--- Log opened Thu Nov 24 00:00:29 2016 | ||
@wiking | mmm | 01:12 |
---|---|---|
@wiking | osx does not have openmp | 01:12 |
@wiking | still | 01:12 |
@wiking | i mean the official clang in it | 01:12 |
@wiking | and i wonder when are we ever gonna get openmp support in the official osx clang | 01:19 |
@sukey | Pull Request #3562 "LinalgRefactor - Matrix elementwise product" synchronized by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/3562 | 04:19 |
@sukey | Pull Request #3563 "LinalgRefactor - Remove old linalg::range_fill and linalg::max" opened by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/3563 | 05:30 |
-!- praisethemoon [~praisethe@41.226.248.123] has joined #shogun | 08:50 | |
-!- praisethemoon [~praisethe@41.226.248.123] has quit [Changing host] | 09:06 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has joined #shogun | 09:06 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has quit [Quit: Leaving] | 10:31 | |
-!- praisethemoon [~praisethe@41.226.248.123] has joined #shogun | 10:58 | |
-!- praisethemoon [~praisethe@41.226.248.123] has quit [Changing host] | 10:58 | |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has joined #shogun | 10:58 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 11:52 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:52 | |
CaBa | wiking: still nothing with xcode 8 either? | 12:28 |
@wiking | - | 12:28 |
@wiking | nada | 12:28 |
@wiking | HeikoS, ping | 12:30 |
@HeikoS | wiking: pong | 12:30 |
@wiking | just a sec | 12:30 |
@HeikoS | CaBa: this weekend I will check the parpameter tree! | 12:30 |
@HeikoS | wiking: I want to have another chat with you on video soon | 12:30 |
@HeikoS | have some ideas about how to push the stuff we discussed with gunnar | 12:30 |
@wiking | kk | 12:30 |
@wiking | ping me anytime | 12:31 |
@HeikoS | ok | 12:32 |
@HeikoS | wiking: I mean basically | 12:32 |
@HeikoS | this idea of pushing the framework one level higher | 12:32 |
@HeikoS | in the sense that it simply offers an environment to *easily* write scientific code | 12:32 |
@HeikoS | and offers all the help: math, data managment, modularity, interfaces, testing, build | 12:33 |
CaBa | wiking: is there still any correspondence between the apple llvm / clang releases and the upstream original version numbers? | 12:33 |
@HeikoS | and then all the algos we have are simply plugins for that | 12:33 |
CaBa | so confusing... | 12:33 |
CaBa | HeikoS: oh, great, thanks! | 12:33 |
@HeikoS | that is cool | 12:33 |
@HeikoS | wiking: it would also simplify the build etc | 12:33 |
@HeikoS | wiking: but I want to discuss how to approach this | 12:33 |
@HeikoS | wiking: because I have the feeling that it sounds like "rewrite" ;) | 12:33 |
CaBa | .oO(he said the r-word. now there's postapocalyptical silence...) | 12:47 |
@wiking | HeikoS, | 12:48 |
@wiking | so there are couple of amazing stuff | 12:48 |
@wiking | :) | 12:48 |
@wiking | get_kernel_matrix in kernel | 12:49 |
@wiking | it seems we are lucky | 12:49 |
@HeikoS | ah | 12:49 |
@HeikoS | man | 12:49 |
@wiking | that it's correct in linux | 12:49 |
@HeikoS | this thing | 12:49 |
@wiking | everywhere else that is incorrect | 12:49 |
@HeikoS | haha | 12:49 |
@wiking | i'm currently writing a unit test | 12:49 |
@HeikoS | good! | 12:49 |
@wiking | that make sure that it's correct | 12:49 |
@wiking | and then change the implementation to use openmp | 12:49 |
@HeikoS | wiking: do you have a bit time now for a chat? | 12:49 |
@HeikoS | wiking: yep that sounds good | 12:49 |
@HeikoS | wiking: like in 5-10 mins for say 15 mins | 12:50 |
