--- Log opened Thu May 26 00:00:12 2016 | ||
-!- thoralf [~thoralf@ip5b4189cf.dynamic.kabel-deutschland.de] has quit [Quit: Konversation terminated!] | 00:20 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 260 seconds] | 02:26 | |
shogun-buildbot | build #1007 of nightly_none is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_none/builds/1007 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 03:18 |
---|---|---|
shogun-buildbot | build #5 of clang - thread analysis is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20thread%20analysis/builds/5 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 03:44 |
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has quit [Quit: Page closed] | 03:44 | |
shogun-buildbot | build #4 of clang - undefined behaviour analysis is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/clang%20-%20undefined%20behaviour%20analysis/builds/4 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 03:48 |
-!- sonne|osx_ [~sonne@x4e33d889.dyn.telefonica.de] has joined #shogun | 05:56 | |
-!- sonne|osx [~sonne@x4db48e9e.dyn.telefonica.de] has quit [Ping timeout: 260 seconds] | 05:57 | |
-!- sonne|osx_ is now known as sonne|osx | 05:57 | |
shogun-buildbot | build #1 of FC23 - libshogun - aarch64 is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/1 blamelist: Viktor Gal <viktor.gal@maeth.com> | 05:58 |
-!- Saurabh7 [~Saurabh7@59.93.219.168] has joined #shogun | 06:18 | |
Saurabh7 | wiking, lisitsyn , Is it okay if I make methods like SGMatrix<T>::clone() const ? | 06:23 |
-!- sanuj [~sanuj@61.0.30.9] has joined #shogun | 06:47 | |
Saurabh7 | also things like SGMatrix::equals(other) | 06:57 |
shogun-buildbot | build #4 of memleak - valgrind is complete: Failure [failed memory check] Build details are at http://buildbot.shogun-toolbox.org/builders/memleak%20-%20valgrind/builds/4 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 07:18 |
@wiking | shogun-buildbot: force build --branch=develop 'FC23 - libshogun - aarch64' | 07:23 |
shogun-buildbot | build #2 forced | 07:23 |
shogun-buildbot | I'll give a shout when the build finishes | 07:23 |
@wiking | Saurabh7: you can make anything const | 07:56 |
@wiking | that does not change the content of the class | 07:56 |
@wiking | but anyways the compiler will error | 07:56 |
@wiking | when a const function is doing something that it's not allowed | 07:57 |
Saurabh7 | ok | 07:57 |
Saurabh7 | but densefeature->get_feature_matrix() | 07:57 |
Saurabh7 | if i want to make this return type const | 07:57 |
Saurabh7 | so that the copy doesnt change anything | 07:57 |
Saurabh7 | how would i do that | 07:58 |
Saurabh7 | since its returning value | 07:58 |
Saurabh7 | wiking, const SGMatrix<ST> CDenseFeatures<ST>::get_feature_matrix() doesnt help | 07:58 |
shogun-buildbot | build #2 of FC23 - libshogun - aarch64 is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/2 | 08:02 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 08:12 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 08:12 | |
@wiking | shogun-buildbot: force build --branch=develop 'FC23 - libshogun - aarch64' | 08:13 |
shogun-buildbot | build #3 forced | 08:13 |
shogun-buildbot | I'll give a shout when the build finishes | 08:13 |
@wiking | Saurabh7: what do you mean by const SGMatrix<ST> doesn't help? | 08:14 |
Saurabh7 | wiking, I change values of elements in the returned matrix | 08:14 |
Saurabh7 | of the copy | 08:15 |
Saurabh7 | and the original is also changed | 08:15 |
@wiking | Saurabh7: ah you mean that you can still change the values in the matrix | 08:15 |
Saurabh7 | yes looks like it | 08:15 |
Saurabh7 | any way arnd i am missing ? | 08:17 |
Saurabh7 | wiking, one more thing | 08:22 |
Saurabh7 | for train CMachine->train(features) onyl features are passed | 08:22 |
Saurabh7 | but labels are members of Machine | 08:23 |
Saurabh7 | how to train in parallel then? | 08:23 |
Saurabh7 | set_labels() isnt suitable for parallel train on different subsets | 08:24 |
shogun-buildbot | build #1136 of nightly_default is complete: Failure [failed test notebooks] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1136 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 08:47 |
shogun-buildbot | build #3 of FC23 - libshogun - aarch64 is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/3 | 08:51 |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 09:18 | |
-!- mode/#shogun [+o besser82] by ChanServ | 09:18 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 09:19 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 09:19 | |
@wiking | shogun-buildbot: force build --branch=develop 'FC23 - libshogun - aarch64' | 09:19 |
shogun-buildbot | build #4 forced | 09:19 |
shogun-buildbot | I'll give a shout when the build finishes | 09:19 |
-!- Saurabh7 [~Saurabh7@59.93.219.168] has quit [Remote host closed the connection] | 09:22 | |
-!- sanuj [~sanuj@61.0.30.9] has quit [Ping timeout: 272 seconds] | 10:02 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 276 seconds] | 10:10 | |
-!- sanuj [~sanuj@61.0.30.9] has joined #shogun | 10:41 | |
sanuj | lisitsyn, there? | 11:15 |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has joined #shogun | 11:34 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:34 | |
sanuj | HeikoS, ! | 11:39 |
@HeikoS | sanuj: jojo | 11:39 |
sanuj | travis still fails on my PR | 11:39 |
@HeikoS | lets see | 11:40 |
@HeikoS | I added few more things to the cookbook | 11:40 |
@HeikoS | we now have static calls | 11:40 |
@HeikoS | where is your PR? | 11:40 |
sanuj | in java, still int turns into double | 11:40 |
sanuj | HeikoS, https://github.com/shogun-toolbox/shogun/pull/3197 | 11:40 |
@HeikoS | not possible | 11:40 |
@HeikoS | rebase your fork and maybe resend the PR | 11:40 |
@HeikoS | check travis output if my patch | 11:41 |
@HeikoS | it is correct in there | 11:41 |
@HeikoS | must be some other error | 11:41 |
@HeikoS | github was a bit funny yesterday | 11:41 |
@HeikoS | might be that | 11:41 |
sanuj | okay | 11:41 |
sanuj | i'll rebase again | 11:41 |
@HeikoS | uh | 11:41 |
@HeikoS | we have another one of those int issues | 11:41 |
@HeikoS | https://travis-ci.org/shogun-toolbox/shogun/jobs/132996284 | 11:41 |
shogun-buildbot | build #4 of FC23 - libshogun - aarch64 is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/4 | 11:42 |
-!- c4goldsw [82e15d88@gateway/web/cgi-irc/kiwiirc.com/ip.130.225.93.136] has joined #shogun | 11:45 | |
