [~hushell@c-24-21-169-136.hsd1.or.comcast.net] has quit [Quit: WeeChat 0.3.7] | 13:45 | |
@iglesiasg | I'got a doubt about matlab static | 13:52 |
@iglesiasg | say I compile against it | 13:53 |
@iglesiasg | IIRC this creates some sort of file that matlab uses to have access to Shogun | 13:53 |
@iglesiasg | just using this file in another computer where Shogun has not been installed, will it work well to use the matlab static interface? | 13:53 |
@wiking | heh the travis got autorestarted ;) | 14:00 |
sonne|work | iglesiasg: no you need libshogun and the same arch | 14:02 |
@iglesiasg | sonne|work, I see | 14:02 |
@iglesiasg | sonne|work, I am trying to figure out how to do so Georg can test our LMNN too causing him the lest possible pain :) | 14:03 |
@iglesiasg | least* | 14:03 |
@wiking | iglesiasg: !!! | 14:03 |
@wiking | 14:01 < jon_r> also if you're a committer to that project you can force a build restart | 14:03 |
@wiking | 14:01 < wiking> jon_r: how? | 14:03 |
@wiking | 14:02 < jon_r> cog -> restart build | 14:03 |
@wiking | 14:02 -!- DracoBlue [~DracoBlue@h-213.61.228.229.host.de.colt.net] has joined #travis | 14:03 |
@wiking | 14:02 < jon_r> but you need push access to the git repo to have that option afaik | 14:04 |
@wiking | 14:02 < wiking> jon_r: on the travis page i suppose | 14:04 |
@wiking | 14:02 < jon_r> ya | 14:04 |
@wiking | 14:02 < wiking> jon_r: i do have | 14:04 |
@wiking | 14:02 < jon_r> travis piggy backs off gh permissions | 14:04 |
@wiking | 14:03 < jon_r> oh and you need to resync if you've only been added recently ;) | 14:04 |
@wiking | 14:03 < wiking> jon_r: resync how? | 14:04 |
@wiking | if u know what i mean... | 14:04 |
@iglesiasg | wtf | 14:04 |
@iglesiasg | yeah.. | 14:04 |
sonne|work | iglesiasg: well try to build with as few libs as possible (not atlas,glpk, ...) and disable-optimization or just -O2? | 14:04 |
@iglesiasg | wiking, so it is that easy to restart travis build?! | 14:05 |
@wiking | iglesiasg: yes :) | 14:05 |
@iglesiasg | facepalm... | 14:05 |
@wiking | iglesiasg: although it doesn't work for me yet | 14:05 |
@iglesiasg | wiking, aham, why so? | 14:05 |
@wiking | i've try to do the sync | 14:05 |
@wiking | but still i dont get rebuild stuff there | 14:05 |
@wiking | only when i check my own repo | 14:05 |
@iglesiasg | the sync part didn't really get it | 14:05 |
@wiking | iglesiasg: on travis site | 14:06 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/builds/9934180 | 14:06 |
@wiking | right topis corner | 14:06 |
@wiking | there's a cogwheel | 14:06 |
@wiking | right? | 14:06 |
@iglesiasg | sonne|work, you mean that he should do this in his machine? Or that I can do that and ship to him some sort of shared library? | 14:06 |
@iglesiasg | wiking, right | 14:06 |
@wiking | if you click on it you should have | 14:07 |
@iglesiasg | aah that one can actually sign in in travis | 14:07 |
sonne|work | iglesiasg: you on yours if it is easier for him but I guess it is easiest if he does it with your guidance | 14:07 |
sonne|work | iglesiasg: I mean it is not really difficult | 14:07 |
@wiking | restart build | 14:07 |
@wiking | like for example here | 14:07 |
@wiking | https://travis-ci.org/shogun-toolbox/shogun/builds/9933391 | 14:07 |
@wiking | ah works for me now | 14:07 |
@wiking | u? | 14:07 |
@wiking | see good to hang out in #travis ;) | 14:08 |
@iglesiasg | wiking, I see it in the second link too | 14:08 |
@iglesiasg | wiking, but not in the first one | 14:08 |
@iglesiasg | ah ok, but that one is currently building | 14:08 |
@wiking | yeah you can only do this with finished builds as it seems | 14:09 |
@iglesiasg | wiking, this is really handy! | 14:09 |
@wiking | < jon_r> you can restart individual jobs as well | 14:09 |
@iglesiasg | let's find out how | 14:10 |
@wiking | iglesiasg: https://travis-ci.org/shogun-toolbox/shogun/jobs/9934181 | 14:10 |
@wiking | and there you can do a restart job | 14:10 |
@wiking | cog wheel etc | 14:10 |
@iglesiasg | awesome | 14:10 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 14:11 | |
travis-ci | [travis-ci] it's Fernando Iglesias'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/9934180 | 14:11 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 14:11 | |
@wiking | heh finally that job died :( | 14:11 |
@wiking | it took like fucking ages | 14:11 |
@iglesiasg | wiking, does it affect your work with cmake that we submit new things? I mean because it uses up travis | 14:12 |
@wiking | iglesiasg: well yeah but it's ok | 14:12 |
@wiking | i hope to be finish today | 14:12 |
@wiking | now the small little shits i'm fixing | 14:12 |
@wiking | and then it's done | 14:12 |
@wiking | mostly it's done | 14:12 |
@iglesiasg | all right | 14:12 |
@wiking | it's just now hacking around with travis | 14:12 |
@iglesiasg | but I will reduce the number of things I push in the mean time | 14:12 |
@wiking | btw | 14:12 |
@wiking | i hope travis has an API | 14:13 |
@iglesiasg | what do you want to do? | 14:13 |
@wiking | because then we could just restart jobs via the irc client as well | 14:13 |
@iglesiasg | aham like we do with the buildbot | 14:13 |
@wiking | indeed | 14:15 |
@wiking | restart - restarts a build or a job | 14:15 |
@wiking | https://github.com/travis-ci/travis#readme | 14:16 |
@wiking | Travis CI Client (CLI and Ruby library) | 14:16 |
@wiking | there you go | 14:16 |
@wiking | sonne|work: can u install that on the ircbot machine? | 14:17 |
sonne|work | wiking: ahh no it is sufficient if we can click on failed builds | 14:22 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has joined #shogun | 14:42 | |
travis-ci | [travis-ci] it's Fernando Iglesias'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/9934340 | 14:42 |
-!- travis-ci [~travis-ci@ec2-54-235-16-133.compute-1.amazonaws.com] has left #shogun [] | 14:42 | |
-!- foulwall [~user@116.114.154.68] has joined #shogun | 14:45 | |
@wiking | seems like gmock/gtest does not like c++0x | 14:46 |
@wiking | GaussianDistribution.sample_2d is having some problems atm it seem | 14:50 |
@wiking | lol | 14:53 |
@wiking | http://about.travis-ci.org/docs/user/osx-ci-environment/ | 14:53 |
@wiking | what is the magic key? | 14:53 |
@wiking | language: objective-c | 14:54 |
@wiking | woah | 14:54 |
@wiking | coool | 14:54 |
@wiking | we have a mac unit testing on travis from now on | 14:54 |
-!- van51 [~van51@athedsl-380772.home.otenet.gr] has joined #shogun | 14:56 | |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has joined #shogun | 15:00 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9936663 | 15:00 |
