--- Log opened Wed Jan 24 00:00:32 2018 | ||
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4116 opened by guruhegde | 05:39 |
---|---|---|
-!- qiongyue [b9e10c6c@gateway/web/freenode/ip.185.225.12.108] has joined #shogun | 07:54 | |
-!- qiongyue [b9e10c6c@gateway/web/freenode/ip.185.225.12.108] has quit [Quit: Page closed] | 09:25 | |
@wiking | lisitsyn, i just had an idea | 09:35 |
@wiking | so here https://github.com/shogun-toolbox/shogun/issues/4043 | 09:35 |
@wiking | there's the crazy stuff | 09:35 |
@wiking | that we'll change GaussianKernel k(2.0) to | 09:35 |
@wiking | Kernel k = kernel("GaussianKernel") | 09:35 |
@wiking | k.set_width(2.0) | 09:35 |
@wiking | etc etc | 09:36 |
lisitsyn | yes | 09:36 |
@wiking | which is crazy becauce more LoC | 09:36 |
@wiking | but yeah lets say its ok | 09:36 |
@wiking | but | 09:36 |
lisitsyn | ehm? what loc/ | 09:36 |
@wiking | line of code | 09:36 |
@wiking | to do the same thing | 09:36 |
lisitsyn | why? | 09:36 |
@wiking | but nevermind now | 09:36 |
lisitsyn | no | 09:36 |
@wiking | wait | 09:36 |
@wiking | lemme finish | 09:36 |
@wiking | :) | 09:36 |
@wiking | so what about that at least in python | 09:36 |
@wiking | we extend the ctors of objs | 09:36 |
@wiking | with swig | 09:36 |
@wiking | that it accepts kwargs** | 09:36 |
@wiking | and then that's a syntactic sugar | 09:37 |
lisitsyn | yes it is easy | 09:37 |
@wiking | for doing self.tag(key, value) | 09:37 |
lisitsyn | yes | 09:37 |
@wiking | because imo this is crazy | 09:37 |
@wiking | https://github.com/shogun-toolbox/shogun/pull/4115/files#diff-850db6d5082e9cf6bbf0ddb8e5ad2ab3R15 | 09:37 |
lisitsyn | it is not because GaussianKernel(2.0) makes no sense | 09:37 |
lisitsyn | :) | 09:37 |
@wiking | 3 lines vs 9 | 09:38 |
@wiking | :) | 09:38 |
lisitsyn | but k.set_width adds some sense | 09:38 |
lisitsyn | wiking: it is better as we don't use kwargs | 09:38 |
@wiking | ? | 09:38 |
lisitsyn | wiking: is it 10, 0.1 or 0.1, 10? :) | 09:38 |
@wiking | i see what you mean | 09:38 |
@wiking | but now if we do a refact | 09:38 |
lisitsyn | can you remember if first arg is CACHE SIZE and the second one is width :) | 09:38 |
@wiking | we could extend the objs | 09:38 |
@wiking | with a custom ctor for python | 09:39 |
@wiking | that does kwargs | 09:39 |
@wiking | no? | 09:39 |
lisitsyn | wiking: or what if you don't care about cache size? :) | 09:40 |
@wiking | nono | 09:40 |
@wiking | i'm not trying to defend | 09:40 |
@wiking | the old way | 09:40 |
lisitsyn | wiking: yes we can add a lot of sugar for python | 09:40 |
@wiking | i'm just saying that since we wanna do it better | 09:40 |
@wiking | than the old one | 09:40 |
lisitsyn | wiking: we can also do k.width = 2 | 09:40 |
@wiking | i dont think that the current version is soooo much better | 09:40 |
lisitsyn | because it can be reduced to __setattr_ | 09:40 |
@wiking | yeah | 09:40 |
@wiking | that too | 09:40 |
@wiking | i'm saying maybe | 09:41 |
@wiking | this patch | 09:41 |
@wiking | or the whole issue | 09:41 |
@wiking | is way too early | 09:41 |
@wiking | as we could fine tune this api | 09:41 |
@wiking | much much more | 09:41 |
@wiking | i mean if we start to diverge too much in between lang interfaces | 09:42 |
@wiking | it'll become a bit of a mess | 09:42 |
@wiking | but i would say syntactic sugars like this would be good to have | 09:42 |