@HeikoS | weiiiird | 11:46 |
@HeikoS | sanuj: ah no, that is something else, caused by "wrong" example code | 11:47 |
@HeikoS | so all should be good | 11:47 |
shogun-buildbot | build #14 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/14 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 11:51 |
sanuj | okay | 11:52 |
shogun-buildbot | build #2876 of bsd1 - libshogun is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2876 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 11:52 |
@HeikoS | sanuj: whats the stand up for today? | 11:55 |
@HeikoS | arianepaola: whats the stand up for today? | 11:55 |
@HeikoS | Saurabh7-: ^ | 11:55 |
c4goldsw | ^ | 11:56 |
c4goldsw | Oops, typo | 11:57 |
@HeikoS | :) | 11:57 |
@HeikoS | sanuj: wow | 12:03 |
@HeikoS | same on my machine | 12:03 |
@HeikoS | sorry about dismissing it | 12:03 |
@HeikoS | investigating | 12:03 |
@HeikoS | sanuj: pretty simple solution ;) | 12:04 |
@HeikoS | see my comment | 12:04 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 12:09 | |
-!- Saurabh7 [~Saurabh7@1.39.15.64] has joined #shogun | 12:10 | |
Saurabh7 | HeikoS, hi | 12:12 |
@HeikoS | Saurabh7 hi | 12:12 |
@HeikoS | just checking on daily progress | 12:13 |
Saurabh7 | HeikoS, why do we pass only features to machine->train() | 12:13 |
@HeikoS | what else would you pass? | 12:13 |
Saurabh7 | because | 12:13 |
Saurabh7 | labels | 12:13 |
@HeikoS | labels? | 12:13 |
@HeikoS | ah yeah | 12:13 |
@HeikoS | I know | 12:13 |
@HeikoS | bad design I guess | 12:13 |
Saurabh7 | beacuse when running in parallel | 12:13 |
Saurabh7 | labels are members | 12:13 |
Saurabh7 | so cant do same thing like am doing with features | 12:14 |
@HeikoS | internally, it is the same as setting them via setter | 12:14 |
Saurabh7 | but its set outisde of train() | 12:14 |
@HeikoS | not sure I get it | 12:14 |
@HeikoS | if you do things in parallel | 12:14 |
Saurabh7 | how do i say it | 12:14 |
@HeikoS | you need multiple CMachine instances, no? | 12:14 |
@HeikoS | that operator on the same data | 12:14 |
@HeikoS | otherwise the states get confused | 12:15 |
Saurabh7 | so how do i do it in cross validation | 12:15 |
@HeikoS | so you just set labels and features | 12:15 |
@HeikoS | that are different for each fold | 12:15 |
@HeikoS | and the machine is also different | 12:15 |
@HeikoS | but they all share data | 12:15 |
@HeikoS | i.e. raw data and raw labels | 12:15 |
@HeikoS | but have different objects with different subsets on them | 12:15 |
@HeikoS | and different machines with different internal states (like solutions) | 12:16 |
@HeikoS | does that help? | 12:16 |
Saurabh7 | I thought the machien is same | 12:16 |
Saurabh7 | sut different subsets on labels and features | 12:16 |
@HeikoS | Saurabh7 how would that work? | 12:16 |
@HeikoS | but labels and features are member of the machine | 12:16 |
@HeikoS | how can they be different? | 12:16 |
@HeikoS | for one machine | 12:16 |
@HeikoS | Saurabh7 think of linear regression | 12:18 |
@HeikoS | training the machine means solving a linear system | 12:18 |
@HeikoS | and storing the result in a variable, i.e. w and b | 12:19 |
Saurabh7 | yeah doesnt make sense | 12:19 |
Saurabh7 | ok let me look into it more | 12:19 |
Saurabh7 | thanks | 12:19 |
@HeikoS | this w and b will be different for every fold right? | 12:19 |
@HeikoS | so if you do things one after another, you can use the same | 12:19 |
@HeikoS | but if you do multiple machines, there will be many w and b | 12:19 |
@HeikoS | so you need to instantiate shallow copies of everything needed for training inside the thread | 12:19 |
@HeikoS | and set the particular subset on features on labels, | 12:20 |
Saurabh7 | ok | 12:20 |
@HeikoS | Saurabh7 shall we write some pseudo-code? | 12:20 |
@HeikoS | would that help you? | 12:20 |
@HeikoS | Saurabh7 maybe draft the x-validation in a gist, not too many details | 12:20 |
@HeikoS | and then we can discuss in there | 12:21 |
@HeikoS | makes sense? | 12:21 |
Saurabh7 | HeikoS, ok will do | 12:21 |
@HeikoS | Saurabh7 I am around for a while, so lets sort this out now :) | 12:23 |
c4goldsw | HeikoS: Hey, I'm trying to get Shogun working using the Octave interface, but I'm getting build errors when running CMake (http I've installed liboctave-dev as suggested in the installation guide, and I have octave 4.0.0 installed. I don't know much about cmake, so the errors don't mean a lot to me either. Any suggestions of what could be wrong? | 12:25 |
@HeikoS | c4goldsw: can you paste a gist of the error? | 12:26 |
c4goldsw | Here are the errors I'm getting: https://gist.github.com/c4goldsw/01bb55f80b87fb24a4fc412ee52075b1 | 12:26 |
@HeikoS | c4goldsw: can you paste the full thing? | 12:26 |
@HeikoS | your octave seems not detected | 12:26 |
c4goldsw | Sure | 12:26 |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun | 12:27 | |
c4goldsw | I updated the same gist. | 12:27 |
shogun-buildbot | build #5 of FC23 - libshogun - aarch64 is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/5 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com> | 12:29 |
@HeikoS | octave-config | 12:30 |
@HeikoS | if you type that into your terminal, what happens? | 12:31 |
@HeikoS | octave-config -v | 12:31 |
c4goldsw | It isn't installed | 12:31 |
@HeikoS | this is the problem then | 12:31 |
@HeikoS | your octave installation | 12:31 |
@HeikoS | it should return | 12:31 |
@HeikoS | something like | 12:31 |
c4goldsw | I'll do that right now and tell you if it can install or not afterwards. | 12:31 |
@HeikoS | 3.8.1 | 12:31 |
@HeikoS | which OS? | 12:31 |
c4goldsw | Ubuntu 14.04 LTS | 12:32 |
@HeikoS | c4goldsw: thats easy then | 12:32 |
@HeikoS | check this | 12:32 |
@HeikoS | https://github.com/shogun-toolbox/shogun/blob/develop/configs/Dockerfile | 12:32 |
Saurabh7 | HeikoS, https://gist.github.com/Saurabh7/6efb54e9fb0834f4d65bccf6a4426546 : how its working right now | 12:32 |
@HeikoS | its the config we run on travis | 12:32 |
c4goldsw | Great, everything works. | 12:32 |
c4goldsw | apt-get install is a beautiful thing (after coming from Windows) | 12:32 |
@HeikoS | yep :) | 12:34 |
@HeikoS | let me know how things go | 12:34 |
@HeikoS | Saurabh7 I commented | 12:34 |
Saurabh7 | HeikoS, yeah i had missed the machine part | 12:36 |
Saurabh7 | was trying with jsut features and labels phew | 12:36 |
@HeikoS | Saurabh7 takes a while to get the head around shared memory parallelisation :) | 12:38 |
