--- Log opened Thu Jun 22 00:00:23 2017 | ||
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3812 | 00:30 |
---|---|---|
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3812 merged by vigsterkr | 00:30 |
@sukey | [https://github.com/shogun-toolbox/shogun] vigsterkr pushed 2 commits: | 00:30 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/41f2ca4995d517c0729363e0f854bddb03af997c | 00:30 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/2def19500536f1408914d7cb5a2247d547204157 | 00:30 |
-!- sukey [~nodebot@ks312251.kimsufi.com] has quit [Remote host closed the connection] | 00:46 | |
-!- sukey [~nodebot@ks312251.kimsufi.com] has joined #shogun | 00:47 | |
-!- mode/#shogun [+o sukey] by ChanServ | 00:47 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 01:51 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 01:51 | |
shogun-buildbot | build #70 of nightly jessie deb is complete: Failure [failed dput] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly%20jessie%20deb/builds/70 blamelist: Giovanni De Toni <giovanni.det@gmail.com>, Viktor Gal <vigsterkr@gmail.com>, MikeLing <sabergeass@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 03:01 |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-oewvncwnqrkuhecv] has joined #shogun | 03:17 | |
@wiking | shogun-buildbot: force build --branch=develop 'trusty - libshogun - viennacl' | 03:32 |
shogun-buildbot | build #286 forced | 03:32 |
shogun-buildbot | I'll give a shout when the build finishes | 03:32 |
shogun-buildbot | build #286 of trusty - libshogun - viennacl is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/286 | 03:35 |
@sukey | [https://github.com/shogun-toolbox/shogun] vigsterkr pushed 2 commits: | 04:11 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/6e0fe4ca2f929806fbbeb54b14b00c7bf3516125 | 04:11 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/dc5c4f2af33e7d5f9dd7ed413dc4e05af26d0dc5 | 04:11 |
shogun-buildbot | build #287 of trusty - libshogun - viennacl is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/287 blamelist: Viktor Gal <viktor.gal@maeth.com> | 04:22 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/7f0ec8896951cd870a03aca057435276c5bb0d4f by vigsterkr | 04:23 |
shogun-buildbot | build #289 of trusty - libshogun - viennacl is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/289 | 04:30 |
-!- travis-ci [~travis-ci@ec2-54-221-162-143.compute-1.amazonaws.com] has joined #shogun | 05:02 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/245613627 | 05:02 |
-!- travis-ci [~travis-ci@ec2-54-221-162-143.compute-1.amazonaws.com] has left #shogun [] | 05:02 | |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/a86dfda35b5d7d178fb3fbe55e409ab7afd914cb by vigsterkr | 06:08 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/df72b5f7f095f7c6ef9278e4ba22c91b9791db1b by vigsterkr | 07:26 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/d027609dbb4a19b121a441636c913a6720ffaf45 by vigsterkr | 07:35 |
-!- travis-ci [~travis-ci@ec2-54-82-70-90.compute-1.amazonaws.com] has joined #shogun | 08:03 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/245644875 | 08:03 |
-!- travis-ci [~travis-ci@ec2-54-82-70-90.compute-1.amazonaws.com] has left #shogun [] | 08:03 | |
-!- TingMiao [uid229534@gateway/web/irccloud.com/x-ekdazixzuyuobllp] has joined #shogun | 08:17 | |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has joined #shogun | 09:09 | |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has quit [Ping timeout: 260 seconds] | 09:16 | |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has joined #shogun | 09:17 | |
-!- geektoni_ [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has joined #shogun | 09:21 | |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has quit [Ping timeout: 260 seconds] | 09:22 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3848 synchronized by geektoni | 09:49 |
shogun-buildbot | build #218 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/218 | 10:46 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 10:53 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 10:53 | |
micmn | wiking: ping | 11:11 |
@wiking | pong | 11:11 |
@wiking | micmn, rebase rebase rebase :) | 11:11 |
micmn | yeah I'm on it | 11:11 |
@wiking | thnx | 11:11 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3711 synchronized by MikeLing | 11:11 |
micmn | some changes in class_list.py that I had to merge | 11:12 |
micmn | and the global fixture thing too | 11:12 |
micmn | I was looking at the SVMOcas unittest | 11:12 |
micmn | https://github.com/MikeLing/shogun/blob/41f2ca4995d517c0729363e0f854bddb03af997c/tests/unit/classifier/svm/SVMOcas_unittest.cc | 11:12 |