lisitsyn | wiking: we can add it definitely but it needs hacking the generator | 09:43 |
@wiking | you mean .i | 09:43 |
lisitsyn | because it would be supported only in py | 09:43 |
@wiking | mmm | 09:43 |
@wiking | not only | 09:43 |
@wiking | but yes | 09:43 |
@wiking | never used this though Python | 09:44 |
@wiking | http://www.swig.org/Doc3.0/Python.html#Python_nn67 | 09:44 |
@wiking | ;) | 09:44 |
@wiking | http://www.swig.org/Doc3.0/Ruby.html#Ruby_nn67 | 09:44 |
@wiking | ruby has it | 09:44 |
@wiking | :) | 09:44 |
@wiking | bu tyeah i'll check around | 09:45 |
@wiking | but first somebody should fix | 09:53 |
@wiking | doxygen + swig | 09:53 |
@wiking | :) | 09:53 |
@wiking | as that is utterly broken atm | 09:53 |
@wiking | ;( | 09:53 |
@wiking | mmm | 10:03 |
@wiking | ppl it seems there's something wrong with CGaussian since the watchers | 10:03 |
@wiking | frame #2: 0x000000010339e622 libshogun.18.dylib`shogun::CSGObject::Self::has(shogun::BaseTag const&) const [inlined] std::__1::map<shogun::BaseTag, shogun::AnyParameter, std::__1::less<shogun::BaseTag>, std::__1::allocator<std::__1::pair<shogun::BaseTag const, shogun::AnyParameter> > >::find(this=0x0000000000000000, __k=0x00007ffeefbfe528) const at map:1222 | 10:03 |
@wiking | and this is from frame #7: 0x000000010359e1d1 libshogun.18.dylib`shogun::CGaussian::register_params(this=0x000000010c409ed0) at Gaussian.cpp:344 | 10:03 |
@wiking | :( | 10:04 |
@wiking | dunno how this didn't come up all the tests we have | 10:04 |
@wiking | :D | 10:04 |
@wiking | but this is on my machine | 10:04 |
@wiking | lisitsyn, you've compiled lately/ran unit test on your osx? | 10:05 |
lisitsyn | wiking: if we don't switch to setters a few things would be impossible :) | 10:09 |
lisitsyn | like plugins and stuff | 10:09 |
lisitsyn | and actually | 10:10 |
lisitsyn | we can't switch to new parameters either | 10:10 |
lisitsyn | because this way we get it automagically | 10:10 |
lisitsyn | but other way we need to rewrite all the getters | 10:10 |
lisitsyn | it is just a way to enforce we don't use get_fdfds anymore | 10:11 |
@wiking | yeye | 10:11 |
@wiking | again | 10:11 |
@wiking | my point is that maybe we should do a bit more ++ on the api | 10:12 |
@wiking | for the modulars | 10:12 |
@wiking | before we start to refactor the metaexamples | 10:12 |
@wiking | i understand the why and hows | 10:12 |
@wiking | and made peace with it :) | 10:12 |
@wiking | just trying to compile shogun now on a linux | 10:13 |
@wiking | and see if i see the same bug | 10:13 |
@wiking | :S | 10:13 |
@wiking | miju | 10:17 |
@wiking | we have a problemo on os | 10:17 |
@wiking | x | 10:17 |
@wiking | X | 10:17 |
@wiking | :( | 10:17 |
@wiking | oh noooooo | 10:19 |
@wiking | i think i know | 10:19 |
@wiking | :)))) | 10:19 |
@wiking | std::unique was standing in my way :S | 10:21 |
@wiking | should never do a line of code before coffee | 10:22 |
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has joined #shogun | 11:50 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 11:50 | |
@HeikoS | lisitsyn: https://travis-ci.org/shogun-toolbox/shogun/jobs/332418575#L2332 | 11:51 |
@HeikoS | wiking: https://travis-ci.org/shogun-toolbox/shogun/jobs/332418575#L2332 | 11:51 |
@HeikoS | ideas? | 11:51 |
@wiking | both gc and clang? | 11:52 |
@HeikoS | y | 11:52 |
@HeikoS | here is gcc: https://travis-ci.org/shogun-toolbox/shogun/jobs/332418574#L2208 | 11:53 |
@wiking | mmm | 11:53 |