-!- travis-ci [~travis-ci@ec2-54-234-6-103.compute-1.amazonaws.com] has left #shogun [] | 15:00 | |
@wiking | well apartently they are down now, we'll have soon an osx builder with travis | 15:03 |
@wiking | https://travis-ci.org/vigsterkr/subgrep/jobs/9941955 | 15:03 |
@wiking | hehe lol | 15:03 |
@wiking | we can even build for iOS :DD | 15:04 |
sonne|work | ohh nice | 15:05 |
sonne|work | wiking: that is excellent news | 15:06 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 15:06 | |
@wiking | you mean the ios builds? :D | 15:08 |
@wiking | hihi | 15:08 |
@wiking | but we would need then that cdash becuause the travis bots could then create packages and upload it to cdash | 15:08 |
@wiking | and then we could distrib it easily | 15:08 |
sonne|work | wiking: no osx :D | 15:11 |
@wiking | :P | 15:12 |
@wiking | ok need to finish jinja2 targets | 15:12 |
@wiking | and then still fucking find out why gmock fails with c++0x | 15:12 |
sonne|work | wiking: why do we need c++0x? | 15:17 |
@wiking | sonne|work: atomic | 15:18 |
sonne|work | but why not allow that only with c++11? | 15:21 |
@wiking | why would we want to restrict it to that? | 15:22 |
@wiking | if it's available | 15:22 |
-!- foulwall [~user@116.114.154.68] has quit [Ping timeout: 240 seconds] | 15:23 | |
sonne|work | it is tough to support all kind of settings / environments | 15:25 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Ping timeout: 245 seconds] | 15:27 | |
@wiking | hahahah | 15:40 |
@wiking | http://stackoverflow.com/questions/9830121/is-ther-a-good-workaround-for-gccs-sorry-unimplemented-cannot-expand-next | 15:40 |
@wiking | "Update GCC." | 15:40 |
@wiking | :D | 15:40 |
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Ping timeout: 245 seconds] | 15:42 | |
lisitsyn | wiking: how to use X? don't use X | 15:43 |
@wiking | ok | 15:44 |
@wiking | fuck it | 15:44 |
lisitsyn | :D | 15:45 |
@wiking | i'll just fucking disable c++0x | 15:45 |
@wiking | when testing is enabled... | 15:45 |
@wiking | or? :) | 15:45 |
@wiking | or no... i will not enable it on gcc 4.6 or older... | 15:46 |
@wiking | althoguh in that case c++11 will be already available | 15:49 |
@wiking | but ok ... :D | 15:49 |
-!- HeikoS [~heiko@nat-166-201.internal.eduroam.ucl.ac.uk] has joined #shogun | 15:50 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:50 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 15:50 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun | 15:51 | |
@wiking | does anybody have 4.6.4 gcc lying around? | 15:52 |
iglesiasg | 4.6.3 here, sorry | 15:53 |
@wiking | yeah that's debian/ubuntu | 15:54 |
@wiking | or 4.7.x if later | 15:54 |
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Remote host closed the connection] | 15:56 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has joined #shogun | 16:02 | |
-!- naywhayare [~ryan@spoon.lugatgt.org] has quit [Quit: leaving] | 16:12 | |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has joined #shogun | 16:45 | |
-!- lisitsyn [~lisitsin@mxs.kg.ru] has quit [Quit: Leaving.] | 17:10 | |
@HeikoS | wiking: around? | 17:26 |
@sonney2k | van51, Gli utenti costituiscono una comunità collaborativa, in cui tutti i membri, grazie anche ai progetti tematici e ai rispettivi luoghi di discussione, coordinano i propri sforzi nella redazione delle voci. | 17:40 |
@sonney2k | gtg | 17:40 |
iglesiasg | training italian? | 17:40 |
van51 | iglesiasg: we 're training on italian :P | 18:07 |
van51 | among others | 18:08 |
-!- gsomix [~gsomix@178.45.109.87] has joined #shogun | 18:30 | |
gsomix | good evening | 18:31 |
@wiking | HeikoS: yes | 18:59 |
@HeikoS | wiking: hi! | 18:59 |
@wiking | yo | 18:59 |
@HeikoS | so I was wondering where to start with the ipython stuff | 18:59 |
@HeikoS | I have a script that can run scripts locally and store notebook copies | 18:59 |
@wiking | mmm so that script you've sent me | 18:59 |
@HeikoS | I can write something that excecutes all notebooks in a foder | 18:59 |
@wiking | basically runs the notebook? | 18:59 |
@HeikoS | yes | 18:59 |
@wiking | ok cool | 18:59 |
@HeikoS | and stores a copy with the output | 18:59 |
@wiking | we dont need more | 18:59 |
@wiking | let me make it | 19:00 |
@HeikoS | wiking: if you want, feel free ;) probably faster than me | 19:00 |
@wiking | i mean add it to cmake | 19:00 |
@wiking | as a new target | 19:00 |
@wiking | make notebooks | 19:00 |
@wiking | that'll basically run the notebooks | 19:00 |
@HeikoS | awesome | 19:00 |
@wiking | ok i'll do that | 19:00 |
@HeikoS | wiking: can we then upload them to shogun webspace? | 19:00 |
@wiking | but let me finish first the jinja | 19:00 |
@wiking | in cmake | 19:00 |
@HeikoS | wiking: ok | 19:00 |
@HeikoS | wiking: I need more of your jinga skills afterwards | 19:01 |
@wiking | yeah we need a webspace on shogun's website for it | 19:01 |
@wiking | create a user or something | 19:01 |
@wiking | to upload it there | 19:01 |
@HeikoS | wiking: yes, and then we need to generate a html file with all the nbviwer links | 19:01 |
@wiking | sonne|work: ^ ideas how to do that? | 19:01 |
@wiking | HeikoS: that should be pretty straightforward | 19:01 |
@HeikoS | wiking: cool | 19:01 |
@wiking | ok lemme get a coffee | 19:02 |
@wiking | and i'll finish now the jinja | 19:02 |
@HeikoS | do you need a template for the html or so? | 19:02 |
@wiking | btw good news: we are going to have osx builds with travis ;) | 19:02 |
@HeikoS | wiking: ok cool, I would like to get this working asap so let me know how I can help | 19:02 |
@HeikoS | wiking: wow you rock man! :) | 19:02 |
@HeikoS | when is cmake ready to take over? | 19:02 |
@wiking | HeikoS: would help a lot if we would have a template | 19:02 |
@wiking | HeikoS: hope it's today | 19:02 |
@HeikoS | whoooo! | 19:02 |
@wiking | i need the jinja fixed | 19:02 |
@wiking | and then the osx job fixed | 19:02 |
@wiking | and that's it basically | 19:03 |
@wiking | ok brb | 19:03 |
@wiking | getting some coffee | 19:03 |
@HeikoS | ok cool! | 19:03 |
-!- van51 [~van51@athedsl-380772.home.otenet.gr] has quit [Quit: Leaving.] | 19:11 | |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 19:18 | |
@wiking | lisitsyn: yo | 19:18 |
lisitsyn | wiking: hey what's up | 19:18 |
@wiking | lisitsyn: can u find out how we can replace this: FILE(GLOB_RECURSE UNITTEST_SRC_TMP RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *${EXT_SRC_TEST_TMP}) | 19:19 |
@wiking | lisitsyn: this is basically now generating the list of files that needs to be compiled or something | 19:19 |
lisitsyn | wiking: need to know what's wrong with it | 19:19 |
@wiking | ok so this | 19:19 |