micmn | same labels used for train and test? | 11:12 |
@wiking | mikeling, ^ | 11:13 |
mikeling | micmn: yes | 11:13 |
@wiking | https://github.com/MikeLing/shogun/blob/41f2ca4995d517c0729363e0f854bddb03af997c/tests/unit/classifier/svm/SVMOcas_unittest.cc#L23 | 11:14 |
@wiking | isn't this a typo? | 11:14 |
@wiking | CBinaryLabels* ground_truth = (CBinaryLabels*)mockData->get_labels_train() | 11:14 |
@wiking | ? | 11:14 |
@wiking | or? | 11:14 |
micmn | I guess it's the same, train and test of the same size and labels split 50/50? but... | 11:15 |
@wiking | ah but still i guess it's a bit misleading | 11:16 |
@wiking | the naming | 11:16 |
mikeling | wiking: micmn They are the same | 11:16 |
@wiking | so for the ground trught | 11:16 |
mikeling | the train and test label | 11:16 |
@wiking | *truth semantically it should be _train | 11:16 |
micmn | yeah | 11:16 |
@wiking | mikeling, basically the order of the examples are the same that's what you are saying | 11:17 |
@wiking | they are not the same examples | 11:17 |
@wiking | but thye are having the same labels... | 11:17 |
@wiking | it'd be still good to have it fixed | 11:18 |
@wiking | CBinaryLabels* ground_truth = (CBinaryLabels*)mockData->get_labels_train() | 11:18 |
@wiking | so that semantically one gets | 11:18 |
@wiking | what is this about | 11:18 |
@wiking | https://github.com/MikeLing/shogun/blob/41f2ca4995d517c0729363e0f854bddb03af997c/tests/unit/classifier/svm/SVMOcas_unittest.cc#L36 | 11:18 |
@wiking | should compare with get_labels_test() | 11:18 |
@wiking | even if it's the same (as the order of it is the same) | 11:18 |
@wiking | just to keep the semantics correct | 11:18 |
mikeling | wiking ok, I will fix it in next push | 11:19 |
@wiking | k thnx | 11:19 |
mikeling | np | 11:19 |
micmn | wiking, mikeling: mmm what should I use for RegressionLabels? | 11:26 |
@wiking | create one :) | 11:26 |
mikeling | micmn: we don''t have that fixture for now :) | 11:27 |
micmn | ok thx :) | 11:27 |
micmn | wiking, mikeling: sorry to bother you again | 12:21 |
micmn | https://github.com/MikeLing/shogun/blob/41f2ca4995d517c0729363e0f854bddb03af997c/tests/unit/classifier/svm/SVMOcas_unittest.cc#L40 | 12:21 |
micmn | aren't the data shared by all tests? | 12:21 |
@wiking | yeah | 12:22 |
@wiking | that should not be there | 12:22 |
@wiking | :) | 12:22 |
mikeling | micmn: yes, I notice that is not necessary | 12:22 |
@wiking | thnx for the review | 12:22 |
@wiking | :))) | 12:22 |
@wiking | feel free to fix | 12:22 |
micmn | in fact I think that SG_REF is needed | 12:22 |
micmn | otherwise the machine destructor will unref those | 12:23 |
mikeling | wiking: btw, I'm still not understand how sg_ref works | 12:23 |
@wiking | micmn, yep | 12:23 |
-!- geektoni_ [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has quit [Quit: Page closed] | 12:23 | |
@wiking | micmn, but that should be somewhere within the creation | 12:23 |
@wiking | of those objects | 12:23 |
@wiking | i mean fixtures | 12:23 |
micmn | maybe in get_features_train() and similar? | 12:24 |
@wiking | yes | 12:24 |
@wiking | that is an adequate place | 12:24 |
@wiking | although | 12:24 |
@wiking | you only want to have that once ++ | 12:24 |
@wiking | not every time you get | 12:24 |
@wiking | as in case | 12:24 |
@wiking | CSVMOcas* ocas = new CSVMOcas(1.0, train_feats, ground_truth); | 12:24 |
@wiking | you'll have a ++ on ref anyways | 12:24 |
@wiking | micmn, see what i mean? | 12:24 |
micmn | ah right | 12:24 |
micmn | so in the environment setup | 12:25 |
@wiking | yes | 12:25 |
@wiking | that'd be the best | 12:25 |
@wiking | HeikoS, ping? | 12:27 |
@HeikoS | wiking: ponmg | 12:27 |
@wiking | HeikoS,there were couple of things... you should check some of the discussions we had with sergey and fernando yesterday | 12:28 |
@wiking | once you are alie | 12:28 |
@wiking | *alive | 12:28 |
@HeikoS | yeah I kind of am | 12:28 |
@HeikoS | what were they about so I can find them? | 12:28 |
@wiking | HeikoS, just check the last rant on #gsoc | 12:28 |
@HeikoS | in slack? | 12:29 |
@wiking | yep | 12:29 |
@wiking | lovely -rwxr-xr-x 1 root admin 103M Jun 22 15:40 /usr/local/lib/python2.7/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so | 12:31 |
@wiking | :) | 12:31 |
@HeikoS | wiking: you gotta summaries things for me, I am not following the gsoc channel :D | 12:36 |
@wiking | ro:) | 12:36 |
@wiking | life is a shit | 12:37 |
@wiking | :) | 12:37 |
@wiking | that's in summary | 12:37 |
@HeikoS | i see | 12:40 |
@HeikoS | thanks for the update :D | 12:40 |
@wiking | no worries :) | 12:41 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3856 opened by micmn | 12:57 |