@wiking | shito | 11:53 |
@wiking | i've only seen this with protobuf previously | 11:53 |
@wiking | coz of different abis | 11:53 |
@wiking | but its weird as you supposed to be compiling the things iwth the very same compiler flags | 11:54 |
@wiking | so :O | 11:54 |
@HeikoS | is there maybe the class_list object missing when linking the unit tests? | 11:54 |
@HeikoS | https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html | 11:55 |
@HeikoS | If you get linker errors about undefined references to symbols that involve types in the std::__cxx11 namespace or the tag [abi:cxx11] then it probably indicates that you are trying to link together object files that were compiled with different values for the _GLIBCXX_USE_CXX11_ABI macro. This commonly happens when linking to a third-party library that was compiled with an older version of GCC. If the third-party librar | 11:55 |
@HeikoS | mmmmmmh | 11:55 |
@HeikoS | why would that be the case? | 11:56 |
@wiking | HeikoS, see my comment above | 11:56 |
@wiking | i mean i wasjust telling u this | 11:56 |
@wiking | :) | 11:56 |
@HeikoS | yep | 11:56 |
@wiking | gotta go ... lunch | 11:56 |
@wiking | bbl | 11:56 |
@HeikoS | enjoy | 11:56 |
@wiking | lisitsyn, y0 | 12:56 |
@HeikoS | wiking: it is the std::string | 13:16 |
@HeikoS | there they changed the abi | 13:16 |
@HeikoS | I will replace with const char* for now | 13:16 |
@HeikoS | this is temporary anyways, until we have full tag coverage | 13:16 |
@wiking | mmmm | 13:16 |
@wiking | dont push it into devleop | 13:17 |
@wiking | lplz | 13:17 |
@wiking | plz | 13:17 |
@wiking | as i'm already using it | 13:17 |
@wiking | for some other thigns | 13:17 |
@HeikoS | using what? | 13:18 |
@wiking | std::set<std::string> parameter_names() const; | 13:18 |
@wiking | so if u r having problems here | 13:18 |
@HeikoS | I am talking std::set<const char*> available_objects(); | 13:18 |
@HeikoS | the other one is fine | 13:18 |
@HeikoS | so all good | 13:19 |
@wiking | mmm | 13:19 |
@wiking | but what i dont get | 13:19 |
@wiking | is that why would that matter | 13:19 |
@wiking | i mean std::string | 13:19 |
@wiking | you are using the same libc++ | 13:19 |
@wiking | or i mean libstdc++ | 13:19 |
@wiking | or? | 13:19 |
@wiking | where do you think you switch stdc++ implementaiton? | 13:19 |
@HeikoS | no idea | 13:19 |
@wiking | so then | 13:19 |
@wiking | ? | 13:19 |
@HeikoS | yeah | 13:19 |
@wiking | i mean that should be fixed | 13:19 |
@HeikoS | but whatever | 13:19 |
@HeikoS | this method will not stay around | 13:20 |
@wiking | because this is just masking the problem | 13:20 |
@HeikoS | neither will the test | 13:20 |
@HeikoS | all temp stuff to help us swtich to tags | 13:20 |
@wiking | yeah but i mean | 13:20 |
@wiking | we have a bigger problem | 13:20 |
@wiking | of not using thesame libstdc++ somewhere | 13:20 |
@HeikoS | ha | 13:20 |
@HeikoS | actually | 13:20 |
@wiking | i mean that could be the root of many other segfaults | 13:20 |
@wiking | that we've seen around lately | 13:20 |
@HeikoS | [ 62%] Linking CXX executable ../../bin/shogun-unit-test | 13:20 |
@HeikoS | /home/heiko/git/shogun/tests/unit/base/SGObject_unittest.cc:555: error: undefined reference to 'shogun::available_objects()' | 13:20 |
@HeikoS | I think this is not abi related after all | 13:21 |
@HeikoS | this is what I get when changing to const char* | 13:21 |