@wiking | lisitsyn: from cmake's manual: ". (We do not recommend using GLOB to collect a list of source files from your source tree. If no CMakeLists.txt file changes when a source is added or removed then the generated build system cannot know when to ask CMake to regenerate" | 19:20 |
lisitsyn | got it | 19:20 |
@wiking | so we need to find a why to do this | 19:20 |
@wiking | for sure with add_custom_target/add_custom_command combo it's possible | 19:21 |
@wiking | see e.g. class_list.cpp generation | 19:21 |
@wiking | if u find a way | 19:21 |
@wiking | we need basically a macro | 19:21 |
@wiking | that have 2 func args: path and filename pattern | 19:21 |
@wiking | thnx | 19:21 |
lisitsyn | will be back in a minute | 19:21 |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has quit [Quit: Leaving.] | 19:21 | |
-!- lisitsyn [~lisitsyn@fb2-lo1.global63.net] has joined #shogun | 19:24 | |
@wiking | and could anybody tell me what is the wrong with this bash line: if [ -z $OSX ] ; then sudo apt-get update -qq ; else ; brew update ; fi | 19:25 |
@wiking | /home/travis/build.sh: line 125: syntax error near unexpected token `;' | 19:25 |
@wiking | /home/travis/build.sh: line 125: `if [ -z $OSX ] ; then sudo apt-get update -qq ; else ; brew update ; fi' | 19:25 |
@wiking | i have no idea why | 19:25 |
@HeikoS | wiking, lisitsyn https://github.com/shogun-toolbox/shogun/issues/1348 | 19:31 |
@HeikoS | what do you think on this? | 19:31 |
@wiking | HeikoS: how? | 19:32 |
lisitsyn | HeikoS: I am okay with it | 19:32 |
@wiking | ah ok dyncasting | 19:32 |
@wiking | sounds cool | 19:33 |
@wiking | but you should add try{ } catch{} | 19:33 |
@HeikoS | this we we would get rid of things | 19:33 |
@wiking | for the bad_cast exception | 19:33 |
@HeikoS | wiking: fine with me | 19:33 |
@HeikoS | wiking: we are not really using exceptions in shogun are we? | 19:33 |
@wiking | yeah there's shogun_exception ;) | 19:33 |
@HeikoS | does SG_ERROR throw it? | 19:33 |
lisitsyn | HeikoS: I wish we had static polymorphism :D | 19:34 |
lisitsyn | not possible though | 19:34 |
lisitsyn | HeikoS: yes | 19:34 |
@wiking | ShogunException | 19:34 |
@wiking | rather | 19:34 |
@HeikoS | wiking: so we can refine that later via the template for the obtain_from_generic | 19:34 |
@HeikoS | wiking, lisitsyn so if anyone could do some more magic with the templated stuff? :) | 19:35 |
@HeikoS | then we can finally remove all those annoying obtain_from_generic calls | 19:35 |
@wiking | HeikoS: you mean you want this to be autogenerated? | 19:36 |
@HeikoS | wiking: yes | 19:36 |
@HeikoS | for all classes | 19:36 |
@wiking | mmm wait | 19:36 |
@wiking | can't we just template it | 19:36 |
@wiking | ? | 19:36 |
@wiking | like | 19:36 |
@HeikoS | wiking: no, swig | 19:36 |
@wiking | obtain_from_generic<ClassName>(generic) | 19:36 |
@HeikoS | this is only needed for modular interfaces with typefree languages | 19:37 |
@HeikoS | wiking: that was my first idea but would be messy | 19:37 |
@wiking | coz swig cannot handle it or what? | 19:37 |
@HeikoS | wiking: in c++ it works fine | 19:37 |
@HeikoS | wiking: so then the method names have to be added to modular | 19:37 |
@HeikoS | like RealFeatures, ShortFeatures etc | 19:37 |
@HeikoS | translating templated classes to modular is kinda annoying | 19:38 |
@wiking | HeikoS: we can autogen it there ;) | 19:38 |
@HeikoS | wiking: true, but think of the compile time | 19:38 |
@wiking | well in case of c++ it'd be better :) | 19:39 |
@wiking | i dont know about swig :D | 19:39 |
@HeikoS | I would rather not mess around with it ;) | 19:39 |
@wiking | lisitsyn: any idea bout that bash script? | 19:39 |
lisitsyn | wiking: no ; looked correct to me | 19:39 |
lisitsyn | I tried to randomly replace them | 19:39 |
lisitsyn | no help | 19:39 |
lisitsyn | :D | 19:39 |
@wiking | heheh yeah i've done that too | 19:39 |
@wiking | didn't help | 19:39 |
@wiking | i have no idea... fucker | 19:40 |
@wiking | anyways now i have to figure out | 19:40 |
@wiking | how to give the path to the clone_unittest.cc.py script the class_list.cpp path | 19:40 |
@wiking | :S | 19:40 |
lisitsyn | wiking: if [ -z $OSX ]; then sudo apt-get update -qq; else brew update ; fi | 19:40 |
lisitsyn | I think you need dis | 19:40 |
@wiking | testing :P | 19:41 |
lisitsyn | wiking: it is like everybody uses glob | 19:45 |
@wiking | lisitsyn: i know but still | 19:46 |
@sonney2k | gsomix, please try to get the tests back to green! | 19:48 |
gsomix | sonney2k, working on it | 19:49 |
gsomix | sonney2k, but we need to think twice. | 19:51 |
gsomix | sonney2k, one line in Fortran order is one column (one feature vector), right? | 19:51 |
-!- lambday [67157d4f@gateway/web/freenode/ip.103.21.125.79] has joined #shogun | 19:51 | |
lambday | HeikoS: hi | 19:52 |
@HeikoS | lambday: hi! :) | 19:52 |
@HeikoS | how is it going? | 19:52 |
lambday | HeikoS: how was your NIPS thing? | 19:52 |
lambday | HeikoS: bad :( the machine I work on in the lab had its HD crashed and is being replaced | 19:52 |
@HeikoS | lambday: NIPS was submitted yesterday night later | 19:53 |
@HeikoS | lambday: oh, so you have no place to work now? | 19:53 |
lambday | HeikoS: my laptop takes 20 mins to compile shogun.. I am planning to buy a new one this week.. | 19:53 |
lambday | HeikoS: I can work though.. | 19:54 |
@HeikoS | lambday: wow thats long | 19:54 |
@HeikoS | lambday: will you get a new institute machine soon? | 19:54 |
lambday | HeikoS: 1GB ram, core 2 duo, you can imagine | 19:54 |
lambday | HeikoS: yes I can manage.. | 19:54 |
@HeikoS | lambday: haha I had that a few years ago and freaked out whenever I tried to compile shogun ;) | 19:55 |
lambday | but I have to set it all up, git, valgrind gterst gmock etc | 19:55 |
@HeikoS | lambday: just disable any modular bindings | 19:55 |
@HeikoS | lambday: ah painful ... | 19:55 |
lambday | HeikoS: yes that might help | 19:55 |
@HeikoS | sonney2k: I would like to activate line and file for warnings and errors, what do you think? | 19:55 |
lambday | HeikoS: any suggestions on buying a good laptop? | 19:55 |
@HeikoS | lambday: get a good one! | 19:55 |
@HeikoS | lambday: I prefer thinkpads, but these are expensive | 19:56 |
@HeikoS | but good quality is sometimes better than an extremely fast machine | 19:56 |
@HeikoS | SSDs are nice, a lot of ram is noce, 4 cores are nice | 19:56 |
pickle27 | +1 for thinkpad | 19:56 |
pickle27 | or ideapad | 19:56 |
lambday | lenovo | 19:56 |
@wiking | lisitsyn: \o/ that worked! | 19:56 |
@HeikoS | but the best thing about thinkpads is the high quality of the parts like keyboard display etc | 19:56 |