Saurabh7 | HeikoS, yeah now its sounds really stupid :) | 12:38 |
Saurabh7 | HeikoS, also how do i const SGMatrix<ST> CDenseFeatures<ST>::get_feature_matrix() make this safe for copies | 12:39 |
sanuj | HeikoS, silly me | 12:39 |
sanuj | didn't see that | 12:39 |
@HeikoS | sanuj: :D no worries | 12:39 |
sanuj | fixed and pushed | 12:39 |
Saurabh7 | sry this : SGMatrix<ST> CDenseFeatures<ST>::get_feature_matrix() | 12:39 |
@HeikoS | Saurabh7 I am thinking we might want to introduce a class of read-only features | 12:39 |
@HeikoS | but for a first prototype | 12:39 |
@HeikoS | you can just define the matrix const | 12:39 |
sanuj | today i'll add some stuff in SGObject for tags and parameters | 12:40 |
@HeikoS | but I think the CDenseFeatures wont accept that right? | 12:40 |
sanuj | like unorderedmap | 12:40 |
sanuj | etc | 12:40 |
sanuj | lisitsyn, there? | 12:40 |
Saurabh7 | HeikoS, const SGMatrix<ST> CDenseFeatures<ST>::get_feature_matrix() | 12:40 |
Saurabh7 | i tried this but still can change the matrix | 12:40 |
@HeikoS | Saurabh7 yeah , but you wont be able to pass that | 12:40 |
@HeikoS | I see | 12:40 |
@HeikoS | dont worry for now | 12:40 |
@HeikoS | we will sort this out at some point | 12:40 |
@HeikoS | sanuj: ^ | 12:40 |
@HeikoS | also might be interesting for you | 12:41 |
@HeikoS | wiking: ah man we need to clean up the const usage in Shogun | 12:41 |
@HeikoS | very important for multithread | 12:41 |
c4goldsw | HeikoS: I've ran into some issues whilst building shogun with the octave interface enabled: https://gist.github.com/c4goldsw/01bb55f80b87fb24a4fc412ee52075b1 | 12:54 |
@wiking | HeikoS: here? | 12:54 |
c4goldsw | Wait, I just updated it again. | 12:55 |
@HeikoS | wiking: yes | 12:55 |
@wiking | ok sorry | 12:55 |
@HeikoS | c4goldsw: that is an issue related to hdf5, maybe just disable it | 12:55 |
@wiking | yeah tha tthing needs to be sorted out | 12:55 |
@wiking | but | 12:55 |
@wiking | first of all | 12:55 |
@HeikoS | -DUSE_HDF5=Off | 12:55 |
@HeikoS | c4goldsw: ^ | 12:55 |
@wiking | what the fuck is BAHSIC doing? | 12:55 |
@HeikoS | in what sense? | 12:55 |
c4goldsw | I'll try that, thanks. | 12:55 |
@wiking | because i dont know why but the unit test fails on arm64 | 12:56 |
@wiking | for it | 12:56 |
@HeikoS | I see | 12:56 |
@HeikoS | well not sure what it is doing on there | 12:56 |
@wiking | sanuj: when you write EXPECT_EQ | 12:56 |
@wiking | the first arg is the expected | 12:56 |
@wiking | the second arg is the actual | 12:56 |
@wiking | not the other way around | 12:56 |
@HeikoS | why do unit tests fail on arm64 if they dont fail on others? | 12:56 |
@wiking | HeikoS: it's called portability | 12:57 |
@wiking | :) | 12:57 |
@HeikoS | wiking: yeah | 12:57 |
@HeikoS | it is doing something non portable | 12:57 |
@wiking | i'm trying to figure out | 12:57 |
@wiking | why and how | 12:57 |
@HeikoS | disable the unit test for now | 12:57 |
@wiking | no | 12:57 |
@HeikoS | we are refactoring that anyways | 12:57 |
@wiking | that's not a fucking solution | 12:57 |
@wiking | :) | 12:57 |
sanuj | wiking, in gtest? | 12:57 |
@wiking | sanuj: yes | 12:57 |
@HeikoS | heavily working on the code atm | 12:57 |
sanuj | okay | 12:57 |
@wiking | HeikoS: same for Statistics? | 12:57 |
@HeikoS | wiking: it will be changed anyways | 12:57 |
@HeikoS | wiking: no | 12:57 |
@wiking | log_det_general | 12:57 |
@HeikoS | ah | 12:57 |
@wiking | because that fails as well | 12:58 |
@HeikoS | that uses eigen | 12:58 |
@HeikoS | but ok that we have t ofix | 12:58 |
@HeikoS | can you give me a link | 12:58 |
@wiking | http://buildbot.shogun-toolbox.org/builders/FC23%20-%20libshogun%20-%20aarch64/builds/5/steps/test/logs/stdio | 12:58 |
@HeikoS | hsic can be disabled | 12:58 |
@wiking | there you can see | 12:58 |
@wiking | that only for one matrix | 12:58 |
@wiking | there's a problem with logdet general | 12:58 |
@HeikoS | kk | 12:58 |
@HeikoS | heavy eigen3 | 12:59 |
@HeikoS | ah I think I know whats wrong | 12:59 |
@HeikoS | will fix later | 12:59 |
@HeikoS | with HSIC, as said, just disable | 13:00 |
@wiking | HeikoS: can you tell me? | 13:00 |
@HeikoS | will do | 13:00 |
@wiking | because i can fix it then | 13:00 |
@HeikoS | wiking: ah ok | 13:00 |
@HeikoS | check line 652 | 13:00 |
@wiking | i mean yep | 13:00 |
@wiking | float64_t result=CMath::INFTY | 13:00 |
@HeikoS | the if clause is not entered | 13:00 |
@wiking | i think it's entered | 13:00 |
@wiking | that's the problem | 13:00 |
@wiking | check_u==detP | 13:01 |
@wiking | is true | 13:01 |
@HeikoS | oh ok | 13:01 |
@wiking | hence it has a value | 13:01 |
@HeikoS | did you valgrind? | 13:01 |
@wiking | result | 13:01 |
@HeikoS | no memory error? | 13:01 |
@HeikoS | maybe eigen behaves different on that arch? | 13:01 |
@HeikoS | ok need to check more then | 13:01 |
@HeikoS | debug | 13:01 |
@HeikoS | the code itself doenst do anything fancy | 13:02 |
@HeikoS | the eigen calls do most work | 13:02 |
@wiking | Hebtw | 13:02 |
@wiking | there are leaks | 13:02 |
@wiking | in sgobj | 13:02 |
@wiking | or rather | 13:02 |
@wiking | CKLInference | 13:02 |
@HeikoS | wiking: should tell Wu | 13:02 |
@HeikoS | we need to tell him to work on feature branchs anyway | 13:03 |
@HeikoS | too many merges | 13:03 |
@HeikoS | no description, no discussion | 13:03 |
@HeikoS | if he leaves, that code is f***ed since noone can fix it | 13:03 |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 13:04 | |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 240 seconds] | 13:06 | |
CaBa | hi guys | 13:06 |
@HeikoS | CaBa: hi | 13:07 |
-!- Saurabh7_ [~Saurabh7@1.39.86.39] has joined #shogun | 13:07 | |
CaBa | back from beautiful helsinki :) | 13:07 |
arianepaola | hello everyone :-) | 13:10 |
-!- Saurabh7 [~Saurabh7@1.39.15.64] has quit [Ping timeout: 246 seconds] | 13:11 | |
@HeikoS | arianepaola: hihi | 13:11 |
-!- Saurabh7_ is now known as Saurabh7 | 13:11 | |
@HeikoS | arianepaola: was just asking about daily status :) all running well? | 13:11 |
@HeikoS | CaBa: nice, what happened there? | 13:13 |
CaBa | HeikoS: BREW, small student-only conference | 13:15 |
sanuj | how are we sorting out cookbooks this week | 13:17 |
sanuj | who is writing which | 13:17 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 13:18 | |
arianepaola | Hi HeikoS | 13:18 |
@HeikoS | sanuj: good point to snyc with others | 13:18 |