@HeikoS | is maybe class_list object not included when linking the unit tests? | 13:21 |
@wiking | mmm | 13:21 |
@wiking | check it | 13:21 |
@wiking | idk | 13:21 |
@wiking | but there's a chance | 13:22 |
@wiking | as we till now never used | 13:22 |
@wiking | any of the methods in class_list.o | 13:22 |
@wiking | afaik | 13:22 |
@wiking | from unit tests | 13:22 |
@HeikoS | yes exactly | 13:23 |
@HeikoS | wiking: btw I will soon remove all jinja tests | 13:25 |
@HeikoS | can all be done with a single test and create_object | 13:25 |
@wiking | i've removed | 13:25 |
@wiking | locally the serialization tests | 13:25 |
@HeikoS | ah ok | 13:25 |
@wiking | i mean that's part of the features/cereal | 13:26 |
@wiking | story | 13:26 |
@HeikoS | then I will only write replacements | 13:26 |
@wiking | to drop all those | 13:26 |
@HeikoS | and push them, then you can drop after in develop | 13:26 |
@wiking | as those are foobar tests | 13:26 |
@wiking | btw have u seen the logs? | 13:26 |
@wiking | mmm i guess not | 13:26 |
@wiking | lemme point u to something | 13:26 |
@wiking | https://gitter.im/shogun-toolbox/shogun/archives/2018/01/24 | 13:27 |
@wiking | beginning | 13:27 |
@HeikoS | the lines of code for setting parameters? | 13:29 |
@wiking | yeah | 13:29 |
@HeikoS | I think we can probably solve that with swig or other tricks | 13:30 |
@HeikoS | like setting multiple parameters at once call | 13:30 |
@wiking | well | 13:30 |
@wiking | unfortunately kwargs story only exists | 13:30 |
@wiking | in a handful of langs | 13:30 |
@HeikoS | I dont like the old API at all with these random cache size vs width things | 13:30 |
@wiking | so that's not like a meta exampleable story | 13:30 |
@HeikoS | that's a consequence of freezing the API | 13:31 |
@wiking | i'm not saying that the old one is good | 13:31 |
@wiking | i'm saying that maybe we should do these refactors | 13:31 |
@wiking | yet | 13:31 |
@HeikoS | I mean suggest something | 13:31 |
@wiking | as i believe we are not yet a stage | 13:31 |
@wiking | yes | 13:31 |
@wiking | this is a suggestion | 13:32 |
@wiking | for example | 13:32 |
@wiking | that we dont do this yet | 13:32 |
@wiking | 2nd that kwargs ctors can be done for some of the langs | 13:32 |
@HeikoS | we are not yet dropping the old stuff | 13:32 |
@HeikoS | just adding the possibility to do it the new way | 13:32 |
@wiking | i understand that | 13:32 |
@HeikoS | so your suggestion is already implemented | 13:32 |
@wiking | where? | 13:32 |
@wiking | i dont see the possibility to do | 13:33 |
@HeikoS | GaussianKernel(2.0, 2.0) is still there | 13:33 |
@wiking | no | 13:33 |
@wiking | it's not kwargs | 13:33 |
@HeikoS | ? | 13:33 |
@HeikoS | sorry I dont understand | 13:33 |
@wiking | python kwargs** | 13:33 |
@wiking | ctor(key1=value1, key2=value2, ...) | 13:34 |
@wiking | that will just internally do a | 13:34 |
@wiking | ctor(); ctor.put(key1, value1).... | 13:34 |
@HeikoS | yes that's what I meant above | 13:34 |
@wiking | that is not implementat | 13:34 |
@wiking | anywhere | 13:34 |
@HeikoS | in python thats how we should do it | 13:34 |
@HeikoS | ah I see now | 13:34 |
@HeikoS | so you mean | 13:34 |
@wiking | you cannot just do | 13:34 |
@HeikoS | lets FIRST implement the kwargs | 13:34 |
@HeikoS | so we dont have to touch everything multiple times | 13:34 |
@HeikoS | in the examples | 13:35 |
@wiking | Kernel(name='Gaussian', width=10, cache=111) | 13:35 |