@HeikoS | just finishing a review | 12:50 |
@wiking | skype hangout whatspapp? | 12:50 |
@wiking | :) | 12:50 |
@wiking | whatsapp has voice | 12:50 |
@wiking | so we can do that | 12:50 |
@wiking | but yeah i'm here | 12:50 |
@wiking | so just ping me | 12:50 |
@HeikoS | cool | 12:52 |
@HeikoS | video maybe | 12:52 |
@HeikoS | ill ping in 5 | 12:52 |
@HeikoS | wiking: ok now? | 12:55 |
@wiking | y | 12:55 |
@wiking | just gimme a link where | 12:56 |
@wiking | :) | 12:56 |
@HeikoS | hangout | 12:56 |
@HeikoS | kk | 12:56 |
@wiking | need a hangout link | 12:56 |
@wiking | invite me to an event | 12:56 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 245 seconds] | 13:30 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 13:32 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 13:32 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 256 seconds] | 14:41 | |
CaBa | wiking: ping | 16:15 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 16:31 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:31 | |
@sukey | Pull Request #3563 "LinalgRefactor - Remove old linalg::range_fill and linalg::max" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3563 | 16:35 |
@sukey | New Commit "Merge pull request #3563 from OXPHOS/linalg_rm_rangefill_and_max | 16:35 |
@sukey | LinalgRefactor - Remove old linalg::range_fill and linalg::max" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/a26d89b95aa61dabdcd82f6090f8392513dd74b9 | 16:35 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 17:07 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 17:22 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:22 | |
@sukey | Pull Request #3538 "temporary workaround for #3460" synchronized by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3538 | 17:22 |
@sukey | Pull Request #3538 "fix for #3460" - https://github.com/shogun-toolbox/shogun/pull/3538 | 17:23 |
@sukey | Pull Request #3538 "fix for #3460" synchronized by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3538 | 17:24 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 268 seconds] | 17:29 | |
@sukey | Pull Request #3534 "LinalgRefactor - Cholesky - CPU only" synchronized by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/3534 | 17:35 |
CaBa | wiking: HeikoS mentioned changes regarding builds on os x recently. both my development and production systems are mac os systems (i7-2620M is my desktop, Xeon-X5675 the production system) | 17:38 |
@sukey | Pull Request #3533 "LinalgRefactor - Remove old linalg::scale" synchronized by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/3533 | 17:39 |
CaBa | wiking: the latter is currently much slower (!) on my small test dataset... | 17:39 |
CaBa | wiking: could there be any explanation for this within shogun? | 17:40 |
-!- praisethemoon [~praisethe@unaffiliated/praisethemoon] has quit [Ping timeout: 250 seconds] | 18:07 | |
-!- HeikoS [~heiko@89.105.104.229] has joined #shogun | 19:03 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 19:03 | |
@sukey | New branch feature/kernel_exp_family_base_measure created on shogun-toolbox/shogun | 19:04 |
@sukey | New Commit "temp commit" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/938a006259b5a520cb29b4417e0ce65e8aceeda2 | 19:04 |
CaBa | HeikoS: ping | 19:23 |
CaBa | 17:37:11 < CaBa> wiking: HeikoS mentioned changes regarding builds on os x recently. both my development and production systems are mac os systems (i7-2620M is my desktop, Xeon-X5675 the production system) | 19:23 |