-!- iglesiasg [~iglesiasg@217.119.234.214] has joined #shogun | 13:05 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 13:05 | |
-!- iglesiasg [~iglesiasg@217.119.234.214] has quit [Client Quit] | 13:06 | |
-!- iglesiasg [~iglesiasg@217.119.234.214] has joined #shogun | 13:06 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 13:06 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 240 seconds] | 13:27 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3751 synchronized by micmn | 13:34 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3856 merged by vigsterkr | 13:44 |
@sukey | [https://github.com/shogun-toolbox/shogun] vigsterkr pushed 2 commits: | 13:44 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/479a93625f7845626d470d9a2fe6412b9392b3d2 | 13:44 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/b6657907cd5aaefa3af47bd4aefa4be320b76bd0 | 13:44 |
@wiking | ok | 13:45 |
@wiking | merged good lets see the other one | 13:45 |
@wiking | micmn, why are those ignores in the macines? | 13:46 |
@wiking | https://github.com/shogun-toolbox/shogun/pull/3751/files#diff-856555dd8e1fe665a77a0707b995f058R80 | 13:46 |
micmn | wiking: machines that fail for various reasons, mostly because they need some inizialization | 13:55 |
@wiking | micmn, ok so those are the ones to be fixed | 13:58 |
@wiking | :) | 13:58 |
@wiking | right? | 13:58 |
@wiking | micmn, what was your fix on the end for the benchmarks cript? | 14:00 |
@wiking | micmn, just because currently mlpack people are running benchmark over shogun | 14:01 |
@wiking | and getting a lot of timeout | 14:01 |
@wiking | i reckon it's the same reason :) | 14:01 |
@wiking | http://masterblaster.mlpack.org/job/benchmark%20-%20shogun/47/consoleFull | 14:02 |
micmn | yeah | 14:05 |
@wiking | so yeah we need a stupid fix | 14:05 |
@wiking | which is stupid | 14:05 |
@wiking | because the only way to fix it | 14:05 |
@wiking | is to disable parallel processing | 14:05 |
@wiking | :D | 14:05 |
micmn | https://gist.github.com/micmn/2b2d61f08b8e303d8d7767f3cddcab3f | 14:06 |
micmn | all shogun's methods should be run in a different process | 14:06 |
@wiking | oh you mean no shoung should be run | 14:07 |
@wiking | prior the fork | 14:07 |
micmn | yeah | 14:07 |
micmn | previously accuracy ecc where computed in the main process | 14:07 |
micmn | *were | 14:07 |
micmn | https://gist.github.com/micmn/2b2d61f08b8e303d8d7767f3cddcab3f#file-lda-py-L134 | 14:08 |
micmn | but this has to be done in every script | 14:08 |
@wiking | micmn, ok i've copy pasted this to the mlpack people | 14:08 |
@wiking | :) | 14:09 |
@wiking | hope they have the resource for it | 14:09 |
@wiking | :) | 14:09 |
@wiking | micmn, ok so unit tests are passing on travis | 14:15 |
@wiking | micmn, since there's no jinja2 on appveyor i should be merging this | 14:16 |
@wiking | only request i have | 14:16 |
@wiking | actually two | 14:16 |
@wiking | b) can we close this https://github.com/shogun-toolbox/shogun/pull/3750 | 14:16 |
@wiking | ? | 14:16 |
micmn | we'll add regression etc later? | 14:16 |
micmn | yeah if we merge the one I updated today | 14:17 |
@wiking | a) i mean | 14:17 |
@wiking | ok fuck | 14:17 |
@wiking | so i mean | 14:17 |
@wiking | there has been quite a lot of comment | 14:17 |
@wiking | by Heiko there | 14:17 |
@wiking | so to use ctags | 14:17 |
@wiking | so let's think about that then | 14:18 |
@wiking | + yeah we can add the other type of machines | 14:18 |
@wiking | in some other PRs | 14:18 |
@wiking | only thing apart from this | 14:18 |
@wiking | there should be an issue for every machine that has problems with serialization | 14:19 |
micmn | ok | 14:19 |
micmn | anyway most of the ignored machines fail because they need some custom inizialization for which the defaiult 'ctor isn't enough | 14:21 |
micmn | *default | 14:21 |
micmn | not directly by serialization problems (...we don't know) | 14:22 |
@wiking | k | 14:23 |
@wiking | let's then address Heiko's comments | 14:23 |
@wiking | in the PR you've updated | 14:23 |
@wiking | and close the other one | 14:23 |
@wiking | and let's start merging | 14:23 |
@wiking | i was trying to get the fucking msvc story fixed | 14:23 |
@wiking | but still couldn't find a good way about the heap story | 14:23 |
@wiking | but yeah those PRs needs rebasing as well | 14:24 |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has joined #shogun | 14:42 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3711 synchronized by MikeLing | 14:43 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3711 synchronized by MikeLing | 14:51 |