@HeikoS | sanuj: next week, put them in the mail you wirte | 13:19 |
sanuj | HeikoS, look at this https://travis-ci.org/shogun-toolbox/shogun/builds/133075635 | 13:19 |
sanuj | can we merge? | 13:19 |
sanuj | what about this week | 13:19 |
arianepaola | I am working on the unit tests | 13:19 |
sanuj | arianepaola, hello :) | 13:20 |
arianepaola | one strange thing that I notice is that after using ubuntu 16.04 things stopped working | 13:20 |
arianepaola | hi sanuj, how are you? | 13:20 |
sanuj | i'm good! | 13:20 |
sanuj | i wanted to ask | 13:21 |
arianepaola | :-) | 13:21 |
arianepaola | me too | 13:21 |
sanuj | which cookbook you are writing this week | 13:21 |
@HeikoS | arianepaola: let us know if something blocks you | 13:21 |
@HeikoS | arianepaola: which unit tests btw? | 13:21 |
arianepaola | lda & flda tests | 13:22 |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 276 seconds] | 13:22 | |
@HeikoS | arianepaola: I see | 13:22 |
@HeikoS | arianepaola: it is good to spend some time on those | 13:22 |
@HeikoS | arianepaola: but please focus on the gsoc project | 13:22 |
@HeikoS | so if you can do something that helps with release AND your gsoc projects, that is much better than just fixing unit tests | 13:22 |
CaBa | HeikoS: you don't happen to have relatives in gütersloh? | 13:22 |
@HeikoS | time for that was more last week, this weel we want to get progress in installation issues | 13:23 |
@HeikoS | wiking: agreed^ | 13:23 |
@HeikoS | ? | 13:23 |
sanuj | lisitsyn, there? | 13:23 |
arianepaola | HeikoS: I have talked with wiking and the tests are for the debian packages to build | 13:23 |
@HeikoS | arianepaola: I see | 13:24 |
@HeikoS | arianepaola: great, then it is helping for both :) | 13:24 |
@HeikoS | so full power ahead thne | 13:24 |
arianepaola | ^^ hey ho let's go :-) | 13:24 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Remote host closed the connection] | 13:28 | |
@wiking | just a sec | 13:30 |
@wiking | arianepaola: what do you mean about 16.04 stopped working? | 13:34 |
-!- Saurabh7 [~Saurabh7@1.39.86.39] has quit [Ping timeout: 260 seconds] | 13:35 | |
arianepaola | getting some build errors, that I did not get on 14.04 | 13:35 |
@wiking | build errors? | 13:36 |
@wiking | with the latest develop? | 13:36 |
@wiking | because:http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/14 | 13:36 |
@wiking | so it compiles on 16.04 | 13:36 |
arianepaola | I have to recreate the vm, seems to be some left overs from the daily images that I used before 16.04 release | 13:40 |
@wiking | where's wu? | 13:40 |
@wiking | arianepaola: i can give you a packer.io template | 13:40 |
@wiking | that can generate a docker or a vm image for u if u like | 13:41 |
arianepaola | if you want to share the packer.io template | 13:41 |
arianepaola | I am using vagrant with docker | 13:42 |
@wiking | HeikoS: Wu keeps using none stdint variables (like int) | 13:42 |
@wiking | arianepaola: lemme send it to you | 13:42 |
arianepaola | ok | 13:42 |
arianepaola | thanks | 13:42 |
@HeikoS | wiking: wu never is in irc, have to send him emails (cc me as well) | 13:43 |
-!- c4goldsw [82e15d88@gateway/web/cgi-irc/kiwiirc.com/ip.130.225.93.136] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 13:44 | |
@wiking | HeikoS: generated_cpp-linear_svm | 13:46 |
@wiking | leaks like hell | 13:46 |
@HeikoS | wiking: interesting :) | 13:46 |
@wiking | http://buildbot.shogun-toolbox.org/memcheck/20160526-0202.html | 13:46 |
@HeikoS | I guess thats not unreferencing | 13:46 |
@HeikoS | every object is instantiated using lisitsyn 's some | 13:47 |
@HeikoS | ill trace it locally | 13:48 |
lisitsyn | щщзы | 13:48 |
lisitsyn | oops | 13:48 |
lisitsyn | :D | 13:48 |
@HeikoS | wiking: while it should be fixed, no big problem | 13:48 |
@HeikoS | as this is definitely caused by the cpp translation | 13:48 |
arianepaola | lisitsyn: schschzy? google translate seems not to work on that one | 13:49 |
lisitsyn | arianepaola: wrong keyboard layout | 13:50 |
lisitsyn | have to switch them | 13:50 |
lisitsyn | sometimes forget to :) | 13:50 |
arianepaola | haha | 13:50 |
@wiking | HeikoS: yeah all generated stuff leaks | 13:51 |
@HeikoS | wiking: all cpp ? | 13:51 |
@HeikoS | the others shouldnt I guess | 13:51 |
lisitsyn | easy to fix | 13:51 |
lisitsyn | just do it! | 13:51 |
lisitsyn | :D | 13:51 |
@wiking | we have a lot of leaks again | 13:52 |
@wiking | :) | 13:52 |
@HeikoS | lisitsyn: none of the some instantiations are deleted | 13:52 |
@HeikoS | any ideas? | 13:52 |
@HeikoS | its all the new operators that valgrind lists | 13:52 |
lisitsyn | interviewing someone right now, will check in a few mins | 13:52 |
@HeikoS | hahahaha | 13:53 |
@HeikoS | ask him | 13:53 |
lisitsyn | her! | 13:53 |
@wiking | ask her | 13:55 |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 13:55 | |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 13:56 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 14:02 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 14:02 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 14:09 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 14:09 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 14:18 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 14:18 | |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [] | 14:18 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 14:26 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has left #shogun [] | 14:26 | |
-!- arianepaola [~ariane@unaffiliated/arianepaola] has joined #shogun | 14:36 | |
-!- Saurabh7 [~Saurabh7@59.93.219.168] has joined #shogun | 14:51 | |
sanuj | lisitsyn, there? | 15:47 |
lisitsyn | sanuj: sup | 15:59 |
sanuj | yo | 15:59 |
sanuj | tags in aer have id also | 15:59 |
sanuj | lisitsyn, do we keep it? | 16:00 |
lisitsyn | yeah probably it makes sense | 16:00 |
sanuj | i didn't have id in my implementation | 16:00 |
sanuj | what is it for | 16:00 |
lisitsyn | sanuj: you can generate unique ids | 16:00 |
sanuj | okay | 16:00 |
sanuj | cool | 16:00 |
lisitsyn | to make them hashable fast | 16:00 |
lisitsyn | hash = id | 16:00 |
sanuj | i see | 16:00 |
sanuj | and what are d pointers | 16:00 |
lisitsyn | sanuj: something like that | 16:01 |
lisitsyn | https://wiki.qt.io/D-Pointer | 16:01 |
sanuj | and i don't understand how unique_ptr is diff from normal c++ pointer | 16:01 |
sanuj | except that only one unique pointer can uniquely own an object | 16:01 |