CaBa | 17:37:41 < CaBa> wiking: the latter is currently much slower (!) on my small test dataset... | 19:23 |
CaBa | 17:38:52 < CaBa> wiking: could there be any explanation for this within shogun? | 19:23 |
CaBa | HeikoS: this does not by any change ring a bell or so? just checking. | 19:23 |
-!- HeikoS [~heiko@89.105.104.229] has quit [Ping timeout: 265 seconds] | 19:29 | |
@wiking | CaBa, mmm | 20:42 |
@wiking | you use gcc for build no? | 20:42 |
CaBa | wiking: gcc-6 for both shogun and my application in the meantime. same versions on both systems (including stdlib). i had different versions of protobuf and hdf5 installed, but bumped that in sync also in the meantime. same effect. | 21:11 |
CaBa | wiking: it also doesn't depend where i build shogun and my program. if i move the lib + my bin to the other system, the performance changes | 21:11 |
CaBa | wiking: oh i think i know what this might be related to... | 21:27 |
@wiking | mmm | 21:29 |
@wiking | that is weird though | 21:29 |
@wiking | xeon is slower? | 21:29 |
@wiking | mmm | 21:29 |
@wiking | shooulnd't that be faster? :D | 21:29 |
CaBa | wiking: i figured it out... | 21:33 |
@wiking | what was it? | 21:33 |
CaBa | wiking: the xeon box has 24 cores... after running it through a profiler, i figured he was mostly busy with locks | 21:33 |
@wiking | is it us | 21:33 |
CaBa | wiking: too many threads | 21:33 |
@wiking | looooooooooocks | 21:34 |
@wiking | which locks? | 21:34 |
CaBa | wiking: set_num_threads(4) -> now i have the same performance as on my laptop | 21:34 |
@wiking | ah lol | 21:34 |
@wiking | ok | 21:34 |
@wiking | but fuck | 21:34 |
@wiking | CaBa, i have some fixes | 21:34 |
@wiking | i hate that you have to compile in order to change num threads | 21:34 |
@wiking | (in case of c++ | 21:34 |
@wiking | so i have now a new env var | 21:34 |
@wiking | SHOGUN_NUM_THREADS | 21:34 |
CaBa | oh, runtime | 21:34 |
@wiking | will push to develop as soon as i have a fix for kernel.cpp | 21:34 |
CaBa | neat | 21:35 |
@wiking | mmm btw | 21:36 |
@wiking | when yo urun | 21:36 |
@wiking | ctest | 21:36 |
CaBa | wiking: training / testing never uses more than 4fold cpu power for me... not even on my actual large datasets. maybe i should be turning that down to 4 in general... | 21:36 |
@wiking | don't you have several tests failing? | 21:36 |
CaBa | wiking: never run that manually. is that part of the default build target? | 21:36 |
@wiking | nono what i mean | 21:36 |
@wiking | if you do this | 21:36 |
@wiking | cmake -DENABLE_TESTING .... | 21:37 |
@wiking | and then once compiled | 21:37 |
@wiking | run ctest -R unit | 21:37 |
@wiking | does all the test passes for you? | 21:37 |
@wiking | btw what are you using for the depedency libraries | 21:37 |
@wiking | ? | 21:37 |
CaBa | i'll check, one moment | 21:37 |
@wiking | brew? | 21:37 |
CaBa | yes | 21:37 |
@wiking | ok so your eigen is already 3.3.0 | 21:37 |
CaBa | /usr/local/Cellar/eigen/3.3.0 (484 files, 6.4M) * | 21:37 |
CaBa | yep | 21:37 |
@wiking | yeah | 21:37 |
@wiking | ok it's not that big a deal | 21:38 |
@wiking | PCA will fail for you | 21:38 |
@wiking | the unit test | 21:38 |
@wiking | but it's just the test's fault | 21:38 |
@wiking | as it does not consider that if EV is really an eigen vector | 21:38 |
CaBa | -DENABLE_TESTING is not the default? | 21:38 |
@wiking | than -1.0*ev is an eigen vector as well | 21:38 |
@wiking | noup | 21:38 |
@wiking | i dont think so | 21:38 |
CaBa | ok, so i have to rerun | 21:39 |
@wiking | yeah sorry | 21:39 |
@wiking | btw | 21:39 |
CaBa | np, 24 cores :P | 21:39 |