mikeling | wiking: micmn The label name has been fixed in https://github.com/shogun-toolbox/shogun/pull/3711/files#diff-78679d0fd2c2058165e04034b113e7dcR23 :) | 14:52 |
mikeling | sorry for the duplicate push... | 14:53 |
@wiking | mmm | 14:53 |
@wiking | coll | 14:53 |
@wiking | thnx | 14:53 |
@wiking | added some comments | 14:55 |
mikeling | wiking: Thank you. | 14:56 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3711 synchronized by MikeLing | 15:01 |
@wiking | mikeling, fix the license info | 15:04 |
@wiking | but no need for more test like this only convert other unit test that uses some custom data generation | 15:05 |
@wiking | if possible | 15:05 |
@wiking | as otherwise the test that micmn is working on | 15:05 |
@wiking | is actually gonna cover this | 15:05 |
mikeling | wiking: copy and past this part of comment https://github.com/shogun-toolbox/shogun/pull/3812/files#diff-9739eb22131152d61edc04fe3812921bR2 as license, right? | 15:10 |
mikeling | like this data generator https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/classifier/svm/LibLinear_unittest.cc#L23, do we need to create a new fixture for it or just make it use the GaussianCheckerboard fixture we have | 15:12 |
micmn | wiking: should I rebase the open PRs now? it will clutter travis... | 15:17 |
@wiking | yes indee | 15:17 |
@wiking | d | 15:17 |
micmn | yes = I rebase? :p | 15:18 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3848 synchronized by geektoni | 15:19 |
geektoni | wiking: I've upgraded #3848 and I've also tested it inside a Shogun docker container, but I wasn't able to reproduce Travis error :(( | 15:23 |
@wiking | geektoni, well | 15:26 |
@wiking | geektoni, allowed to fail? :) | 15:26 |
@wiking | does gtest has such flag? | 15:27 |
@wiking | micmn, fuck travis :) | 15:27 |
@wiking | let's kill it | 15:27 |
geektoni | mmh I could disable the tests for now | 15:27 |
lisitsyn | travis is gone already, wiking | 15:28 |
geektoni | and investigate on this problem later on. | 15:28 |
lisitsyn | :P | 15:28 |
@wiking | lisitsyn, :D kallaaaakaalaaaaaaaa kaaaaaaaaaaaaaaaalaaaa | 15:28 |
@wiking | NYICKI | 15:29 |
@wiking | konjiec | 15:29 |
@wiking | travis | 15:29 |
@wiking | :D | 15:29 |
@wiking | geektoni, SG_PAUSE_COMP | 15:34 |
@wiking | ;) | 15:34 |
@wiking | nico | 15:34 |
@wiking | i mean | 15:34 |
@wiking | nice-o machine | 15:34 |
@wiking | :) | 15:34 |
@wiking | lets see | 15:34 |
@wiking | worst case u can disable | 15:34 |
@wiking | if it still fails on travisO | 15:34 |
geektoni | wiking: mmh it will take a loong time before my PR will get tested by Travis :P | 15:36 |
geektoni | the queue is a bit crowded ;) | 15:36 |
@wiking | mmm | 15:37 |
@wiking | lalal | 15:37 |
@wiking | what should i kill? | 15:37 |
@wiking | :) | 15:37 |
@wiking | lisitsyn, pingu | 15:37 |
lisitsyn | wiking: pingo | 15:38 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 15:38 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 15:38 | |
@wiking | lisitsyn, read and comment | 15:38 |
@wiking | plz | 15:38 |
@wiking | tired of this lalal | 15:38 |
lisitsyn | kill travis? | 15:38 |
lisitsyn | but he is already fired | 15:38 |
lisitsyn | ok ok ok | 15:39 |
lisitsyn | really? kill travis? | 15:39 |
@wiking | killed | 15:39 |
@wiking | geektoni, ok so only one before your stuff :) | 15:39 |
geektoni | LOL | 15:40 |
geektoni | thanks | 15:40 |
lisitsyn | you killed travis? :) | 15:40 |
@wiking | we need a fucking way to speed up tests | 15:40 |
@wiking | this is getting again annoying | 15:40 |
@wiking | should we add ccache to the docker img? | 15:40 |
@wiking | :) | 15:40 |
lisitsyn | make tests annoying again? | 15:40 |
@wiking | lisitsyn, well i mean travis is super slow | 15:40 |
@wiking | takes again a long time to get things tests | 15:41 |
@wiking | *tested | 15:41 |
@wiking | what we could do is to add ccache to the docker img | 15:41 |
lisitsyn | how does that help | 15:41 |
@wiking | well | 15:41 |
@wiking | we init it :) | 15:41 |
lisitsyn | hahaha | 15:41 |
@wiking | and that's how we push it to the docker repo | 15:42 |
@wiking | :) | 15:42 |
lisitsyn | it would get useless in a month | 15:42 |
@wiking | and then after every release | 15:42 |
@wiking | we update it | 15:42 |
@wiking | yeah | 15:42 |
@wiking | although it would still be better as it is now | 15:42 |
@wiking | i mean | 15:42 |
@wiking | ok | 15:42 |
@wiking | then question | 15:42 |
@wiking | anybody KNOWS how the fuck we could have something similar | 15:42 |
@wiking | like an EBS? :) | 15:42 |
@wiking | on aws | 15:42 |
@wiking | for travis | 15:42 |
@wiking | we do that with the buildbot builders | 15:42 |
lisitsyn | s3fs | 15:42 |
@wiking | lisitsyn, speed? | 15:42 |
lisitsyn | I didn't benchmark | 15:43 |