@HeikoS | wiking: | 16:04 |
@HeikoS | party! | 16:04 |
@HeikoS | sonne just signed the final doc | 16:04 |
@HeikoS | now we have done everything | 16:04 |
@HeikoS | just have t owait | 16:04 |
@HeikoS | and then we are eV | 16:04 |
sanuj | eV? | 16:05 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:05 | |
@HeikoS | sanuj: https://en.wikipedia.org/wiki/Eingetragener_Verein | 16:05 |
lisitsyn | sanuj: yeah then please read about that :P | 16:05 |
sanuj | lisitsyn, yeah :P | 16:05 |
@wiking | HeikoS: cool | 16:07 |
@wiking | HeikoS: wait for papers? | 16:07 |
@HeikoS | wiking: yes | 16:07 |
@HeikoS | but all we should get now is the letter of recognition | 16:08 |
@HeikoS | and a bill :) | 16:08 |
@wiking | ah ok | 16:08 |
@wiking | cool cool | 16:08 |
@wiking | after that | 16:08 |
@wiking | we should transfer the ownership | 16:08 |
@HeikoS | yep | 16:08 |
@wiking | of fatbot and hedonismbot to ev | 16:08 |
@HeikoS | exactly | 16:08 |
@HeikoS | and have a "donate" button | 16:08 |
@wiking | (that is sonne|osx should do that :P) | 16:08 |
@wiking | yep | 16:08 |
@wiking | and i think we can apply to some VAT exemption on aws as well | 16:09 |
@wiking | afaik | 16:09 |
@HeikoS | yep | 16:09 |
-!- OXPHOS [9d8b131c@gateway/web/freenode/ip.157.139.19.28] has joined #shogun | 16:16 | |
OXPHOS | HeikoS hey thanks for the comments | 16:19 |
@HeikoS | OXPHOS: hi there | 16:19 |
@HeikoS | thanks for the cookbooks! | 16:19 |
OXPHOS | HeikoS not sure what the linalg should be like..so I tried to find sth to do :P | 16:20 |
@HeikoS | what do you mean? | 16:21 |
OXPHOS | lambday wiking: Hi so what do you think the linalg should be like? a.dot(b) or dot(a, b)? | 16:21 |
@HeikoS | I see | 16:21 |
OXPHOS | HeikoS ^ | 16:21 |
@HeikoS | OXPHOS: well there is cereal | 16:21 |
@HeikoS | linalg discussion should just go on the side | 16:21 |
@HeikoS | keep in mind you want to spend most of your time coding | 16:22 |
@HeikoS | OXPHOS: I think we agreed on the dot(a,b) right? | 16:22 |
OXPHOS | Kinda need wiking to review my current cereal as well... | 16:23 |
OXPHOS | nope I think yesterday wiking has different idea. Too much branches with current dot(a, b) design | 16:23 |
OXPHOS | *many | 16:23 |
OXPHOS | whatever | 16:23 |
@HeikoS | I see | 16:23 |
@HeikoS | haha yeah indeed | 16:24 |
@HeikoS | where is your cereal? | 16:24 |
OXPHOS | HeikoS: https://github.com/shogun-toolbox/shogun/pull/3202 | 16:24 |
@HeikoS | well he added comments | 16:25 |
@HeikoS | if you resolve this, it can be merged | 16:25 |
@HeikoS | OXPHOS: this is just the cmake, is there any progress with the prototype? | 16:26 |
OXPHOS | really..it's mainly formatting issue. I'd like to talk about the realization. It's in the comment | 16:26 |
OXPHOS | I thought anyway the code won't run even if I submit the pr.. | 16:26 |
@HeikoS | OXPHOS: but this PR is just about adding a dependency | 16:27 |
@HeikoS | so the discussion in there should be about that | 16:27 |
@HeikoS | focussing on getting it merged | 16:27 |
@HeikoS | other discussions should happen elsehwere | 16:27 |
@HeikoS | you can always send "fake" PRs with non working code (just say so in comment) | 16:27 |
@HeikoS | and discuss there | 16:27 |
@HeikoS | this one can be merged once the formatting is done | 16:28 |
@HeikoS | so go ahead with that, we merge, you send another one and we can discuss there | 16:28 |
@HeikoS | keeping things focussed | 16:28 |
OXPHOS | okay then. but you have any comments on what i have now? ;) the ugly for-loop save. I'm no it now. | 16:28 |
@HeikoS | you mean element wise loading-saving? | 16:29 |
OXPHOS | yes | 16:29 |
@HeikoS | thats what we do in our code as well | 16:29 |
@HeikoS | I think should be possible to avoid this | 16:30 |
@HeikoS | but as a first go, it might be ok | 16:30 |
@HeikoS | can optimize later, it is more important to get a prototype going asap for now | 16:30 |
OXPHOS | sure | 16:31 |
@HeikoS | so I think should be ok for now | 16:31 |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 16:31 | |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Remote host closed the connection] | 16:33 | |
shogun-buildbot | build #258 of deb1 - libshogun - PR is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/258 blamelist: OXPHOS | 16:36 |
shogun-buildbot | build #259 of deb1 - libshogun - PR is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/259 blamelist: OXPHOS | 16:38 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:38 | |
lambday | OXPHOS: hey! | 16:43 |
lambday | wiking: ping | 16:43 |
@wiking | ok sorry | 16:43 |
@wiking | was in a meeting | 16:43 |
@wiking | what'sup | 16:43 |
@wiking | ? | 16:43 |
@wiking | (reading logs) | 16:43 |
lambday | wiking: remember the discussion we had yesterday about branching in the linalg methods | 16:43 |
lambday | if-else too much? | 16:43 |
@wiking | ok first of all | 16:43 |
@wiking | we should benchmark that branching | 16:44 |
@wiking | second of all | 16:44 |
@wiking | where's the cereal review stuff OXPHOS | 16:44 |
OXPHOS | Heyhey I decreased the error of current dot(a, b) from over 20 to 3 yesterday ... so still a little bit work to make that work. But I can try a.dot(b) if you decide to go that way | 16:44 |
@wiking | CEREAL | 16:44 |
@wiking | :) | 16:44 |
@wiking | where's that | 16:45 |
OXPHOS | wiking: just a little patch [in] https://github.com/shogun-toolbox/shogun/pull/3202 | 16:45 |
OXPHOS | comment for now | 16:45 |
lambday | OXPHOS: nono.. just go ahead with that we have discussed | 16:45 |
OXPHOS | lambday okay | 16:45 |
lambday | OXPHOS: let me know if you need help with anything | 16:46 |
@wiking | OXPHOS: there you go | 16:47 |
@wiking | OXPHOS: if you have any questions lemme know | 16:47 |
@wiking | and lets continue with the stuff | 16:47 |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has joined #shogun | 16:47 | |
@wiking | because serialization is straightforward | 16:47 |
OXPHOS | lambday: thx should be fine. just that complicated *static_cast<GPU_Vector<T>*>(a) need a little bit more care | 16:47 |
@wiking | linalg needs more and more discussion | 16:47 |
OXPHOS | wiking: got it! | 16:48 |
@wiking | OXPHOS: cool... ping me here when you have those changes | 16:48 |
OXPHOS | wiking: sure! | 16:48 |
@wiking | it should actually make the travis work | 16:48 |
lambday | OXPHOS: yeah.. remember to use const Vector& everywhere in the linalg methods | 16:48 |