lambday | but I don't get why they are so pricey | 19:57 |
lambday | oh | 19:57 |
@HeikoS | lambday: you can get a similarly fast machine for much less, but that is all plastic cheap ;) | 19:57 |
lambday | HeikoS: the problem is, most of the laptops we get comes with windows 8 preinstalled and it creates some problem with linux dual boot since it doesn't use BIOS but UEFI | 19:58 |
@HeikoS | lambday: yeah I had that | 19:58 |
@HeikoS | uefi and dualboot is messy | 19:58 |
@HeikoS | Thats why I deleted windows :D | 19:58 |
@HeikoS | btw lenovo sells thinkpads without any OS in germany, maybe also in india? | 19:58 |
@HeikoS | I bought mine without OS | 19:58 |
@HeikoS | student deal | 19:59 |
@HeikoS | also 15% discount | 19:59 |
lambday | ahan! | 19:59 |
lambday | I gotta check | 19:59 |
@wiking | fuck why wasn't i using ccache before... | 19:59 |
@HeikoS | wiking: haha :) | 19:59 |
lambday | HeikoS: I had my eye on an amazing HP pavilion (core i7 3rd gen, 8gb ddr3, 1tb hd).. but then I learned it comes with win8 | 19:59 |
@HeikoS | lambday: ccache ^ | 20:00 |
@wiking | HeikoS: i mean for shogun... but we would still need ccache-swig enabled as well... it's magic ;P | 20:00 |
@wiking | i've tested it manually :P | 20:00 |
lambday | HeikoS: yes I remember you advicing me using that :) | 20:00 |
@wiking | cache size 848.0 Mbytes | 20:00 |
@wiking | lol | 20:00 |
@wiking | that is only shogun :P | 20:00 |
@wiking | comeooooon i want to see jinja2 working | 20:01 |
@HeikoS | wiking: installation file is also pretty big :D | 20:01 |
lambday | HeikoS: oh and regarding that type casting thing, yes I thought of static method but then thought of having a syntactic sugar | 20:02 |
lambday | HeikoS: I am not sure which kinds of things we can use with swig | 20:02 |
lambday | :-/ | 20:02 |
lambday | I gotta keep in mind for those classes whom I am planning to expose to modular interface | 20:02 |
@HeikoS | lambday: yes | 20:03 |
@HeikoS | lambday: no templates in SWIG | 20:03 |
lambday | HeikoS: all concrete? | 20:03 |
@HeikoS | you have to create a name for every fixed template | 20:03 |
lambday | oh | 20:03 |
@HeikoS | wiking: btw for ccache did you use the symlink install way? | 20:03 |
@wiking | HeikoS: don't lie there's template in swig :P | 20:03 |
lambday | and for methods there is no boundary, right? I can use operator overloading and stuffs? | 20:03 |
@HeikoS | wiking: we are not using them though | 20:04 |
@wiking | HeikoS: no... cmake autodetect/enables it for ya | 20:04 |
lambday | wiking: there is? :-/ | 20:04 |
@HeikoS | lambday: yes | 20:04 |
@HeikoS | wiking: whooooo | 20:04 |
@wiking | HeikoS: cmake is your friend | 20:04 |
@wiking | ;) | 20:04 |
@HeikoS | when is it ready? :) | 20:04 |
@wiking | well i'm just testing jinja2 generation | 20:04 |
@wiking | let's hope now it works | 20:04 |
@HeikoS | cool | 20:05 |
lambday | HeikoS: alright.. it worked though.. it would be easy to replace with static method later if I find problems with swig | 20:05 |
@HeikoS | lambday: ok then do it | 20:05 |
@HeikoS | lambday: remind me of the state | 20:05 |
@HeikoS | what is next/mmissing? | 20:05 |
lambday | alright | 20:05 |
lambday | HeikoS: oh and Lanczos ended up giving amazing accuracy! | 20:05 |
@wiking | ok the clone_unittest.cc is generated... | 20:06 |
lambday | I actually useda bit different algorithm than that of Krylstat.. | 20:06 |
@wiking | so let's hope that it compiles it as well | 20:06 |
lambday | HeikoS: wiki claims this method to be most numerically stable | 20:06 |
@wiking | yes yes tests/unit/CMakeFiles/shogun-unit-test.dir/clone_unittest.cc.o | 20:06 |
@HeikoS | lambday: nice! | 20:06 |
@HeikoS | lambday: which one? | 20:06 |
@HeikoS | wiking: nice! | 20:06 |
@HeikoS | nice nice nice! | 20:06 |
lambday | HeikoS: the one I used for Lanczos... | 20:06 |
lambday | the main loop | 20:06 |
@HeikoS | lambday: nice that it is accurate! | 20:07 |
@HeikoS | lambday: did you btw test this matrix I sent you? | 20:07 |
@HeikoS | ozone one? | 20:07 |
lambday | HeikoS: no not yet! | 20:08 |
lambday | oh yes I should! | 20:08 |
@HeikoS | lambday: ok that would be next for the solvers etc | 20:08 |
@HeikoS | C^-1 x | 20:08 |
@HeikoS | eig(C) | 20:08 |
lambday | alright | 20:08 |
@HeikoS | lambday: and apart from that? | 20:09 |
@HeikoS | COCG_M all fine? | 20:09 |
@wiking | HeikoS: ok commiting .... let's see travis on fire with the first osx build job \o/ | 20:09 |
@HeikoS | lambday: sorry I lost a bit track last week | 20:09 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 20:09 | |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * a7ef209 / src/interfaces/python_modular/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/a7ef209dc414e84a399a941899e002ebabcec8a0 | 20:09 |
shogun-notifier- | shogun: Fix PythonModular build with C++11 | 20:09 |
shogun-notifier- | shogun: read more about this bug at: | 20:09 |
shogun-notifier- | shogun: https://github.com/swig/swig/pull/70 | 20:09 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * fea7732 / cmake/CheckCXX11Features.cmake: https://github.com/shogun-toolbox/shogun/commit/fea773222608d631c7a2695ae5d8c740679207a4 | 20:09 |
shogun-notifier- | shogun: Fix variadic template handling with -std=c++0x in gcc | 20:09 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * bf73b4a / cmake/CheckCXX11Features.cmake: https://github.com/shogun-toolbox/shogun/commit/bf73b4a0ec1a78fbeef835f1014fe8aa96cf2cee | 20:09 |
shogun-notifier- | shogun: UNSET(_HAS_CXX0X_FLAG) if gcc is <= 4.6.3 | 20:09 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * e0a2dbf / .travis.yml: https://github.com/shogun-toolbox/shogun/commit/e0a2dbf7209d0955bd5382ce6488ed1eddc9cf5c | 20:09 |
shogun-notifier- | shogun: Travis: add OSX job | 20:09 |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 7f0dfcf / CMakeLists.txt,tests/unit/CMakeLists.txt,tests/unit/base/clone_unittest.cc.py: https://github.com/shogun-toolbox/shogun/commit/7f0dfcf837d7c506f68159d445ee7d479f7b4b77 | 20:09 |
shogun-notifier- | shogun: Add cmake target for jinja2 generated unit tests | 20:09 |
@wiking | tadaaaaaaaaaaaaaaaaaaaaaa | 20:09 |
@HeikoS | wiking: whooo :) osx builds so good! | 20:09 |
lambday | HeikoS: no no I myself had to go too slow last week because of all these insti level stuffs :( | 20:09 |
@wiking | HeikoS: build 1700 | 20:09 |
@wiking | :) | 20:09 |
lambday | HeikoS: the one we have is no more cocg_m but cg_m | 20:10 |
@HeikoS | haha | 20:10 |
@HeikoS | lambday: ah yes | 20:10 |
lambday | for individual we'll be using cocg | 20:10 |
lambday | preconditioned | 20:10 |