@wiking | well | 21:39 |
@wiking | ld is 1 core | 21:39 |
@wiking | :( | 21:39 |
@wiking | and it'll take fucking 5 minutes | 21:39 |
@wiking | to link the unit test | 21:39 |
@wiking | :( | 21:39 |
@wiking | i'm currently working on a fix for it | 21:39 |
@wiking | but unfortunately i'm not sure if i can get the gcc part working as well | 21:39 |
@wiking | the clang part is tested | 21:39 |
CaBa | i'm on 18204b2, is that enough? | 21:41 |
@wiking | lemme check | 21:41 |
@wiking | mmm i think i dont even have that locally | 21:41 |
@wiking | :D | 21:41 |
CaBa | kk ;) | 21:42 |
CaBa | it's just 1 commit ago i just realized | 21:42 |
CaBa | not much changed the last 2 days | 21:42 |
CaBa | [ 99%] Built target shogun-unit-test | 21:43 |
@wiking | yeah | 21:43 |
@wiking | that'll be 5 minute | 21:43 |
@wiking | s:) | 21:43 |
CaBa | ;) | 21:43 |
CaBa | [ 99%] Built target python_doxy2swig | 21:44 |
CaBa | the evil 99%... almost like a good old microsoft progress bar ;) | 21:44 |
@wiking | :> | 21:45 |
@wiking | heheh | 21:45 |
@wiking | check if build/tests/unit/shogun-unit-test exists | 21:45 |
@wiking | if it is ok | 21:45 |
@wiking | then in a separate shell | 21:45 |
@wiking | cd build && ctest -R unit | 21:45 |
CaBa | The following tests FAILED: 84 - unit-PeriodicKernelTest (Failed) | 21:48 |
@wiking | yep yep | 21:48 |
@wiking | you'lll have a summary on the end | 21:48 |
CaBa | that's kinda the end | 21:49 |
CaBa | 99% tests passed, 1 tests failed out of 271 | 21:50 |
CaBa | Total Test time (real) = 10.20 sec | 21:50 |
CaBa | you mean that? | 21:50 |
@wiking | mmmm | 21:50 |
@wiking | on the end | 21:50 |
@wiking | ctest should give you the failed ones | 21:50 |
@wiking | a list of it | 21:50 |
CaBa | 99% tests passed, 1 tests failed out of 271 | 21:51 |
CaBa | Total Test time (real) = 10.20 sec | 21:51 |
CaBa | The following tests FAILED: 84 - unit-PeriodicKernelTest (Failed) | 21:51 |
CaBa | Errors while running CTest | 21:51 |
CaBa | that are the last lines. | 21:51 |
@wiking | lol | 21:52 |
@wiking | can you run it one more time? | 21:52 |
@wiking | btw you use gcc to compile right? | 21:52 |
CaBa | yes, still ;) | 21:52 |
CaBa | i re-ran already | 21:52 |
CaBa | same output | 21:52 |
@wiking | same error? | 21:52 |
@wiking | can you do this then | 21:53 |
CaBa | there is a more detailed log thouth | 21:53 |
@wiking | ctest -R unit-PeriodicKernelTest -V | 21:53 |
@wiking | and pastebin it | 21:53 |
CaBa | wiking: https://gitlab.unique-internet.de/snippets/10/raw | 21:53 |
@wiking | ok | 21:54 |
@wiking | uname -a? | 21:54 |
@wiking | for me it's | 21:54 |
@wiking | Darwin rdr.local 16.1.0 Darwin Kernel Version 16.1.0: Thu Oct 13 21:26:57 PDT 2016; root:xnu-3789.21.3~60/RELEASE_X86_64 x86_64 | 21:54 |
CaBa | Darwin systembio4.charite.de 15.6.0 Darwin Kernel Version 15.6.0: Thu Sep 1 15:01:16 PDT 2016; root:xnu-3248.60.11~2/RELEASE_X86_64 x86_64 | 21:54 |
CaBa | no sierra yet | 21:54 |
@wiking | k | 21:54 |
@wiking | ok i'll need your help | 21:54 |
@wiking | as soon as i think i'm ready with the fixes | 21:54 |
CaBa | shoot | 21:54 |
@wiking | which would be another | 21:54 |
@wiking | 12 hours | 21:55 |
@wiking | or so | 21:55 |
CaBa | what's the matter? what's different for you>? | 21:55 |
@wiking | for me many other tests are failing | 21:55 |
@wiking | :) | 21:55 |
@wiking | can you try it on your laptop? | 21:55 |
CaBa | laptop is on el capitan as well | 21:56 |
CaBa | i keep them as sync as possible config wise ;) | 21:56 |
CaBa | btw | 21:57 |