@wiking | etc | 13:35 |
@HeikoS | yep +1 | 13:35 |
@wiking | i dont see the point of pushign metaexmaples | 13:35 |
@wiking | all the time | 13:35 |
@wiking | just because we farted out a new idea yesterday | 13:35 |
@wiking | lets first try to get into where we wanna be | 13:35 |
@wiking | and then do those refactors | 13:35 |
@HeikoS | so I think the steps should be | 13:40 |
@HeikoS | 1) kwargs for python prototype | 13:40 |
@HeikoS | 2) meta language ability to express this | 13:40 |
@HeikoS | 3) transition | 13:40 |
@HeikoS | Ill update the issue with that | 13:40 |
@HeikoS | wiking: you know what the state of std::map in swig is? | 13:41 |
@wiking | afaik most of the languages supports it | 13:42 |
@wiking | but github.com/swig/swig | 13:42 |
@HeikoS | so we could even add a method | 13:42 |
@HeikoS | CSGObject::set(std::map) | 13:42 |
@HeikoS | or put it is called | 13:42 |
@HeikoS | but kwargs is nicer | 13:42 |
@wiking | even R has a std::map wrapper | 13:43 |
@wiking | https://github.com/swig/swig/blob/master/Lib/r/std_map.i | 13:43 |
@wiking | :) | 13:43 |
@wiking | but yeah i would check it before heavily depending on it | 13:43 |
@wiking | :) | 13:43 |
@HeikoS | actually, we can just hide this in the meta lang api | 13:53 |
@HeikoS | just do kwargs in there | 13:53 |
@HeikoS | and then translate it to what works | 13:54 |
@HeikoS | then the fact that we dont have it implemented is not a blocker | 13:54 |
@HeikoS | smv.sg contains svm = SVM("LibSVM", C=1) | 13:54 |
@HeikoS | which right now would be translated into | 13:55 |
@HeikoS | svm = LibSVM() | 13:55 |
@HeikoS | svm.C=1 | 13:55 |
@HeikoS | but then we don't need to refactor the examples later on | 13:55 |
@HeikoS | only change the translation | 13:55 |
@HeikoS | tada | 13:55 |
lisitsyn | that's a good idea | 13:55 |
lisitsyn | if it supports kwargs - do kwargs | 13:55 |
lisitsyn | if not do get | 13:55 |
@HeikoS | nonintrusive nonblocking :) | 13:56 |
@HeikoS | lisitsyn: the linker error is just missing object | 13:56 |
@HeikoS | im deciphering the cmake to fix it | 13:56 |
@HeikoS | lisitsyn: btw what do you think of a put method that takes a std::map? | 13:57 |
@wiking | boom | 14:13 |
@wiking | https://github.com/xoreaxeaxeax/movfuscator/tree/master/validation/doom | 14:13 |
@wiking | :) | 14:13 |
@HeikoS | hehe | 14:16 |
@HeikoS | wiking: I have a question for you | 14:25 |
@HeikoS | class_list.o is included when lishogun.so is linked | 14:25 |
@HeikoS | and then libshogun.so is included when shogun-unit-test is linked | 14:25 |
@HeikoS | so I don't see whats wrong | 14:25 |
@HeikoS | lisitsyn: you around? | 14:26 |
lisitsyn | HeikoS: partially | 14:26 |
@HeikoS | lisitsyn: I changed your method to std::set<const char*> available_objects(); | 14:27 |
@HeikoS | and now the error became /home/heiko/git/shogun/tests/unit/base/SGObject_unittest.cc:555: error: undefined reference to 'shogun::available_objects()' | 14:27 |
@HeikoS | so it is probably not an abi problem | 14:27 |
@HeikoS | lisitsyn: but I still don't understand it | 14:27 |
@HeikoS | lisitsyn: you have ideas? | 14:27 |
lisitsyn | HeikoS: let me check the code once again | 14:28 |
lisitsyn | HeikoS: map is fine but it needs some work | 14:28 |
lisitsyn | I don't see the problem wiking sees :) | 14:28 |
@HeikoS | and weirdly | 14:29 |
@HeikoS | nm -D src/shogun/libshogun.so | grep available_objects | 14:29 |
@HeikoS | 0000000003a1aa46 T _Z17available_objectsv | 14:29 |