@HeikoS | but it is also for families of systems? | 20:10 |
@HeikoS | lambday: yes exactly | 20:10 |
lambday | yes | 20:10 |
@HeikoS | lambday: cool! | 20:11 |
lambday | but I am not too sure about how accuracy that would be to the actual soln | 20:11 |
lambday | how accurate* | 20:11 |
@HeikoS | man, exciting stuff | 20:11 |
@HeikoS | lambday: what do you mean | 20:11 |
@sonney2k | gsomix, no one line in an ascii file is one line in C order | 20:11 |
-!- pickle27 [~Kevin@d67-193-243-174.home3.cgocable.net] has quit [Quit: Leaving] | 20:12 | |
lambday | HeikoS: the matrices I tested with didn't give accuracy as I expected.. although small ones.. | 20:12 |
lambday | compared with direct solvers.. | 20:12 |
lambday | with fixed number of iterations, 1000 | 20:12 |
@HeikoS | lambday: ok, well we will have to test that on larger problems and compare against some other implementation | 20:12 |
gsomix | sonney2k, hm, ok. it seems that we don't really need order settings in CSVFile | 20:13 |
lambday | HeikoS: okay.. yes few libraries have cocg implemented but I haven't found any that has cg_m implemented except krylstat... | 20:13 |
lambday | HeikoS: I'll do some tests | 20:13 |
@HeikoS | lambday: lets use krylstat then :) | 20:13 |
lambday | yes.. alright | 20:14 |
gsomix | sonney2k, we just need to read data line by line and transpose it. | 20:14 |
gsomix | right? | 20:14 |
lambday | HeikoS: alright so I'll do these tests this week... and add cg_m job.. | 20:15 |
lambday | and move on to coloring from next week onwards | 20:15 |
iglesiasg | all right, I will see you later guys | 20:16 |
@sonney2k | gsomix, yes we always read data in line by line | 20:16 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 20:16 | |
gsomix | sonney2k, when I'm introducing order settings I mean that we can read lines either as rows (vectors) or as cols (features). | 20:16 |
gsomix | *I was | 20:16 |
@sonney2k | yes exactly | 20:17 |
gsomix | sonney2k, ok, then let's simplify it. | 20:18 |
@HeikoS | lambday: cool | 20:18 |
@HeikoS | lambday: and then all the stuff is done right? only graph stuff missing | 20:19 |
lambday | HeikoS: yes.. | 20:19 |
lambday | HeikoS: and individual preconditioner thing missing | 20:19 |
@HeikoS | lambday: ah yes | 20:19 |
gsomix | sonney2k, CSVFile should read line by line in SGVector and then transpose it with help of convert_to_matrix | 20:19 |
@HeikoS | ok we should add this too, but later | 20:19 |
@HeikoS | its quite easy | 20:19 |
@HeikoS | just need another operator and some minor changes | 20:20 |
lambday | I plan to do it later after everything seems to work | 20:20 |
lambday | yes | 20:20 |
@sonney2k | gsomix, yeah but you do that already | 20:20 |
@sonney2k | gsomix, only that the sgvector -> sgmatrix step is confusing since c_order in ascii means fortran order o_O | 20:20 |
lambday | HeikoS: I feel bad it I could involve all the time for this I could have finished it all by now :( but then by prof :'( | 20:21 |
lambday | my* | 20:21 |
lambday | this formal verification project is killing me :'( | 20:21 |
@HeikoS | lambday: dont worry | 20:21 |
lambday | man my brain is not designed for multitasking.. I don't know how you manage all these all at once | 20:21 |
@sonney2k | gsomix, btw I need an ascii reader also ... | 20:22 |
@HeikoS | lambday: my brain is neither | 20:22 |
@sonney2k | lambday, mine is neither | 20:22 |
lisitsyn | I tried multitasking and I got no brain now | 20:22 |
@HeikoS | lambday: just organise time properly, then it will work :) | 20:22 |
lambday | that feels good | 20:22 |
@sonney2k | lambday, relax just stick with us and then we can continue anything later too | 20:22 |
gsomix | sonney2k, m? what do you mean? | 20:22 |
@wiking | gsomix: man we need to talk :) | 20:23 |
@wiking | gsomix: in english :) | 20:23 |
lambday | sonney2k: I will... I never had this much fun in anything else! | 20:23 |
@HeikoS | lambday: also dont worry, there is already so much done, the graph thing will be imported so not so much coding, next big challenge is testing the things and then writing interfaces and nice examples | 20:23 |
lambday | lisitsyn: lol man! :D | 20:23 |
gsomix | wiking, english? oh, maybe russian+vodka? | 20:23 |
@HeikoS | lambday: yeah you should stay here :) | 20:23 |
@wiking | gsomix: so wtf is that linereader is doing exactly... dont wanna look at the code ... but basically what are your assumptions... beceause it might be that you wanna read a file that doesn't exist, because of the way cmake is handling builds.... | 20:24 |
@sonney2k | gsomix, well I am dying with this - I want a line in an ascii file to be a column vector in shogun | 20:24 |
@sonney2k | gsomix, so what do you name it? | 20:24 |
@sonney2k | c or fortran order? | 20:24 |
gsomix | sonney2k, line == column vector (feature vector)? it's fortran order. | 20:24 |
lambday | HeikoS: yeah... just once december is over I am again back to work for shogun! | 20:24 |
lisitsyn | wiking: that's crazy! they don't seem to have something well known for that | 20:25 |
@sonney2k | gsomix, no C order :D | 20:25 |
@sonney2k | haha | 20:25 |
* sonney2k dies | 20:25 | |
gsomix | sonney2k, nooo. | 20:25 |
gsomix | sonney2k, look. | 20:25 |
gsomix | step by step | 20:25 |
@sonney2k | it is C order | 20:25 |
gsomix | nope | 20:26 |
@sonney2k | yes | 20:26 |
@sonney2k | one row == one vector | 20:26 |
@sonney2k | fortran would be | 20:26 |
@sonney2k | one column == one vector | 20:26 |
@sonney2k | it is all crazily confusing for ascii | 20:27 |
@sonney2k | the SGVector -> SGMatrix function is correct IMHO | 20:27 |
@sonney2k | and the CSVFile too | 20:27 |
@sonney2k | only the test not | 20:27 |
gsomix | sonney2k, so, maybe we need make just transpose flag? | 20:28 |
@wiking | lisitsyn: i dont know how to disable -Wc++11-narrowing | 20:28 |
gsomix | not order or so | 20:28 |
@sonney2k | gsomix, maybe that is easier to grasp | 20:28 |
lisitsyn | wiking: anything like -Wnoc++11-narrowing? | 20:28 |
@sonney2k | gsomix, so you could document it with one line == one column vector of the matrix | 20:28 |
@wiking | lisitsyn: clang: clang: warning: unknown warning option '-Wnoc++11-narrowing'; did you mean '-Wc++11-narrowing'? | 20:29 |
@wiking | warning: unknown warning option '-Wnoc++11-narrowing'; did you mean '-Wc++11-narrowing'? | 20:29 |
@sonney2k | and one can transpose the result if some transpose flag is set | 20:29 |
lisitsyn | wiking: just guess ;) | 20:29 |
@sonney2k | gsomix, only issue I have currently is that you need twice the memory for loading ascii data | 20:29 |
@wiking | lisitsyn: -Wno-c++11-narrowing | 20:29 |
@wiking | maaaybe | 20:29 |