CaBa | -DCMAKE_CXX_COMPILER=/usr/local/bin/g++-6 -DCMAKE_C_COMPILER=/usr/local/bin/gcc-6 -DCMAKE_INSTALL_PREFIX=~/RESEARCH/external/shogun-release -DCMAKE_BUILD_TYPE=Release -DENABLE_TESTING=ON -DPythonModular=ON -DBUILD_DASHBOARD_REPORTS=OFF -DBUILD_META_EXAMPLES=OFF -DENABLE_COLPACK=OFF -DENABLE_PROTOBUF=OFF | 21:57 |
CaBa | this is my cmake config | 21:57 |
CaBa | just in case this disables things that you are talking about | 21:57 |
@wiking | mmm | 22:02 |
@wiking | ok | 22:02 |
@wiking | i'll have to check with gcc6 | 22:02 |
@wiking | as i've never tried to compile it | 22:02 |
@wiking | anyhow the reason we couldnt upgrade | 22:02 |
@wiking | shogun to 5.0.0 in brew | 22:03 |
@wiking | because of some weird shit with | 22:03 |
@wiking | with some of tests | 22:03 |
CaBa | ok | 22:03 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 22:56 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:56 | |
@sukey | Pull Request #3538 "fix for #3460" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3538 | 23:06 |
@sukey | New Commit "Merge pull request #3538 from karlnapf/bugfix/R_33_modular | 23:06 |
@sukey | fix for #3460" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/48ee7ea13a4d9510164136bc2a6963f8ec1a7c35 | 23:06 |
@sukey | Issue #3460 "Rmodular cannot be built " closed by karlnapf - https://github.com/shogun-toolbox/shogun/issues/3460 | 23:06 |
@wiking | HeikoS, ping | 23:06 |
@wiking | man | 23:06 |
@HeikoS | wiking: pong | 23:06 |
@wiking | i'm just testing our PRNG | 23:06 |
@HeikoS | which one? | 23:07 |
@wiking | any | 23:07 |
@wiking | :) | 23:07 |
@HeikoS | the Random | 23:07 |
@HeikoS | ? | 23:07 |
@HeikoS | ok | 23:07 |
@wiking | CRadom | 23:07 |
@wiking | and comparing it to std:: | 23:07 |
@wiking | (c++11 | 23:07 |
@HeikoS | ah yes | 23:07 |
@wiking | dunno what to do with this shit | 23:07 |
@wiking | because that prng is good for | 23:07 |
@wiking | vectors | 23:07 |
@wiking | but we never use fill array | 23:07 |
@wiking | always | 23:07 |
@HeikoS | ah we loop | 23:07 |
@wiking | for (int i = 0;.... | 23:07 |
@wiking | but my biggest problem | 23:08 |
@wiking | that lets say i remove | 23:08 |
@wiking | the thing and start using c++111 | 23:08 |
@wiking | that is not threadsafe | 23:08 |
@wiking | so we need to box it with a mutex | 23:08 |
@wiking | which is even shittier | 23:08 |
@wiking | so i'm wondering wtf with this whole random | 23:08 |
@HeikoS | mmmh | 23:09 |
@HeikoS | so mutexing it, is it really that bad? | 23:09 |
@HeikoS | mmmmh | 23:09 |
@wiking | well fuck | 23:09 |
@wiking | i mean not that bad | 23:09 |
@wiking | but why would we do it | 23:09 |
@wiking | i mean i'm not so sure if we should create wrappers at all | 23:09 |
@wiking | c++11 has a very nice interface | 23:09 |
@wiking | for randome | 23:09 |
@wiking | *random | 23:09 |
@HeikoS | if it is part if the language itself | 23:10 |
@HeikoS | I think we can use it | 23:10 |
@HeikoS | it is just | 23:10 |
@HeikoS | if we want to change once a cool lib comes out, | 23:10 |
@wiking | mmm yeah | 23:10 |
@wiking | true | 23:10 |
@wiking | interfaces | 23:10 |
@wiking | :))) | 23:10 |
@HeikoS | though to be fair that is mostly for external libs | 23:10 |
@HeikoS | not for c++ stuff | 23:11 |
@HeikoS | I would also use the STL more heavily these days :) | 23:11 |
@wiking | heheh yeah | 23:11 |
@HeikoS | and get rid of DynArray | 23:11 |
@wiking | anyhow i still wanna compare | 23:11 |
@HeikoS | fuck | 23:11 |
@HeikoS | yeah sure | 23:11 |
@wiking | hahaha DynArray | 23:11 |
@wiking | so lemme try to write a unit tests | 23:11 |
@HeikoS | Ill write some dev docs now | 23:11 |