@HeikoS | it is there | 14:29 |
lisitsyn | HeikoS: I forgot namespace | 14:30 |
@HeikoS | ok let me try | 14:30 |
@HeikoS | :) | 14:30 |
lisitsyn | HeikoS: shogun:: | 14:30 |
lisitsyn | sorry my fault | 14:30 |
lisitsyn | :) | 14:30 |
@HeikoS | ah fuck! | 14:30 |
@HeikoS | hahaha | 14:30 |
@HeikoS | lisitsyn: with the kwargs parameter we can just do the meta trick | 14:31 |
@HeikoS | then whoever feels inspired to implement kwargs, they can do it | 14:31 |
@HeikoS | and everything magically updates | 14:31 |
@HeikoS | but we don't want any blockers for moving | 14:31 |
@HeikoS | lisitsyn: actually I like the explicitness of the tags thing much more than those constructors | 14:32 |
@HeikoS | and it is not like we have thousands of parameters | 14:32 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 opened by karlnapf | 14:35 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 karlnapf added label: "good first issue" | 14:35 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 | 14:40 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4114 synchronized by karlnapf | 14:44 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4114 | 14:47 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 | 14:51 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4118 opened by guruhegde | 14:52 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 | 14:55 |
@HeikoS | lisitsyn: https://github.com/shogun-toolbox/shogun/issues/4117 | 14:55 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 | 14:56 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4110 | 14:57 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4101 | 14:57 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4050 closed by karlnapf | 14:57 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4118 | 15:06 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4116 synchronized by guruhegde | 15:10 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4114 merged by karlnapf | 15:58 |
@sukey | [https://github.com/shogun-toolbox/shogun] karlnapf pushed 2 commits: | 15:58 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/1d0c05a54af7ca2a18de0f17988f2f5bb0ab47b3 | 15:58 |
@sukey | https://github.com/shogun-toolbox/shogun/commit/fad3dfe46e2d420024558d5d82ea8106e2a27e6d | 15:58 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/1972 closed by karlnapf | 16:03 |
-!- witness [uid10044@gateway/web/irccloud.com/x-bizuqzedrstjinus] has quit [Quit: Connection closed for inactivity] | 16:06 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4117 | 16:10 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4118 closed by karlnapf | 16:13 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/4047 | 16:17 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/3992 karlnapf added label: "testing" | 16:19 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/3992 karlnapf added label: "BUG" | 16:19 |
@sukey | [https://github.com/shogun-toolbox/shogun] Issue https://github.com/shogun-toolbox/shogun/issues/3992 karlnapf removed label: "good first issue" | 16:19 |
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-peulujplhienhnwb] has joined #shogun | 17:08 | |
-!- travis-ci [~travis-ci@ec2-54-162-9-36.compute-1.amazonaws.com] has joined #shogun | 17:44 | |
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/332823126 | 17:44 |