lisitsyn | shouldn't be that slow | 15:43 |
lisitsyn | ok idea | 15:43 |
@wiking | ok so that could work | 15:43 |
lisitsyn | yes | 15:43 |
@wiking | or just have | 15:43 |
@wiking | a tar.xz | 15:43 |
lisitsyn | mount a volume using fuse that is s3fs | 15:43 |
lisitsyn | caches are append-only anyway | 15:44 |
@wiking | append-only? :D | 15:44 |
@wiking | if you cannot read | 15:44 |
@wiking | what's the point | 15:44 |
@wiking | :D | 15:44 |
@wiking | (ok i'm tired) | 15:44 |
lisitsyn | I mean you just add a file that's MD5 | 15:44 |
lisitsyn | haha ok I was ambiguous | 15:45 |
lisitsyn | I mean there is no issue with parallel write | 15:45 |
lisitsyn | to the same file | 15:45 |
lisitsyn | but there could be some crazy waits if we go parallel | 15:45 |
lisitsyn | due to posix stuff | 15:45 |
@wiking | hehe | 15:45 |
@wiking | yeah i mean this is shitty actually | 15:45 |
@wiking | with those 10 jobs | 15:45 |
@wiking | on travis | 15:46 |
@wiking | mikeling, yes exactly like these needs to be replaced https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/classifier/svm/LibLinear_unittest.cc#L23 | 16:17 |
@wiking | no i mean | 16:17 |
@wiking | the only thing we want | 16:18 |
@wiking | that if we can use the fixture | 16:18 |
@wiking | then let's use that | 16:18 |
@wiking | so this is like a 2 gaussians each being dim = 2 | 16:18 |
@wiking | with 50 samples | 16:18 |
mikeling | yes, so we don't need new fixture right? Just replace the generate_gaussians with the fixture we have right now | 16:19 |
mikeling | got it | 16:19 |
@wiking | yeah | 16:19 |
@wiking | but you need to take care of some details | 16:19 |
@wiking | right? | 16:19 |
mikeling | yes | 16:19 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/classifier/svm/LibLinear_unittest.cc#L49 | 16:19 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 276 seconds] | 16:32 | |
-!- iglesiasg [~iglesiasg@217.119.234.214] has quit [Quit: leaving] | 16:41 | |
geektoni | wiking: https://travis-ci.org/shogun-toolbox/shogun/jobs/245784419#L1983 | 16:51 |
@wiking | dead? | 16:51 |
geektoni | all interfaces now | 16:52 |
geektoni | LOL | 16:52 |
geektoni | apparently | 16:52 |
@wiking | :) | 16:52 |
geektoni | SWIG generate wrong code when dealing with rxcpp::subscriber | 16:52 |
@wiking | geektoni, make_subscriber | 16:52 |
@wiking | geektoni, you can disable that code section from | 16:52 |
@wiking | swig | 16:52 |
@wiking | do you need to expose it to SWIG interfaces | 16:52 |
@wiking | ? | 16:52 |
@wiking | i dont think so | 16:53 |
geektoni | nope | 16:53 |
@wiking | no the actual handlers | 16:53 |
geektoni | mmh, I don't think | 16:53 |
geektoni | so | 16:53 |
@wiking | there's a macro | 16:56 |
@wiking | #ifndef SWIG | 16:57 |
geektoni | yeah I've already seen them around the code | 16:58 |
geektoni | but | 16:58 |
geektoni | I'm not quite sure where is the code that causes SWIG to fail | 16:58 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 16:58 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:58 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3857 opened by MikeLing | 17:00 |
geektoni | wiking: at least gcc now works lol https://travis-ci.org/shogun-toolbox/shogun/jobs/245784417 | 17:01 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3848 synchronized by geektoni | 17:25 |
geektoni | wiking: ^ interfaces should be okay now | 17:26 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3711 synchronized by MikeLing | 17:36 |
@HeikoS | wiking, mikeling btw it is dangerous to test libsvm with gaussians as just by chance, there will be errors. | 17:37 |
@HeikoS | wiking, mikeling we have some datasets with linearly separable data as well that could be used instead | 17:37 |
@HeikoS | wiking: in general, I think we should somehow distinguish between unit tests and tests for statistical correctness (e.g. in a random sense) | 17:37 |
mikeling | HeikoS: you mean we make unit test depend on dataset in data/ folder ? | 17:38 |
@wiking | no | 17:38 |
@HeikoS | mikeling: thats an option, though I know it is not optimal | 17:38 |
@wiking | we do not want to use | 17:38 |
@wiking | any file | 17:38 |
@wiking | from data | 17:38 |
@HeikoS | just saying: | 17:38 |
@wiking | in a unit tests | 17:38 |
@HeikoS | if you use gaussian blobs | 17:38 |
@wiking | that's not a unit test | 17:39 |
@HeikoS | then your test will sporadically fail | 17:39 |
@HeikoS | unless you fix the seed | 17:39 |
-!- geektoni [c1cdd252@gateway/web/freenode/ip.193.205.210.82] has quit [Ping timeout: 260 seconds] | 17:39 | |
@HeikoS | which is implicitly fixing a dataset, which is also not good | 17:39 |