@HeikoS | have a bit of time | 23:11 |
@wiking | to std_normal_distro | 23:11 |
@wiking | k | 23:11 |
@HeikoS | cool | 23:12 |
@HeikoS | I merged the R fix btw | 23:12 |
@HeikoS | manual speaks clear langauge | 23:12 |
@HeikoS | no R includes from external C | 23:12 |
@HeikoS | I mean the whole sg_print_functions.cpp business is a bit outdated anyways | 23:12 |
@HeikoS | but no priority | 23:12 |
@wiking | yes | 23:12 |
@wiking | btw | 23:13 |
@wiking | this whole shit with | 23:13 |
@wiking | Cmath | 23:13 |
@wiking | should go into a toilet | 23:13 |
@wiking | especially the random part | 23:13 |
@wiking | we should put all that to deprecated | 23:13 |
@wiking | aaaaaaaaaand | 23:13 |
@wiking | start adding acutally optimal implementation | 23:13 |
@wiking | of log, exp etc | 23:13 |
@wiking | since we have actually a wrapper for it in CMath | 23:13 |
@HeikoS | yep | 23:18 |
@HeikoS | make it issues | 23:18 |
@HeikoS | and put them to a milestone | 23:18 |
@wiking | hahahah | 23:18 |
@sukey | New Commit "Merge pull request #3533 from OXPHOS/linalg_scale_remove | 23:18 |
@sukey | LinalgRefactor - Remove old linalg::scale" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/71828bd8b4bb750a04d92fa42c245a3baadfdf00 | 23:18 |
@sukey | Pull Request #3533 "LinalgRefactor - Remove old linalg::scale" merged by karlnapf - https://github.com/shogun-toolbox/shogun/pull/3533 | 23:18 |
@HeikoS | wiking: I am serious | 23:18 |
@wiking | i mean nono | 23:19 |
@wiking | i know | 23:19 |
@wiking | but | 23:19 |
@wiking | this is fucking scary | 23:19 |
@wiking | look at this | 23:19 |
@wiking | ./examples/undocumented/libshogun/basic_random 1000000 | 23:19 |
@wiking | 35.8699 ms | 23:19 |
@HeikoS | ouch | 23:19 |
@wiking | ~/a.out 1000000 std | 23:19 |
@wiking | 140.559 ms | 23:19 |
@wiking | so | 23:19 |
@wiking | actually | 23:20 |
@wiking | :))) | 23:20 |
@wiking | we are not tHAT bad | 23:20 |
@wiking | :> | 23:20 |
@wiking | a.out - c++11 | 23:20 |
@HeikoS | do these things pass the same tests for randomness? | 23:20 |
@wiking | what do you mean pass the same tests? | 23:20 |
@wiking | i mean they are both std normal distrib | 23:20 |
@wiking | do we have a test for this? or should i actually test it myself? :) | 23:20 |
@wiking | like on a qq plot | 23:21 |
@wiking | i mean i would presume c++11 has a true std normal distrib implemented :D | 23:21 |
@HeikoS | c++11 will be correct | 23:22 |
@HeikoS | but is ours? | 23:22 |
@HeikoS | there are more sohpisticated tests than qq plot | 23:22 |
@wiking | ok so can u just give me one | 23:22 |
@wiking | :) | 23:22 |
@wiking | i mean thi sis the random | 23:22 |
@wiking | that we use like everywhere | 23:22 |
@wiking | ;)))) | 23:22 |
@wiking | jsut saying | 23:22 |
@wiking | so if its not correct we are fucked anyways | 23:22 |
@wiking | :) | 23:22 |
@HeikoS | https://www.random.org/analysis/ | 23:23 |
@wiking | TEST(Random, random_std_normal_quantiles) | 23:23 |
@wiking | we have this test | 23:23 |
@wiking | actually implemented by u | 23:23 |
@wiking | so that should assure that we are really generating a std normal distro | 23:24 |
@wiking | no? | 23:24 |
@HeikoS | no it is not enough | 23:24 |
@wiking | :D | 23:24 |
@HeikoS | just to do quick early alarms | 23:24 |
@HeikoS | if something is terribly wrong | 23:24 |
@HeikoS | but if we have code written by ourselves that generates random numbers | 23:25 |
@wiking | yeah but ok | 23:25 |
@HeikoS | we need to test for more subtle artifacts | 23:25 |
@wiking | well it's not our own code | 23:25 |