@wiking | :) | 20:29 |
gsomix | sonney2k, yeah. either twice read or twice memory. | 20:30 |
@sonney2k | gsomix, I would rather read the data twice ( it will be cached anyway if small) | 20:30 |
@sonney2k | anyways I am off for today | 20:30 |
@sonney2k | cu | 20:30 |
gsomix | sonney2k, ok! | 20:30 |
@wiking | lisitsyn: yeah it was mine as well :P | 20:30 |
lisitsyn | wiking: so did it work? | 20:30 |
@sonney2k | gsomix, I used to read data twice and so was able to process BIIIG datasets | 20:30 |
@wiking | lisitsyn: we'll see :) | 20:31 |
gsomix | sonney2k, ok, thanks! | 20:31 |
@sonney2k | gsomix, so just do the same. reading twice is not an issue due to cache (if the file fits in memory) | 20:31 |
@sonney2k | gsomix, so you would just read the first line | 20:31 |
@sonney2k | gsomix, determine how many columns you have | 20:31 |
@sonney2k | gsomix, then count the number of lines | 20:31 |
@sonney2k | then alloc structures and read it for real | 20:32 |
* sonney2k afk | 20:32 | |
@wiking | sonney2k: ping | 20:38 |
@wiking | HeikoS: so ipython | 20:39 |
@wiking | HeikoS: this is the purger: https://gist.github.com/minrk/3719849 | 20:39 |
@wiking | HeikoS: but where's the generator | 20:39 |
@wiking | ? | 20:39 |
@wiking | HeikoS: ah ok other mail | 20:42 |
@wiking | lemme test them | 20:42 |
gsomix | wiking, FILE* fin=fopen("io/LineReader_unittest.cc", "r"); | 20:43 |
@wiking | gsomix: ok this will change.. | 20:44 |
-!- gsomix_ [~gsomix@178.45.71.206] has joined #shogun | 20:47 | |
-!- gsomix [~gsomix@178.45.109.87] has quit [Ping timeout: 240 seconds] | 20:48 | |
-!- travis-ci [~travis-ci@ec2-107-21-195-131.compute-1.amazonaws.com] has joined #shogun | 20:48 | |
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: http://travis-ci.org/shogun-toolbox/shogun/builds/9954976 | 20:48 |
-!- travis-ci [~travis-ci@ec2-107-21-195-131.compute-1.amazonaws.com] has left #shogun [] | 20:48 | |
-!- zxtx [~zv@cpe-75-83-151-252.socal.res.rr.com] has quit [Ping timeout: 276 seconds] | 20:51 | |
-!- gsomix_ [~gsomix@178.45.71.206] has quit [Ping timeout: 246 seconds] | 20:53 | |
lambday | I got an error while trying to compile shogun in my laptop... /shogun/mathematics/ajd/ApproxJointDiagonalizer.h:58:65: error: call of overloaded 'SGMatrix(NULL, int, int)' is ambiguous | 20:57 |
lambday | anyone else got that? | 20:57 |
lisitsyn | lambday: I am checking | 20:59 |
lambday | lisitsyn: https://github.com/lambday/shogun/blob/feature/log_determinant/src/shogun/mathematics/ajd/ApproxJointDiagonalizer.h#L58 | 21:00 |
lisitsyn | lambday: thanks | 21:00 |
lisitsyn | hmm | 21:00 |
lambday | but last one is ref counting right? | 21:00 |
lambday | and that's by default true | 21:00 |
lambday | so 3 params should be fine | 21:00 |
lisitsyn | schtupeed C++ | 21:00 |
lambday | hehe | 21:01 |
lisitsyn | he thinks he could convert int to bool | 21:01 |
lisitsyn | and NULL to int | 21:01 |
lambday | shtoopead my laptop | 21:01 |
lambday | argh | 21:01 |
lisitsyn | THAT'S WHY WE NEED NULLPTR | 21:01 |
lisitsyn | :D | 21:01 |
lambday | but but | 21:01 |
lambday | I mean does it compile fine in your machine? | 21:01 |
lisitsyn | that's exactly what is nullptr for | 21:01 |
lisitsyn | lambday: trying to compile it now | 21:01 |
lambday | lisitsyn: mine is gcc version 4.6.3 20120306 (Red Hat 4.6.3-2) (GCC) | 21:02 |
lambday | i686 | 21:03 |
lisitsyn | lambday: I see! | 21:03 |
lambday | so yours work fine | 21:03 |
lisitsyn | I think so | 21:03 |
lambday | :| | 21:03 |
lisitsyn | lambday: just make it more explicit | 21:03 |
lisitsyn | ehmm how.. :D | 21:04 |
lambday | and what? :-/ | 21:04 |
lambday | shit pickle is not here :( | 21:04 |
lambday | lisitsyn: could you please tell me a bit about the nullptr thing? | 21:04 |
lisitsyn | lambday: it won't be supported by your compiler ;) | 21:05 |
lisitsyn | lambday: it is a NULL from C++11 that can't be converted to int | 21:05 |
-!- hushell [~hushell@24.21.169.136] has joined #shogun | 21:05 | |
lambday | okay | 21:05 |
lisitsyn | at least with no brute force like reinterpret cast | 21:05 |
lisitsyn | so your compiler would not mix things up | 21:05 |
-!- gsomix_ [~gsomix@109.169.150.49] has joined #shogun | 21:05 | |
lambday | errr... reinterpret cast | 21:06 |
lisitsyn | lambday: why reinterpret cast is ok | 21:06 |
lisitsyn | in a right place | 21:06 |
lisitsyn | say you want some heterogeneous container - you could use void* pointers here | 21:06 |
lisitsyn | the only way is that reinterpret_cast then | 21:07 |
lisitsyn | shit happens :D | 21:07 |
lambday | alright | 21:07 |
lambday | hmm | 21:07 |
lisitsyn | lambday: okay my solution would be to add false | 21:07 |
lambday | so, | 21:07 |
lisitsyn | to the failing line | 21:07 |
lisitsyn | NULL,0,0,false | 21:07 |
lambday | alright | 21:07 |
lambday | yeah that should work | 21:08 |
lisitsyn | then issue a PR and I'll merge it | 21:08 |
lisitsyn | I see no better way right now | 21:08 |
lisitsyn | it is truly ambiguous | 21:08 |
lambday | lisitsyn: okay.. the only problem is, I already have some stuffs pushed to my fork.. | 21:09 |
lisitsyn | lambday: I can fix it | 21:09 |
lambday | lisitsyn: which are not ready to be merged yet | 21:09 |
lisitsyn | trust me I am an engineer | 21:09 |
lisitsyn | :D | 21:09 |
lambday | lol you're the boss :D | 21:09 |
lambday | so, I should add a separate commit with only that, and you can merge one particular commit and not the other ones | 21:09 |
lambday | (still no idea about this git thing :( ) | 21:10 |
lambday | oh | 21:10 |
lambday | shit I misinterpreted | 21:10 |
lisitsyn | lambday: I'll commit it | 21:11 |
lisitsyn | just rebase | 21:11 |
lambday | lisitsyn: yes I got you now | 21:11 |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * 7d3f37f / src/shogun/mathematics/ajd/ApproxJointDiagonalizer.h: https://github.com/shogun-toolbox/shogun/commit/7d3f37f7b5bc6f11bdf3e1caa6259d853e5d3343 | 21:11 |
shogun-notifier- | shogun: Fixes ambiguity that drives GCC 4.6.3 crazy | 21:11 |
lambday | lisitsyn: lol@the commit msg :D | 21:11 |
lisitsyn | lambday: http://cdn.lolchamp.com/wp-content/uploads/2012/09/trust-me-i-am-an-engineer.jpg | 21:11 |
lisitsyn | lambday: http://img.pandawhale.com/post-2829-Trust-me-I-m-an-engineer-pGLV.jpeg | 21:11 |
lambday | lisitsyn: hahahaha! | 21:11 |
lambday | lol | 21:12 |
lambday | I saw the first one on facebook a long time back! :D | 21:12 |
lisitsyn | that's shogun way | 21:12 |
lambday | the 2nd one is mindblowing | 21:12 |
lambday | lol.. if you see the codebase I am working on in my lab you'll faint in 20 milisecond | 21:12 |
@wiking | lo hushell | 21:14 |