@wiking | because it should automagically download cereal | 16:49 |
OXPHOS | wiking: makes more sense | 16:49 |
OXPHOS | lambday: got it | 16:49 |
-!- sanuj [~sanuj@61.0.30.9] has quit [Ping timeout: 272 seconds] | 16:52 | |
-!- sanuj [~sanuj@61.0.30.9] has joined #shogun | 16:53 | |
@wiking | arianepaola: just sent you the generator | 16:54 |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:56 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:56 | |
@wiking | GSoC-ers | 16:56 |
@wiking | i did not see a standup update from anybody todya | 16:56 |
@wiking | :))) | 16:56 |
@wiking | lemme see the logs | 16:56 |
c4goldsw | Hey HeikoS, you're in London, right? | 16:57 |
@HeikoS | yes | 16:57 |
@HeikoS | why? | 16:57 |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has joined #shogun | 16:57 | |
travis-ci | it's Heiko Strathmann'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/133115126 | 16:57 |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has left #shogun [] | 16:57 | |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 16:58 | |
@wiking | leagoetz_: i need standup updates :))) | 16:58 |
@wiking | time to ring the magical alarm bell f | 16:58 |
@wiking | :P | 16:58 |
c4goldsw | I'm visiting a friend there for the weekend, who lives in the SOAS / UCL area - would you have 30 minutes or so to grab coffee? | 16:58 |
@HeikoS | haha ;) | 16:58 |
@wiking | and gsocers | 16:59 |
leagoetz_ | Ok, ok, will ring | 16:59 |
@wiking | please check this regularly http://buildbot.shogun-toolbox.org/memcheck | 16:59 |
@wiking | we are leaking quite extensively | 16:59 |
@HeikoS | arianepaola, sanuj, Saurabh7, OXPHOS ^ | 17:00 |
@wiking | i would really appreciate if you would send in prs to stop leaks | 17:00 |
-!- sanuj [~sanuj@61.0.30.9] has quit [Ping timeout: 276 seconds] | 17:00 | |
leagoetz_ | wiking: just saw your email - looks like you've done the ringing already :0 | 17:00 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 240 seconds] | 17:00 | |
leagoetz_ | arianepaola: sanuj: Saurabh7: OXPHOS: please put your updates in IRC | 17:01 |
leagoetz_ | thanks! | 17:01 |
@wiking | leagoetz_: heheheheh yeah the impatient part of me | 17:01 |
OXPHOS | log - 0525 done: 2 cookbook pr; linalg working prototype. 0526 todo: finish cookbook pr; finish cereal cmake; hopefully cereal prototype and linalg prototype | 17:02 |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Remote host closed the connection] | 17:03 | |
@wiking | thnx | 17:04 |
sonne|work | HeikoS: happy after all? | 17:08 |
@HeikoS | 0525, OXPHOS? :) | 17:08 |
@HeikoS | sonne|work: very happy, you should be as well | 17:08 |
@HeikoS | as you are now responsible for our finances ;D | 17:08 |
OXPHOS | HeikoS: 26th just started.. | 17:09 |
c4goldsw | HeikoS Exi was wondering if it’s just a mistake or it’s deliberate that the Minimizer class is not inherited from SGObject, hence serialization and other SGObject functionalities are not available… (like sg_ref, sg_unref) | 17:11 |
c4goldsw | cuse me, I should have been more specific - I arrive tomorrow and I'm there for the weekend. | 17:11 |
c4goldsw | ? | 17:11 |
@HeikoS | OXPHOS: haha, good working times ;) | 17:11 |
@HeikoS | c4goldsw: no I think it was left out on purpose | 17:11 |
@HeikoS | but that is kind of before plugin was thought of | 17:12 |
@HeikoS | so I think we should change | 17:12 |
@wiking | ?? | 17:12 |
@wiking | so what? :) | 17:13 |
@wiking | i need an answer :) | 17:13 |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 17:13 | |
c4goldsw | Heiko, that was accidentally copy pasted from someone else's text (I don't know how that ended up there lol). What I meant to say was "Excuse me, I should have been more specific - I arrive tomorrow and I'm there for the weekend" | 17:13 |
@wiking | :))) | 17:14 |
@wiking | c4goldsw: that was my email | 17:14 |
@wiking | HeikoS: if we dont hear about you after tomorrow dont worry we'll call the authorities that c4goldsw has kidnapped you ;) | 17:15 |
@HeikoS | ehm ... | 17:15 |
@wiking | jooooooking :) | 17:15 |
@HeikoS | I am a bit lost | 17:16 |
@HeikoS | sitting in a talk ;) | 17:16 |
@wiking | hahahah | 17:16 |
c4goldsw | I wouldn't want to pay Ryanair anything extra, their baggae policy is too expensive. | 17:16 |
@wiking | it's a derivative of X | 17:16 |
c4goldsw | ;) | 17:16 |
@wiking | c4goldsw: hahahahah | 17:16 |
@wiking | you can eat him :) | 17:16 |
c4goldsw | wiking I'll pass. | 17:17 |
@wiking | :D | 17:17 |
@wiking | is there a backport of valgrind to jessie | 17:17 |
@wiking | (3.11) | 17:17 |
OXPHOS | wiking: may i ask, hot to find URL_MD5? | 17:23 |
OXPHOS | *how | 17:23 |
@wiking | ah you download the tar.gz or whatever archive | 17:23 |
@wiking | and run a md5sum <file> | 17:23 |
OXPHOS | have to download md5sum first :P | 17:25 |
sonne|work | HeikoS: hurray o_O | 17:25 |
sonne|work | HeikoS: it is always difficult for me to have an appointment with $WHOMEVER in the middle of the day ... | 17:25 |
@wiking | OXPHOS: apt-get install | 17:25 |
@wiking | or yum install | 17:25 |
@HeikoS | sonne|work: hopefully we dont need more | 17:26 |
@wiking | OXPHOS: MD5 (v1.1.2.tar.gz) = 34d4ad174acbff005c36d4d10e48cbb9 | 17:27 |
OXPHOS | I only have brew..thx wiking | 17:27 |
@wiking | OXPHOS: ah it's md5 on osx | 17:28 |
@wiking | it's system stuff | 17:28 |
OXPHOS | aha | 17:28 |
sonne|work | HeikoS: luckily I 'only' paid 25 EUR | 17:29 |
@wiking | man | 17:30 |
@wiking | linking the unit test | 17:30 |
@wiking | binary is a fucking pain | 17:30 |
@wiking | ideas? | 17:30 |
@wiking | is there like a caching mechanism for linking? :) | 17:31 |
@wiking | ccache does not do linking :) | 17:32 |
@HeikoS | wiking: agree | 17:32 |
@HeikoS | wiking: I think we should just split unit test per plugin | 17:32 |
@HeikoS | ie one binary per plugin | 17:32 |
@HeikoS | and one binary for core | 17:32 |
@HeikoS | then no constant re-linking when developing code | 17:32 |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has joined #shogun | 17:33 | |
-!- leagoetz [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 260 seconds] | 17:37 | |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has quit [Quit: Leaving.] | 17:38 | |
-!- leagoetz_ [~leagoetz@nat-169-164.internal.eduroam.ucl.ac.uk] has quit [] | 17:42 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] | 17:43 | |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has joined #shogun | 17:43 | |