@HeikoS | so having something that generates linearly separable data, and then checking 100% accuracy is a better idea than gaussians | 17:39 |
@HeikoS | gaussians with compact support for example | 17:39 |
mikeling | HeikoS: ok, maybe CMath::ini_random(10) for now? | 17:39 |
mikeling | * init_random(10) | 17:40 |
@HeikoS | mikeling: thats not really nice either | 17:41 |
@HeikoS | what data are you using? | 17:41 |
@HeikoS | or fixture? | 17:41 |
@HeikoS | what does it generate? | 17:41 |
@HeikoS | mikeling: ? | 17:42 |
@HeikoS | mikeling: you know how to restrict the support of Gaussians? This way you could ensure that they dont overlap, i.e. your linear classifier can perfectly distinguish the classes | 17:43 |
mikeling | HeikoS: mmm, no, i don't know how to restrict the Gaussians | 17:43 |
@HeikoS | mikeling: what class/code are you using? | 17:43 |
mikeling | HeikoS: in https://github.com/shogun-toolbox/shogun/blob/develop/tests/unit/environments/GaussianCheckerboard.h | 17:44 |
@HeikoS | mikeling: uuuh, that code is very hard to read | 17:45 |
mikeling | HeikoS: mmm, which part? | 17:45 |
@HeikoS | wiking: shouldnt have been merged | 17:46 |
@HeikoS | for (index_t i = 0, j = 0; i < data.num_cols; ++i) | 17:46 |
@HeikoS | test_idx[j++] = i; | 17:46 |
@HeikoS | 17:46 | |
@HeikoS | 17:46 | |
@HeikoS | labels[i / 2] = (i < data.num_cols / 2) ? 1.0 : -1.0; | 17:46 |
@HeikoS | i mean are you kidding me? | 17:46 |
@wiking | ? | 17:47 |
@HeikoS | thats horrible | 17:47 |
@HeikoS | to read code | 17:47 |
@HeikoS | all this can be done nicely and readable | 17:47 |
@wiking | sure | 17:47 |
mikeling | HeikoS: basically, I'm doing the same thing in https://github.com/shogun-toolbox/shogun/pull/3812/files#diff-78679d0fd2c2058165e04034b113e7dcL20 | 17:47 |
@wiking | the PR was open for more than 2 weeks | 17:47 |
@HeikoS | mikeling: sure I have seen things like this elsewhere | 17:48 |
@HeikoS | doesnt change the fact that one can also write this nicely | 17:48 |
@HeikoS | ok anyways | 17:49 |
@HeikoS | as for the gaussians | 17:49 |
@wiking | any suggestion for improvement is more than welcome | 17:49 |
@wiking | before or after merge | 17:49 |
@HeikoS | mikeling: you have the PR somewhere? | 17:49 |
@HeikoS | I can comment there | 17:50 |
@wiking | this is not breaking yet any CI | 17:50 |
mikeling | sure, just a second | 17:50 |
@wiking | apart from aestetics | 17:50 |
@HeikoS | hard to maintain, puts people off | 17:50 |
mikeling | HeikoS: here is the pr https://github.com/shogun-toolbox/shogun/pull/3812 | 17:50 |
@wiking | sure | 17:50 |
@wiking | as said | 17:50 |
@HeikoS | will hardly b touched | 17:50 |
@wiking | HeikoS, it was open for a long time | 17:50 |
@wiking | at least 2-3 weeks | 17:51 |
@HeikoS | I have read that | 17:51 |
@wiking | so i mean | 17:51 |
@HeikoS | I know I was gone for a bit, but pls next time ping me in the PR and I will comment | 17:51 |
@HeikoS | you started the whole thing about having multiple dev +1 before merging, remember? | 17:52 |
@wiking | HeikoS, look man | 17:52 |
@wiking | i'm fine doing this here | 17:52 |
@wiking | you were not around | 17:52 |
@wiking | for the last couple of weeks | 17:52 |
@wiking | i'm trying to maintain 3 students input | 17:52 |
@wiking | where one of them is struggling with some very basic things | 17:53 |
@wiking | apart from this i'm trying to do my job | 17:53 |
@wiking | if you think you could have done this better | 17:53 |
@wiking | i'm happy to hear about your input | 17:53 |
@wiking | how i could have done better things | 17:53 |
@wiking | but sorry | 17:53 |
@HeikoS | I will comment | 17:53 |
@wiking | none of the merged PRs | 17:53 |
@wiking | have broken | 17:53 |
@wiking | ANY | 17:53 |
@wiking | CI | 17:53 |
@wiking | whereas my comment of | 17:53 |
@wiking | [23:49] <HeikoS> you started the whole thing about having multiple dev +1 before merging, remember? | 17:54 |
@wiking | was due to the fact | 17:54 |
@wiking | that there was | 17:54 |
@wiking | NUMEROUS cases | 17:54 |
@wiking | where PRs were merged | 17:54 |
@wiking | that broke CIs | 17:54 |
@wiking | but again | 17:54 |
@wiking | if this thing is hurting anybody's aeastetics | 17:54 |
@wiking | mea culpa | 17:54 |
@wiking | this is all i could manage | 17:54 |
@wiking | i can stop it | 17:55 |
lisitsyn | hey | 18:03 |
lisitsyn | I'd vote for merging working stuff as well | 18:03 |
mikeling | sorry....I should do it better | 18:03 |
lisitsyn | polishing is easier | 18:03 |
@HeikoS | (and never happened in the past) | 18:04 |
lisitsyn | what never happened? | 18:13 |
lisitsyn | I think it has happened quite a few times before | 18:13 |
@HeikoS | mikeling: dont worry, I sent a list of things that could be improved in the PR | 18:41 |