@wiking | lo HeikoS | 21:14 |
lisitsyn | lambday: I used to work in way too enterprise company you won't scary me | 21:15 |
lambday | lisitsyn: lol alright but I am scared like anything! everytime I try to fix something its like cutting the green wire on a timebomb | 21:16 |
lambday | lisitsyn: I am sorry | 21:17 |
-!- gsomix_ [~gsomix@109.169.150.49] has quit [Ping timeout: 276 seconds] | 21:17 | |
lambday | there are more such errors I didn't notice | 21:17 |
lambday | ./shogun/mathematics/ajd/JADiagOrth.h:57:63: error: call of overloaded 'SGMatrix(NULL, int, int)' is ambiguous | 21:17 |
lisitsyn | lambday: still fails? | 21:17 |
lisitsyn | ahh | 21:17 |
lisitsyn | grep time | 21:18 |
lambday | and another one at line# 69 | 21:18 |
lambday | yeah grep is better | 21:19 |
lisitsyn | JADiagOrth | 21:19 |
lisitsyn | :D | 21:19 |
lisitsyn | orthodox | 21:19 |
@HeikoS | wiking: yes? | 21:20 |
lambday | hehe | 21:20 |
lambday | lisitsyn: and pickle27 is also working on something called Jedi, right? | 21:20 |
lambday | :D | 21:20 |
lisitsyn | lambday: yes kevin wan kenobi | 21:20 |
lambday | and you are master yoda | 21:21 |
lambday | calm you must remain | 21:21 |
lisitsyn | I to try be | 21:21 |
-!- gsomix_ [~gsomix@109.169.186.0] has joined #shogun | 21:22 | |
-!- gsomix_ is now known as gsomix | 21:22 | |
-!- HeikoS [~heiko@nat-166-201.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 21:22 | |
shogun-notifier- | shogun: Sergey Lisitsyn :develop * 1965e5e / src/shogun/mathematics/ajd/ (6 files): https://github.com/shogun-toolbox/shogun/commit/1965e5ead29f7b9e089d4a75275181e464b7a926 | 21:22 |
shogun-notifier- | shogun: Fixed a few more GCC 4.6.3 errors with ambiguous ambiguity | 21:22 |
lambday | rebasing | 21:23 |
lisitsyn | should work now | 21:23 |
gsomix | lisitsyn, engineering is fun. http://cs5956.vk.me/u26538480/148879729/x_2c69e9eb.jpg :) | 21:24 |
lisitsyn | it feels like a russian dormitory | 21:25 |
gsomix | lisitsyn, hell yeah. | 21:27 |
shogun-buildbot | build #1681 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1681 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 21:30 |
gsomix | lisitsyn, lambday http://cs301313.vk.me/u4095831/148937354/x_ad912382.jpg | 21:32 |
lambday | gsomix: holy mother of god! :D | 21:35 |
lambday | lol my laptop denies to serve anything else when I am compiling shogun in it.. | 21:35 |
lambday | it has been 10 mins and its still compiling :| | 21:36 |
lisitsyn | lambday: I should not have shortened this compilation output | 21:36 |
lisitsyn | at least something to read | 21:36 |
lisitsyn | I don't mind to add a downloader that prints shakespeare to the tty while compiling | 21:37 |
lisitsyn | not sure others would like it though | 21:37 |
-!- gsomix [~gsomix@109.169.186.0] has quit [Ping timeout: 240 seconds] | 21:37 | |
lambday | lisitsyn: you mean, just printing "compiling blah" is good good rather earlier what shogun had was better? | 21:38 |
lambday | is not good* I mean | 21:39 |
lisitsyn | well at least it would be more verbose and you could read things | 21:39 |
lisitsyn | while compiling shogun | 21:39 |
lambday | lisitsyn: yes but then its a hell lot of msgs, specially with eigen3 its impossible to read :( | 21:39 |
lisitsyn | as you have nothing to do | 21:39 |
-!- gsomix [~gsomix@85.26.165.90] has joined #shogun | 21:42 | |
gsomix | and one more http://cs301304.vk.me/u12861443/148937354/x_c4abd3ae.jpg | 21:43 |
lambday | 20 mins and counting... I think it worked.. and I will get disconnected in 13 mins | 21:47 |
lisitsyn | lambday: what's the thing with your connection? | 21:48 |
@wiking | ok | 21:48 |
@wiking | who wants to merge? | 21:48 |
lisitsyn | wiking: merge what? | 21:49 |
@wiking | feature/CMake -> develop | 21:49 |
lisitsyn | oh | 21:49 |
@wiking | last commit is coming up | 21:49 |
lisitsyn | is it the time? | 21:49 |
@wiking | it is time | 21:49 |
@wiking | :) | 21:49 |
lisitsyn | wiking: is it rebased? | 21:49 |
@wiking | lisitsyn: did it yesterday | 21:50 |
lambday | lisitsyn: its such a pain the the ass... its a insti policy.. from 1:30 IST to 5:00 IST they disconnect LAN so that we sleep early instead of watching sitcoms all night and playing games | 21:50 |
@wiking | but we are again apart i'm afraid... although not too much | 21:50 |
@wiking | lisitsyn: i fear that another rebase will kill the shit | 21:50 |
lisitsyn | lambday: man I thought russia is a stupid place to live | 21:50 |
lisitsyn | you win | 21:50 |
lisitsyn | :D | 21:50 |
@wiking | so maybe just a merge into develop and that's it | 21:50 |
@wiking | but i dont know | 21:50 |
lambday | lisitsyn: these IITs implemented this policy few years back and I never saw any such thing anywhere else! | 21:51 |
lambday | their logic is that we don't attend morning classes cause we sleep late :D | 21:51 |
lisitsyn | lambday: yes that's pure no-brainer | 21:52 |
lambday | lisitsyn: tell me about it! | 21:53 |
lisitsyn | lambday: you've got to change place ;) | 21:53 |
lambday | lisitsyn: if I get a job in december I'll be a free bird :D | 21:53 |
lisitsyn | sorry I forgot - are you a masters student now? | 21:54 |
lisitsyn | wiking: just merge it then ;) | 21:54 |
-!- travis-ci [~travis-ci@ec2-54-226-143-194.compute-1.amazonaws.com] has joined #shogun | 21:54 | |
travis-ci | [travis-ci] it's Sergey Lisitsyn'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/9957076 | 21:54 |
-!- travis-ci [~travis-ci@ec2-54-226-143-194.compute-1.amazonaws.com] has left #shogun [] | 21:54 | |
@wiking | lisitsyn: hehehe :) | 21:54 |
lambday | lisitsyn: yes sir | 21:55 |
lisitsyn | lambday: sir? ;) | 21:55 |
@wiking | lisitsyn: i'm still trying to test the f*ing -Wno-c11-narrowmyass flag | 21:55 |
lisitsyn | wiking: so does it help to not narrow your ass? | 21:55 |
@wiking | lisitsyn: dunno... still compiling :) | 21:55 |
@wiking | lisitsyn: wonder what the swig ppl will do about that | 21:56 |
lambday | lisitsyn: you are master yoda after all :) | 21:56 |
@wiking | lisitsyn: at least travis says that that patch is good to merge ;) | 21:56 |
@wiking | https://travis-ci.org/swig/swig/builds/9937766 | 21:56 |
@wiking | :D | 21:56 |
lisitsyn | wiking: they will do the thing to solve it! | 21:56 |
lisitsyn | in a month | 21:56 |
lisitsyn | :D | 21:56 |
@sonney2k | gsomix, heya! | 21:56 |
@wiking | :D | 21:57 |
lisitsyn | lambday: it is 3 minutes for you | 21:57 |
@sonney2k | gsomix, tests fixed? | 21:57 |
lambday | lisitsyn: :'( yeah :'( | 21:57 |
lambday | lisitsyn: and shogun still compiling :'( | 21:58 |
lisitsyn | what would you say before your 'death'? | 21:58 |