travis-ci | it's lambday'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/133127613 | 17:43 |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has left #shogun [] | 17:43 | |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has joined #shogun | 17:44 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 17:44 | |
@HeikoS | wiking: think thats good? | 17:45 |
@wiking | it needs a lot of hacking in cmake :P | 17:51 |
@HeikoS | wiking: yeah | 18:01 |
@HeikoS | mmh | 18:01 |
@HeikoS | lets wait for plugins to arrive first then | 18:01 |
@HeikoS | I mean it would be great if one could have a totally local view onto shogun | 18:01 |
@HeikoS | when developing | 18:01 |
@HeikoS | I dont want to be distracted by tests/compile/interface for anything I am not developing | 18:01 |
@HeikoS | if I am e.g. a scientific user that just wants a nice framework for implementing things | 18:02 |
@HeikoS | and getting all the benefits of build/license etc | 18:02 |
@HeikoS | interface I meant | 18:08 |
-!- sonne|work [~sonnenbu@ip5b414dfc.dynamic.kabel-deutschland.de] has quit [Ping timeout: 250 seconds] | 18:09 | |
@wiking | HeikoS: here? | 18:16 |
@HeikoS | wiking: j | 18:16 |
@HeikoS | aaaa | 18:16 |
@wiking | ok so do you know anything about the Inference | 18:16 |
@wiking | and KLInference | 18:16 |
@wiking | hierachy? | 18:16 |
@wiking | because i have here a question | 18:17 |
@wiking | i know why we leak | 18:17 |
@wiking | but i dont know if it's ok what i do | 18:17 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/machine/gp/KLInference.cpp#L152 | 18:17 |
@HeikoS | checking | 18:18 |
@HeikoS | this is no SGOBject? | 18:18 |
@HeikoS | and that causes iut? | 18:18 |
@wiking | noup | 18:19 |
@wiking | so the problem i was wondering | 18:19 |
@wiking | that in one derived class KLDualInferenceMethod | 18:19 |
@wiking | this minimizer is overridden | 18:19 |
@HeikoS | yeah ok | 18:19 |
@HeikoS | and the old thing not deleted? | 18:19 |
@wiking | but without freeing the old one | 18:19 |
@wiking | yes | 18:19 |
@HeikoS | I see | 18:19 |
@wiking | i though to remove the whole | 18:19 |
@HeikoS | maybe add a check if non-null free | 18:19 |
@HeikoS | maybe not | 18:20 |
@wiking | LBFGSMinimizer | 18:20 |
@wiking | but then realised | 18:20 |
@HeikoS | I would just free before over-writing | 18:20 |
@wiking | that in other derived | 18:20 |
@HeikoS | wiking: ok then I suggest that Wu makes everything CSGObject. (Told him not to before, so needs explain) | 18:20 |
@wiking | that is being used | 18:20 |
@wiking | but it's ok | 18:20 |
@wiking | i mean he has a register thing | 18:20 |
@wiking | so i'll just fix it | 18:20 |
@wiking | i was just wondering first if KLInference could be so abstract | 18:20 |
@wiking | that it does not have a default minimizer | 18:20 |
@HeikoS | might be good to ask wu and tell him the fix, ask if that breaks anything else | 18:22 |
@HeikoS | but he has pretty good test coverage as for results | 18:22 |
@wiking | this->register_minimizer | 18:22 |
@wiking | this will take care of things | 18:22 |
@wiking | https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/machine/gp/KLInference.cpp#L346-L356 | 18:22 |
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has joined #shogun | 18:24 | |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has joined #shogun | 18:29 | |
travis-ci | it's Heiko Strathmann'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/133132833 | 18:29 |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has left #shogun [] | 18:29 | |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has joined #shogun | 19:06 | |
travis-ci | it's Heiko Strathmann'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/133151135 | 19:06 |
-!- travis-ci [~travis-ci@ec2-54-211-151-230.compute-1.amazonaws.com] has left #shogun [] | 19:06 | |
rcurtin | I think someone called my name but I can't find it in my scrollback because this channel has so much activity :) | 19:12 |
-!- sonne|work [~sonnenbu@sams-office-nat.tomtomgroup.com] has quit [Ping timeout: 244 seconds] | 19:15 | |
-!- sonne|work [~sonnenbu@88.134.176.135] has joined #shogun | 19:15 | |
-!- Saurabh7 [~Saurabh7@59.93.219.168] has quit [Ping timeout: 252 seconds] | 19:30 | |
@HeikoS | rcurtin: whooooo :) | 19:32 |
rcurtin | :) | 19:37 |
-!- HeikoS [~heiko@nat-170-9.internal.eduroam.ucl.ac.uk] has quit [Ping timeout: 264 seconds] | 19:40 | |
-!- travis-ci [~travis-ci@ec2-54-81-63-29.compute-1.amazonaws.com] has joined #shogun | 19:47 | |
travis-ci | it's Heiko Strathmann'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/133158029 | 19:47 |
-!- travis-ci [~travis-ci@ec2-54-81-63-29.compute-1.amazonaws.com] has left #shogun [] | 19:47 | |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has joined #shogun | 20:23 | |
travis-ci | it's lambday'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/133171846 | 20:23 |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has left #shogun [] | 20:23 | |
OXPHOS | @wiking hey | 21:05 |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has joined #shogun | 21:12 | |
travis-ci | it's lambday'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/133185623 | 21:12 |
-!- travis-ci [~travis-ci@ec2-54-204-198-241.compute-1.amazonaws.com] has left #shogun [] | 21:12 | |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit [Quit: Page closed] | 21:20 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 21:40 | |
-!- mode/#shogun [+o besser82] by ChanServ | 21:40 | |
@wiking | OXPHOS: yes? | 21:46 |
OXPHOS | wiking you're still around! got a stupid problem | 21:47 |
OXPHOS | I did include<external/Cereal> | 21:47 |
OXPHOS | and in Cereal.cmake, I have | 21:48 |
OXPHOS | ExternalProject_Add( Cereal PREFIX ${CMAKE_SOURCE_DIR}/Cereal DOWNLOAD_DIR ${THIRD_PARTY_DIR}/Cereal | 21:48 |
OXPHOS | the download dir gives error: | 21:48 |
OXPHOS | dir '/Cereal' does not exist after file(MAKE_DIRECTORY) | 21:49 |
@wiking | mmm | 21:49 |
OXPHOS | but when I remove 'Cereal', and make it : DOWNLOAD_DIR ${THIRD_PARTY_DIR}/ | 21:49 |
OXPHOS | it works | 21:49 |
@wiking | can give the source of the cmake? | 21:49 |
@wiking | just to see what do you mean | 21:49 |
@wiking | as otherwise this is a bit out of context | 21:50 |
OXPHOS | 1sec | 21:50 |
OXPHOS | wiking: https://gist.github.com/OXPHOS/b7d9d31cd3841605df53effeb528c8e0 | 21:51 |