mikeling | HeikoS: Thank you! I will address it right away | 18:42 |
@HeikoS | mikeling: I think some of the code I complained about was not written by you so I should blame the person who wrote it instead (which is likely to be myself :D ) | 18:45 |
@HeikoS | wiking: ^ | 18:45 |
@wiking | dunno who wrote the ocas test | 18:45 |
@wiking | i guess a blame could tell | 18:46 |
@HeikoS | I think this stuff comes form the gaussian blobs | 18:46 |
@HeikoS | somehow got copy pasted in there | 18:46 |
@HeikoS | and slightly modified | 18:46 |
@HeikoS | from a messy start | 18:46 |
@HeikoS | but it doesnt matter | 18:46 |
@HeikoS | I put some suggestions in | 18:46 |
@HeikoS | wiking: what are your thoughts about separating unit tests from correctness tests? | 18:47 |
@wiking | micmn, i dont know if i already wrote this or not.... but lemme be sure that i did: so would be great if you could fire an issue against the benchamrks of mlpack about how to fix the shogun openmp problem... just so that this doesn't disappear into the void | 18:48 |
@wiking | thnx heaps | 18:48 |
@HeikoS | wiking: sorry for the waterfall, but what happened to the std::vector thing in the end? | 18:48 |
@wiking | i reported it to zoq (main developer of benchmarks) but just in case having an issue that marks the things would be great | 18:48 |
@wiking | HeikoS, ^ this is a fucking shit btw | 18:49 |
@HeikoS | whats going on there? | 18:49 |
@wiking | openmp + gnu fork | 18:49 |
@wiking | it's deadlock | 18:49 |
@wiking | there's a workaround | 18:49 |
@wiking | but | 18:49 |
@wiking | still | 18:49 |
@wiking | this is a shit | 18:49 |
@HeikoS | ah | 18:49 |
@wiking | so if you for example wanna use shogun | 18:49 |
@wiking | and then fork some processes | 18:50 |
@wiking | and use some shogun more | 18:50 |
@wiking | you are fucked | 18:50 |
@wiking | because the openmp will get into some hanging state | 18:50 |
micmn | wiking: sure, will do | 18:50 |
@wiking | and basically this is what they do in benchmarks of mlpack | 18:50 |
@wiking | micmn, thnx heaps | 18:50 |
@HeikoS | ha | 18:50 |
@HeikoS | is that reported? | 18:50 |
@wiking | and now they ran | 18:50 |
@HeikoS | as in "shogun crashes" ? | 18:50 |
@wiking | against new shogun | 18:51 |
@HeikoS | in their papers? :D | 18:51 |
@HeikoS | cute | 18:51 |
@wiking | and tests timeouts | 18:51 |
@wiking | like a lot | 18:51 |
@wiking | becuase actually the openmp code | 18:51 |
@wiking | is happening | 18:51 |
@wiking | in eigen level | 18:51 |
@wiking | :) | 18:51 |
@wiking | mostly | 18:51 |
@wiking | so we cannot even control anything | 18:51 |
@wiking | ... | 18:51 |
@wiking | anyhow i told it to zoq so it'll get handled | 18:51 |
@wiking | and then they rerun stuff | 18:51 |
@wiking | bt this we should mark somewhere | 18:52 |
@wiking | because it's a major issue | 18:52 |
@wiking | we again dont have a FAQ | 18:52 |
@wiking | right? | 18:52 |
@wiking | coz this would be worth to be mentioned anywhere there | 18:52 |
@wiking | if you fork how to fork | 18:52 |
@wiking | as gnu fork with gnu openmp | 18:52 |
@wiking | will fuck you | 18:52 |
@wiking | HeikoS, doesn't happen with icc | 18:53 |
@wiking | of course... | 18:53 |
@HeikoS | ok | 18:53 |
@wiking | micmn, and the other reason why it'd be cool because you had all the logs of bugs etc... but i saw that you've logged thos in your daily | 18:53 |
@wiking | so that's really great | 18:53 |
@HeikoS | yeah an issue is really important | 18:53 |
@wiking | we can use that later for the faq or something | 18:53 |
@HeikoS | both in shogun and in zoq as well | 18:53 |
@wiking | HeikoS, zoq said he'll port the fix | 18:54 |
@wiking | during th weekend | 18:54 |
@wiking | is | 18:54 |
@wiking | *ish | 18:54 |
@wiking | coz basically this is all linalg :) | 18:54 |
@wiking | so yeah this should be somewhere easily found info | 18:54 |
@wiking | and we cannot do anything about this | 18:54 |
@wiking | i.e. not an issue | 18:55 |
@wiking | this is a feature of gnu openm | 18:55 |
@wiking | *openmp | 18:55 |
@wiking | i'm wondering if this shit comes up with clang's openmp | 18:55 |
@wiking | does it use the same gomp? | 18:55 |
@HeikoS | no idea | 18:55 |
@wiking | i guess no because libiomp5 | 18:56 |
@wiking | buuuut | 18:56 |
@wiking | would be great to figure these shit out... | 18:56 |
@wiking | HeikoS, mmmm correctness... | 18:57 |
@wiking | HeikoS, you mean similar to serializable? | 18:57 |
@wiking | i mean in one way and the ohter | 18:58 |
@HeikoS | wiking: sorry what do you mean? | 18:58 |
@wiking | train -> serialize | 18:58 |
@HeikoS | ah | 18:58 |
@wiking | test | 18:58 |
@HeikoS | yeah I mean | 18:58 |