@wiking | it's DSFTM | 23:25 |
@HeikoS | ah ok | 23:25 |
@HeikoS | then ok | 23:25 |
@wiking | and then on top we apply | 23:25 |
@HeikoS | then the unit test is only for making sure the sign and scale is right | 23:25 |
@wiking | i thnk w ehave a box muller transform implemented | 23:26 |
@sukey | New Commit "Correct link to INTERFACES.md" to shogun-toolbox/shogun by karlnapf: https://github.com/shogun-toolbox/shogun/commit/a5449137a470fdfdbb254e79f31fb5575b0cb11f | 23:26 |
@HeikoS | kk | 23:26 |
@wiking | i have to check which implemntation we hae | 23:27 |
@wiking | but basically we used the uniform distrib | 23:27 |
@wiking | generated by DSFMT | 23:27 |
@HeikoS | btw | 23:27 |
@HeikoS | http://shogun.ml/cookbook/latest/index.html | 23:27 |
@HeikoS | I need the index.html | 23:28 |
@wiking | yes | 23:28 |
@wiking | you needit | 23:28 |
@wiking | :) | 23:28 |
@HeikoS | that seems like I should change it? | 23:28 |
@wiking | what? | 23:28 |
@HeikoS | http://shogun.ml/cookbook/latest | 23:28 |
@HeikoS | so that this is enough | 23:28 |
@wiking | ah you need a redirect for that | 23:28 |
@wiking | in the website | 23:28 |
@HeikoS | shall I add? | 23:28 |
@wiking | go ahead | 23:28 |
@wiking | i mean | 23:28 |
@wiking | honestly | 23:28 |
@wiking | i wouldn't do a redirect | 23:28 |
@wiking | just here | 23:28 |
@HeikoS | ok | 23:29 |
@HeikoS | so change links instead? | 23:29 |
@HeikoS | to include the index.html? | 23:29 |
@wiking | https://github.com/shogun-toolbox/shogun-web2/blob/master/shogun_web.py#L74 | 23:29 |
@wiking | here check the filename | 23:29 |
@wiking | if it's ending on latest or current | 23:30 |
@wiking | then need to append index.html | 23:30 |
@wiking | else do whatever is there | 23:30 |
@wiking | ok finally the first time i'm positively surprised about shogun in a long time | 23:31 |
@wiking | :)))) | 23:31 |
@HeikoS | haha | 23:32 |
@HeikoS | the rng stuff? | 23:32 |
@wiking | yeah | 23:34 |
@wiking | but i'm still wondering | 23:34 |
@wiking | why dont we have like an array filler | 23:34 |
@wiking | for random | 23:34 |
@wiking | i'll make every random related stuff in CMath deprecated | 23:34 |
@wiking | [[deprecated]](C++14) | 23:35 |
@wiking | [[deprecated("reason")]](C++14) | 23:35 |
@wiking | :DDD | 23:35 |
@wiking | c++14 feature | 23:35 |
@wiking | amazing | 23:35 |
@HeikoS | yes | 23:35 |
@HeikoS | would be good to have a deprecated marker in shogun in fact | 23:36 |
@HeikoS | but not sure how much that helps | 23:36 |
@HeikoS | better fix things | 23:36 |
@wiking | man | 23:36 |
@wiking | you know how many palces | 23:36 |
@wiking | we have CMath::random | 23:36 |
@wiking | so i'lljust now put a deprecated shit | 23:36 |
@wiking | everywhere | 23:36 |
@wiking | and then change slowly | 23:36 |
@wiking | mmm | 23:36 |
@wiking | wonder where should i add the deprecated macro | 23:36 |
@wiking | it goes into common.h | 23:37 |
@wiking | :) | 23:37 |
@wiking | i'm wondering whether we should depend on this https://github.com/foonathan/compatibility | 23:40 |
@HeikoS | I am not sure about deprecated | 23:42 |
@HeikoS | what does it help us marking these things? | 23:42 |
@HeikoS | http://buildbot.shogun-toolbox.org/builders/shogun%20-%20website/builds/59/steps/shell/logs/stdio | 23:43 |
@HeikoS | wiking: ^ | 23:43 |
@HeikoS | saw that guy? | 23:43 |
@wiking | yes for the 2nd time | 23:43 |
@wiking | ignore it | 23:43 |
@wiking | :) | 23:43 |
@wiking | :) | 23:43 |
@HeikoS | kk | 23:44 |
@HeikoS | mmh | 23:44 |
@HeikoS | I broke the cookbooks with my redirect | 23:44 |