OXPHOS | If I put the codes in Eigen3.cmake, it can work; If I move Eigen3.cmake to Cereal.cmake, it cannot. Do I need to register sth.? | 21:52 |
@wiking | ok wait | 21:53 |
@wiking | i'm not so sure if i understand | 21:53 |
@wiking | can you show me | 21:53 |
@wiking | how you include this | 21:54 |
@wiking | (detect cereal) | 21:54 |
@wiking | and i would do something like SET(CEREAL_INCLUDE_DIRS SET(EIGEN_INCLUDE_DIR ${THIRD_PARTY_DIR}/include/cereal) | 21:55 |
OXPHOS | updated: https://gist.github.com/OXPHOS/b7d9d31cd3841605df53effeb528c8e0 | 21:55 |
OXPHOS | wiking^ | 21:55 |
OXPHOS | Aha yes I have that | 21:56 |
@wiking | do not add it to SYSTEM_INCLUDES | 21:56 |
@wiking | add it to simple INCLUDES | 21:56 |
@wiking | and use CEREAL_INCLUDE_DIRS not CEREAL_INCLUDE_DIR | 21:56 |
@wiking | (note the prular form | 21:57 |
@wiking | why do you do include<external/Cereal> ? | 21:59 |
@wiking | why not simply | 21:59 |
@wiking | #include <cereal/...> | 21:59 |
OXPHOS | we need to download first right? | 22:00 |
@wiking | yes | 22:06 |
@wiking | https://gist.github.com/OXPHOS/b7d9d31cd3841605df53effeb528c8e0#file-cmakelists-L4 | 22:06 |
@wiking | this will download it | 22:06 |
@wiking | if it's not on the system | 22:06 |
@wiking | but ok | 22:06 |
@wiking | can you update your PR? | 22:07 |
@wiking | with these stuff | 22:08 |
@wiking | it's easier to comment code there | 22:08 |
OXPHOS | sure | 22:09 |
OXPHOS | wiking done | 22:11 |
shogun-buildbot | build #260 of deb1 - libshogun - PR is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/260 blamelist: OXPHOS | 22:12 |
OXPHOS | wiking was messing around to see where went wrong..will change them back | 22:12 |
OXPHOS | *I | 22:12 |
arianepaola | see you all tomorrow | 22:14 |
arianepaola | bye | 22:14 |
@wiking | OXPHOS: ok so i've commented couple of stuff | 22:14 |
OXPHOS | arianepaola bye | 22:15 |
OXPHOS | wiking what do you mean by "check the content"? | 22:15 |
OXPHOS | Currently, with "DOWNLOAD_DIR ${THIRD_PARTY_DIR}/Cereal", I cannot finish cmake | 22:16 |
@wiking | lemme check | 22:16 |
OXPHOS | wiking but with "DOWNLOAD_DIR ${THIRD_PARTY_DIR}/", cereal can be downloaded to root | 22:17 |
@wiking | OXPHOS: what is exactly the problem? | 22:19 |
@wiking | http://pastebin.com/bAwTYRWm | 22:19 |
OXPHOS | wiking: the error is dir '/Cereal' does not exist after file(MAKE_DIRECTORY) | 22:20 |
OXPHOS | it doesn't have the permission to create the folder? | 22:20 |
@wiking | what? | 22:21 |
@wiking | i mean sorry | 22:21 |
@wiking | but i just added this to shogun | 22:21 |
OXPHOS | ?? | 22:21 |
@wiking | and it worked for me | 22:21 |
@wiking | so i really dont knwo what's the problem | 22:21 |
@wiking | please can you pastebin it | 22:21 |
OXPHOS | yes just saw it | 22:21 |
@wiking | but that include dir wont work | 22:21 |
@wiking | gonna have to check some stuff | 22:21 |
OXPHOS | okay so I'll try to fix that first | 22:22 |
@wiking | ok so it needs some more love | 22:22 |
@wiking | see the things in eigen | 22:22 |
OXPHOS | they're exactly the same I would say | 22:23 |
@wiking | no | 22:23 |
@wiking | they are not | 22:23 |
@wiking | :) | 22:23 |
@wiking | please check them and you'll see | 22:23 |
OXPHOS | really..do you mean the args part? | 22:24 |
@wiking | yes | 22:24 |
@wiking | because the thing is that currently cereal is extracted in the wrong place | 22:24 |
@wiking | but that is something else | 22:24 |
@wiking | i think | 22:24 |
@wiking | why did you set | 22:25 |
@wiking | PREFIX ${CMAKE_SOURCE_DIR}/Cereal | 22:25 |
@wiking | ? | 22:25 |
@wiking | CMAKE_BINARY_DIR | 22:25 |
@wiking | should be | 22:25 |
OXPHOS | aa...like I said I was messing around to see what can work | 22:25 |
OXPHOS | for the problem I saw, DOWNLOAD_DIR problem, nothing can save it | 22:26 |
OXPHOS | Cereal.cmake was exactly like Eigen3.cmake in the first place. | 22:27 |
OXPHOS | okay I sudo-ed | 22:27 |
@wiking | OXPHOS: cereal is header only or? | 22:28 |
OXPHOS | wiking yes it is | 22:28 |
@wiking | ok just a sec | 22:28 |
@wiking | mmmm | 22:29 |
@wiking | waaaaaaaaaait | 22:29 |
@wiking | why is this in cereal's cmake | 22:30 |
@wiking | -- Found the following Boost libraries: | 22:30 |
@wiking | -- serialization | 22:30 |
@wiking | -- unit_test_framework | 22:30 |
@wiking | i thought it has 0 dependencies | 22:30 |
@wiking | :S | 22:30 |
OXPHOS | so, if you download the package Cereal, there would be cmakelist, unittests/, etc. Actually, only the include/cereal/ folder has useful stuff | 22:32 |
OXPHOS | this might be the reason wiking | 22:32 |
@wiking | kk | 22:32 |
@wiking | then it's pretty easy | 22:32 |
@wiking | OXPHOS: SET(CEREAL_INCLUDE_DIRS ${CMAKE_BINARY_DIR}/Cereal/src/Cereal/include) | 22:33 |
OXPHOS | aha right | 22:34 |
OXPHOS | -DEIGEN_INCLUDE_INSTALL_DIR:PATH=${THIRD_PARTY_DIR}/include/eigen | 22:34 |
OXPHOS | I don't need this flag right | 22:34 |
@wiking | well in this case no | 22:34 |
@wiking | it should be good like that | 22:34 |
@wiking | and you dont need the compiler flag shit either | 22:34 |
OXPHOS | okay. | 22:35 |
@wiking | OXPHOS: commented on the PR | 22:35 |
OXPHOS | thx! | 22:35 |
@wiking | fix those | 22:35 |
@wiking | and that should work then | 22:35 |
OXPHOS | don't know why I need to sudo the download dir : / but I'll try to bypass it | 22:36 |
OXPHOS | at local for now | 22:36 |
shogun-buildbot | build #261 of deb1 - libshogun - PR is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun%20-%20PR/builds/261 | 23:13 |
@wiking | OXPHOS: make[3]: *** No rule to make target `install'. Stop. | 23:22 |
@wiking | make[2]: *** [Cereal/src/Cereal-stamp/Cereal-install] Error 2 | 23:22 |
@wiking | make[1]: *** [CMakeFiles/Cereal.dir/all] Error 2 | 23:22 |
@wiking | this needs fixing | 23:22 |
@wiking | that means the external cmake script | 23:22 |
OXPHOS | I didn't see anything leads to the 'install' step.. | 23:23 |
@wiking | OXPHOS: check travis | 23:23 |
OXPHOS | I saw but I dont understand why.. | 23:25 |
@wiking | you should disable install step in that external script | 23:30 |
@wiking | INSTALL_COMMAND "" | 23:31 |
OXPHOS | wiking thx lemme try | 23:32 |
OXPHOS | wiking thanks so much. should be fine this time. | 23:39 |
OXPHOS | wiking and if you care to know, I just find out why I couldn't download cereal earlier. The module uses {THIRD_PARTY_DIR} before it is defined..so silly | 23:40 |
--- Log closed Fri May 27 00:00:13 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!