@wiking | would be similar | 18:58 |
@wiking | than the 'correctness' what you mean | 18:58 |
@HeikoS | unit tests are kind of tests for modules working fine | 18:58 |
@wiking | here | 18:58 |
@wiking | right? | 18:58 |
@HeikoS | indepdendent of random data | 18:58 |
@HeikoS | like a+a = 2*a | 18:58 |
@HeikoS | but we have lots of test that are like | 18:58 |
@HeikoS | "linearly separates linearly separable data" | 18:58 |
@HeikoS | which seems a bit different to me | 18:59 |
@HeikoS | serialization is more unit test like | 18:59 |
@wiking | micmn, is just refurbishing this https://github.com/shogun-toolbox/shogun/pull/3751 | 18:59 |
@HeikoS | but different because it is auto generated and covers all classes | 18:59 |
@wiking | so i mean | 19:00 |
@wiking | it is in a way similar | 19:00 |
@wiking | but of course | 19:00 |
@wiking | there any fucking numb data would do it | 19:00 |
@wiking | for the purpose of testing | 19:00 |
@wiking | serialization | 19:00 |
@wiking | micmn, so there we do the refactor still of the python scripto? | 19:00 |
@wiking | :) | 19:00 |
micmn | yeah, you mean Heiko's comments on ctags? | 19:01 |
@wiking | yes | 19:02 |
@wiking | cool | 19:02 |
@wiking | ok i'm gonna check wtf is there | 19:02 |
@wiking | with the https://github.com/shogun-toolbox/shogun/pull/3843 | 19:02 |
@wiking | as for some reason | 19:02 |
@wiking | travis hiccuped | 19:02 |
@wiking | aaah | 19:03 |
@wiking | i know | 19:03 |
@wiking | stupid formatting | 19:03 |
@wiking | :D | 19:03 |
micmn | of course | 19:03 |
* wiking thiking about removing either msvc 14 or 15 from the appveyor... as 1 pr test takes 1 hour | 19:04 | |
@wiking | any objections^ | 19:04 |
@wiking | and ther's no ccache for msvc | 19:05 |
@wiking | lisitsyn, fucking appveyor has cache dir support :) | 19:05 |
@wiking | lisitsyn, Caches lets Travis CI store directories between builds, which is useful for storing dependencies that take longer to compile or download. | 19:06 |
@wiking | argh | 19:06 |
@HeikoS | wiking: no objections from my side | 19:06 |
@wiking | fuck | 19:07 |
@wiking | https://docs.travis-ci.com/user/caching/#ccache-cache | 19:07 |
@wiking | since when ?:) | 19:07 |
@wiking | lisitsyn, ^ :D | 19:07 |
@wiking | what do you think? :) | 19:07 |
@wiking | kill with fire? :D | 19:07 |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 246 seconds] | 19:10 | |
@sukey | [https://github.com/shogun-toolbox/shogun] New branch travis-ccache created | 19:18 |
@sukey | [https://github.com/shogun-toolbox/shogun] New commit https://github.com/shogun-toolbox/shogun/commit/9029307e64c47c3b32573c884782d19ecc31b0e9 by vigsterkr | 19:18 |
lisitsyn | wiking: oh | 19:18 |
lisitsyn | that's good | 19:18 |
@wiking | lets see | 19:19 |
@wiking | if it works | 19:19 |
@wiking | i guess i'll know it in the morning | 19:22 |
@wiking | as i'm off | 19:22 |
@wiking | gnite all | 19:22 |
rcurtin | HeikoS: wiking: I don't think we've published any benchmarks with these failures on our site, I think it has been a long time since any new shogun benchmarks have been done | 19:48 |
rcurtin | definitely we wouldn't want to publish incorrect benchmarks, so I think marcus will update using the patch we got today and then we can re-run | 19:49 |
rcurtin | the benchmarking system has not received the love it should have :( we are trying to change that... | 19:51 |
-!- travis-ci [~travis-ci@ec2-23-22-171-32.compute-1.amazonaws.com] has joined #shogun | 21:07 | |
travis-ci | it's Viktor Gal's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/245879078 | 21:07 |
-!- travis-ci [~travis-ci@ec2-23-22-171-32.compute-1.amazonaws.com] has left #shogun [] | 21:07 | |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-oewvncwnqrkuhecv] has quit [Quit: Connection closed for inactivity] | 22:06 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 22:08 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:09 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 260 seconds] | 22:42 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 22:52 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:52 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Client Quit] | 22:56 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has joined #shogun | 23:17 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 23:17 | |
-!- HeikoS [~heiko@host-92-0-178-129.as43234.net] has quit [Ping timeout: 240 seconds] | 23:26 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3826 synchronized by micmn | 23:46 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/3842 synchronized by micmn | 23:46 |
--- Log closed Fri Jun 23 00:00:25 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!