lambday | lisitsyn: I just wanna die watching a successful compilation :'( | 21:58 |
@wiking | heheh warning: implicit conversion turns floating-point number into bool: 'float64_t' (aka 'double') to 'bool' [-Wimplicit-conversion-floating-point-to-bool] | 21:58 |
@wiking | vector[i]=CMath::atan2(vector[i], x); | 21:58 |
@wiking | wtf man :) | 21:58 |
@wiking | ~^~~~~~~~~~~~~~~~~~~~~~~~~~ | 21:58 |
lisitsyn | wiking: implicit-my-shit-out-to-bool | 21:59 |
@sonney2k | lambday, please email me | 21:59 |
lisitsyn | that's what they were thinking of? some crazy warnings? | 21:59 |
lisitsyn | wiking: where is bool here? | 22:01 |
@wiking | shogun/lib/SGVector.cpp:1410:16: note: in instantiation of member function 'shogun::SGVector<bool>::atan2' requested here | 22:02 |
@wiking | template class SGVector<bool>; | 22:02 |
lisitsyn | sonney2k: need you now | 22:03 |
lisitsyn | see the mail | 22:03 |
lisitsyn | wiking: ohh interesting :D | 22:04 |
-!- lambday [67157d4f@gateway/web/freenode/ip.103.21.125.79] has quit [Ping timeout: 250 seconds] | 22:04 | |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has joined #shogun | 22:05 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 22:05 | |
@iglesiasg | good evening! | 22:05 |
@wiking | hahaha fucking helll | 22:05 |
@wiking | this swig vs c++11 is getting better and better | 22:05 |
lisitsyn | iglesiasg: hey | 22:06 |
shogun-buildbot | build #1680 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://www.shogun-toolbox.org/buildbot/builders/deb1%20-%20libshogun/builds/1680 blamelist: Sergey Lisitsyn <lisitsyn.s.o@gmail.com> | 22:06 |
@wiking | maybe swig + python is just not ready for c++11 | 22:06 |
@wiking | http://pastebin.com/iG0YJDJB | 22:07 |
@wiking | fix ideas? | 22:07 |
lisitsyn | wiking: hemhmem | 22:12 |
lisitsyn | what? | 22:12 |
lisitsyn | wiking: I think swig is not ready | 22:13 |
lisitsyn | :D | 22:13 |
-!- zxtx [~zv@rrcs-76-79-81-162.west.biz.rr.com] has joined #shogun | 22:13 | |
@wiking | ok so PythonModular -> no c++11 ? | 22:14 |
lisitsyn | wiking: well it looks like it is swig generated code | 22:15 |
gsomix | sonney2k, early in morning. and string reading. falling asleep... | 22:15 |
gsomix | nite | 22:15 |
lisitsyn | which can't be amended | 22:15 |
@sonney2k | lisitsyn, well sure you are in then but only if you talk to people | 22:15 |
@sonney2k | otherwise it won't be fun for you | 22:16 |
lisitsyn | sonney2k: again? :) | 22:16 |
@sonney2k | lisitsyn, again what? | 22:16 |
lisitsyn | I think we talked it out | 22:16 |
@sonney2k | lisitsyn, no we just chatted I didn't even have a chance to try my russian on you | 22:17 |
lisitsyn | sonney2k: lets share a plane and then you can try it | 22:17 |
@wiking | mmm maaybe this is only clang + swig + c++11 ... lemme check | 22:17 |
-!- gsomix [~gsomix@85.26.165.90] has quit [Ping timeout: 256 seconds] | 22:20 | |
@wiking | bingo.... it's clang + c++11 + swig | 22:31 |
lisitsyn | wiking: bullshit bingo | 22:33 |
lisitsyn | :D | 22:33 |
@wiking | man this matrix of what fails and not is amazing | 22:37 |
-!- travis-ci [~travis-ci@ec2-54-226-143-194.compute-1.amazonaws.com] has joined #shogun | 22:39 | |
travis-ci | [travis-ci] it's Sergey Lisitsyn'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/9957573 | 22:39 |
-!- travis-ci [~travis-ci@ec2-54-226-143-194.compute-1.amazonaws.com] has left #shogun [] | 22:39 | |
@sonney2k | wiking, I guess we would be fine w/o c++11 | 22:47 |
@wiking | sonney2k: ah it's ok i've realised that the problem is in the case when libc++ is being used instead of libstdc++ | 22:47 |
@wiking | sonney2k: i'm just verifying now this... | 22:48 |
@wiking | but until now it seems that's the case... | 22:48 |
@sonney2k | wiking, wtf is libc++ / libstc++ ? I mean you use g++ / clang++ for a reason | 22:48 |
@sonney2k | to exactly not having to supply some c++ lib | 22:48 |
@wiking | well it's not that simple | 22:48 |
@wiking | ;) | 22:48 |
@wiking | in most linux distros stdc++ is tweaked | 22:49 |
@wiking | for clang to support c++11 | 22:49 |
@wiking | in other OSes | 22:49 |
@wiking | it's not liek that... | 22:49 |
@sonney2k | asdf | 22:49 |
@wiking | so if u want c++11 then u need libc++ | 22:49 |
@iglesiasg | all right my bed is calling for me | 22:59 |
@iglesiasg | good night! | 22:59 |
-!- iglesiasg [~iglesias@s83-179-44-135.cust.tele2.se] has quit [Quit: Ex-Chat] | 22:59 | |
shogun-notifier- | shogun: Viktor Gal :feature/CMake * 7b14039 / .travis.yml,CMakeLists.txt,src/interfaces/python_modular/CMakeLists.txt: https://github.com/shogun-toolbox/shogun/commit/7b14039483482a2c5bf70a5c4f5e0ed97d0640b0 | 23:02 |
shogun-notifier- | shogun: Fix -Wno-c++11-narrowing flag and disable c++11 on Darwin with | 23:02 |
shogun-notifier- | shogun: PythonModular | 23:02 |
shogun-notifier- | shogun: Travis: install jinja2 on osx job | 23:02 |
@wiking | ok this should do it | 23:02 |
@wiking | ok | 23:03 |
@wiking | fuckit | 23:03 |
@wiking | who's up for a merge? | 23:03 |
@sonney2k | wiking, merge dammit! | 23:05 |
@sonney2k | DOOOOOOOOOOOOOOOOOOOOOOOO it! | 23:05 |
@wiking | sonney2k: do u reckon this is not going to break develop branch? | 23:05 |
@wiking | because i did a weird rebase+merge in feature/CMake branch | 23:06 |
@sonney2k | sure it should not break | 23:06 |
@sonney2k | otherwise I will send over same mad vodka drinking russians | 23:06 |
@sonney2k | (accompanied by their bears) | 23:07 |
@sonney2k | to give you a treat | 23:07 |
@wiking | :> | 23:07 |
@wiking | dunno man... | 23:07 |
@wiking | ok i'll do locally a git-flow feature finish CMake | 23:08 |
@wiking | should i do a rebase...? | 23:09 |
@wiking | or just finish | 23:09 |
@wiking | ok i'll read this fucking gitflow manual | 23:09 |
@sonney2k | wiking, wait rebase is not possible on a published branch | 23:12 |
@sonney2k | wiking, you have to merge somehow so I guess finish is the only possible thing | 23:12 |
@wiking | sonney2k: i've rebased develop | 23:14 |
@wiking | i mean | 23:14 |
@wiking | in my feature/CMake i did a git rebase develop | 23:14 |
@wiking | nevermind i'll try this feature finish | 23:15 |
@wiking | how convenient for travis to die just now | 23:16 |
@sonney2k | wiking, but then you have to force push your branch right? | 23:17 |
@wiking | wasn't force pushing | 23:20 |
@wiking | pulled -> merge -> push | 23:20 |
@sonney2k | hmmhh | 23:21 |
@wiking | god bless the restart job button on travis | 23:21 |
@sonney2k | then that wasn't a reabse | 23:21 |
lisitsyn | wiking: so what's happenning? | 23:59 |
--- Log closed Thu Aug 08 00:00:06 2013 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!