-!- travis-ci [~travis-ci@ec2-54-162-9-36.compute-1.amazonaws.com] has left #shogun [] | 17:44 | |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4103 synchronized by shubham808 | 17:57 |
@sukey | [https://github.com/shogun-toolbox/shogun] Pull Request https://github.com/shogun-toolbox/shogun/pull/4103 synchronized by shubham808 | 17:59 |
-!- travis-ci [~travis-ci@ec2-54-161-146-113.compute-1.amazonaws.com] has joined #shogun | 18:41 | |
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/332823126 | 18:41 |
-!- travis-ci [~travis-ci@ec2-54-161-146-113.compute-1.amazonaws.com] has left #shogun [] | 18:41 | |
Trixis | Hm there seems to be some sort of an issue with director kernels and crossvalidation in python | 19:02 |
Trixis | now i just need to get the code to actually write the segfault or whatever into logs, instead of core dump followed by process hanging | 19:02 |
@HeikoS | Trixis: uh that is untested land :) | 19:03 |
@HeikoS | but yeah let's fix it | 19:03 |
Trixis | hahaha | 19:03 |
Trixis | it printed a proper error *once* | 19:04 |
@HeikoS | I am sure than can be fixed | 19:04 |
@HeikoS | it is just a lot of framework stuff piled up in that case | 19:05 |
Trixis | ah | 19:05 |
Trixis | and even then it was very vague, might be something to do with the director kernels getting cloned when cross validation runs? | 19:05 |
Trixis | ill try a minimal working example that replicates it | 19:06 |
@HeikoS | cool thanks appreciated | 19:06 |
@HeikoS | send it and I will try to help | 19:06 |
@HeikoS | Ill have to run now though | 19:09 |
@HeikoS | talk tomorrow? | 19:09 |
Trixis | kk | 19:09 |
@wiking | Trixis, just push the thingy into an issue | 19:16 |
-!- shubham808 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has joined #shogun | 19:17 | |
shubham808 | Hi | 19:19 |
Trixis | wiking: ye, probs tomorrow, gotta run now as well | 19:20 |
@wiking | kk | 19:20 |
@wiking | nw | 19:20 |
Trixis | ill just use custom kernels in the meantime | 19:20 |
shubham808 | @HeikoS i have been working on #4103 | 19:23 |
shubham808 | @lambday has used completely different jobs and aggregators here which store the result in complex form and they are converted to scalar only when they are finalized | 19:25 |
-!- Prashanth [uid170915@gateway/web/irccloud.com/x-peulujplhienhnwb] has quit [Quit: Connection closed for inactivity] | 19:26 | |
-!- shubham808 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has quit [Quit: Page closed] | 19:35 | |
-!- shubham808 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has joined #shogun | 20:12 | |
-!- shubham808 [0e8bf0fb@gateway/web/freenode/ip.14.139.240.251] has quit [Client Quit] | 20:13 | |
-!- yash_ [01ba2545@gateway/web/freenode/ip.1.186.37.69] has joined #shogun | 20:37 | |
-!- yash_ [01ba2545@gateway/web/freenode/ip.1.186.37.69] has quit [Ping timeout: 260 seconds] | 21:28 | |
-!- gp [~gp@ip4d15fa06.dynamic.kabel-deutschland.de] has joined #shogun | 22:46 | |
-!- Ansh [~Ansh@157.41.142.184] has joined #shogun | 23:28 | |
-!- HeikoS [~heiko@host86-129-231-92.range86-129.btcentralplus.com] has quit [Ping timeout: 260 seconds] | 23:38 | |
-!- AnshBalde [~Ansh@157.41.142.184] has joined #shogun | 23:46 | |
-!- Ansh [~Ansh@157.41.142.184] has quit [Ping timeout: 256 seconds] | 23:47 | |
-!- AnshBalde [~Ansh@157.41.142.184] has quit [Ping timeout: 252 seconds] | 23:52 | |
--- Log closed Thu Jan 25 00:00:34 2018 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!