@HeikoS | can you quickly press "examples" on website and tell me what causes that? | 23:45 |
@HeikoS | http://shogun.ml/cookbook/latest/index.html | 23:45 |
@HeikoS | vs | 23:45 |
@HeikoS | http://shogun.ml/cookbook/latest | 23:45 |
@HeikoS | maybe I revert and add back the index.html | 23:45 |
@HeikoS | seems easier | 23:45 |
@wiking | ? | 23:45 |
@wiking | didn't you test locally? | 23:46 |
@HeikoS | I did | 23:47 |
@wiking | the css seems to be fucked up | 23:47 |
@wiking | btw | 23:47 |
@wiking | http://shogun.ml/cookbook/latest/ | 23:47 |
@wiking | its 404 | 23:47 |
@wiking | and that shouldn't be | 23:47 |
@HeikoS | yeah | 23:48 |
@HeikoS | mmg | 23:48 |
@HeikoS | I revert this one and leave index.html | 23:48 |
@HeikoS | no nerves for this now | 23:48 |
@wiking | why? | 23:48 |
@HeikoS | locally I get http://127.0.0.1:8203/cookbook/latest | 23:48 |
@HeikoS | so cannot test | 23:48 |
@wiking | just fix it correctly | 23:48 |
@wiking | <link rel="stylesheet" href="_static/basic.css" type="text/css" /> | 23:48 |
@wiking | you have this | 23:48 |
@wiking | it's not absolute but relative | 23:48 |
@wiking | if you get http://shogun.ml/cookbook/latest -> http://shogun.ml/cookbook/latest/ | 23:48 |
@wiking | same for current | 23:48 |
@wiking | and that's all | 23:49 |
@wiking | pretty simple | 23:49 |
@HeikoS | not sure I understand what you are saying, but thats because i actually dont know stuff about this :) | 23:49 |
@HeikoS | so lets see | 23:49 |
@wiking | but fuck | 23:49 |
@wiking | i mean | 23:49 |
@wiking | <link rel="stylesheet" href="_static/basic.css" type="text/css" /> | 23:49 |
@wiking | thi sshit referes | 23:49 |
@wiking | to something that is relative path | 23:49 |
@wiking | now if you have | 23:49 |
@wiking | http://shogun.ml/cookbook/latest | 23:50 |
@wiking | and then append _static/basic.css | 23:50 |
@wiking | then you get | 23:50 |
@wiking | http://shogun.ml/cookbook/latest_static/basic.css | 23:50 |
@wiking | which is wrong obviously | 23:50 |
@wiking | hence the shitty looking stuff now | 23:50 |
@wiking | so if your url is | 23:50 |
@wiking | .endswith('latest') or endswith('current') | 23:50 |
@wiking | you should append a / | 23:50 |
@wiking | to the url itself | 23:51 |
@wiking | not to the file reading part | 23:51 |
@wiking | but i mean | 23:51 |
@wiking | the hillarious part is | 23:51 |
@wiking | that this has been solved | 23:51 |
@wiking | like 15 years ago | 23:51 |
@wiking | in apache | 23:51 |
@wiking | now fucking python shit fucking flask | 23:51 |
@wiking | needed of course | 23:51 |
@wiking | to reinvent the fucking wheel | 23:51 |
@wiking | and here we are | 23:51 |
@wiking | 2016 | 23:51 |
@wiking | you have to deal with correct fucking URLs | 23:52 |
@wiking | i wouldn't actually do a redirect | 23:54 |
@wiking | for /examples | 23:54 |
@wiking | because it's fucking stupid | 23:54 |
@wiking | just add a new route on top of | 23:54 |
@wiking | cookbook | 23:54 |
@HeikoS | no idea | 23:54 |
@HeikoS | I reverted the commit for now, easier to separate out what fucked up this way | 23:55 |
@wiking | ok | 23:55 |
@wiking | but then dont change things in future? :) | 23:55 |
@HeikoS | I want it different | 23:55 |
@HeikoS | but I have like one hour now that I wanted to write some docs | 23:55 |
@HeikoS | and not deal with http links ;) | 23:55 |
@wiking | yeah but fuck | 23:55 |
@wiking | then why do you do it | 23:55 |
@wiking | ? | 23:55 |
@HeikoS | wrong decision | 23:57 |
--- Log closed Fri Nov 25 00:00:30 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!