--- Log opened Tue Jul 26 00:00:38 2016 | ||
-!- 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] | 00:34 | |
-!- GandalfTheWizard [~Emma@112.10.170.216] has joined #shogun | 00:48 | |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has joined #shogun | 04:14 | |
-!- GandalfTheWizard [~Emma@112.10.170.216] has quit [Ping timeout: 240 seconds] | 04:29 | |
@wiking | OXPHOS: if you are touching CHAIDTree | 05:13 |
---|---|---|
@wiking | do you mind to fix a bug there? | 05:13 |
OXPHOS | wiking: I can try | 05:14 |
@wiking | OXPHOS: lemme dig it out | 05:14 |
@wiking | OXPHOS: one is rather simple: http://buildbot.shogun-toolbox.org/static_analysis/2016-07-13-010034-30511-1/report-e95725.html#EndPath | 05:16 |
OXPHOS | wiking: not sure how to read it..what is the problem there? | 05:20 |
@wiking | see the 5 steps | 05:20 |
@wiking | it basically describe you a situation | 05:21 |
@wiking | where you gonna have a null pointer exception | 05:21 |
@wiking | when you do CDynamicObjectArray* children=node->get_children(); | 05:21 |
@wiking | coz | 05:21 |
@wiking | node_t* node=current | 05:21 |
@wiking | but there's never a check whether current != NUL | 05:21 |
@wiking | L | 05:21 |
@wiking | got it? | 05:21 |
OXPHOS | so there could be situation when current=NULL | 05:22 |
OXPHOS | just we need a check there | 05:22 |
OXPHOS | wiking^ | 05:23 |
@wiking | yeah i mean the easiest is that you add an assertation | 05:24 |
@wiking | in the beginning of the function call of CLabels* CCHAIDTree::apply_from_current_node(SGMatrix<float64_t> fmat, node_t* current) | 05:24 |
@wiking | that current != NULL | 05:24 |
@wiking | it's a simple input value check for the function | 05:26 |
OXPHOS | wiking: kk i see | 05:27 |
OXPHOS | wiking: and for the sgvector serialization - any comments on the refcount? https://github.com/shogun-toolbox/shogun/pull/3226 | 05:28 |
@wiking | yeah i'm gonna look into that now | 05:28 |
@wiking | ok so another thing is | 05:28 |
@wiking | can you login here: https://scan.coverity.com | 05:28 |
@wiking | with your github accoutn? | 05:28 |
OXPHOS | wiking: "You are not associated with any projects" | 05:29 |
OXPHOS | In order to start using Scan you need to be associated with at least one project. | 05:30 |
@wiking | ah | 05:30 |
@wiking | can you tell me what is your email address associated with your github account? | 05:30 |
OXPHOS | engelzora@gmail.com | 05:30 |
OXPHOS | I can do "add me to the project" | 05:31 |
@wiking | send you an invite | 05:31 |
OXPHOS | wiking: thx. I logged in. | 05:34 |
@wiking | ok can u see now the shogun project? | 05:34 |
@wiking | you should be able to view the defects | 05:34 |
@wiking | i wonder if you can open this link: https://scan8.coverity.com/reports.htm#v11736/p10202/fileInstanceId=12508917&defectInstanceId=4294867&mergedDefectId=135945 | 05:35 |
@wiking | but once you can go to the defects you should try to look at the 135945 CID | 05:35 |
@wiking | that's the CHAID related resource leak | 05:35 |
@wiking | the thing is that it's rather complicated :) | 05:35 |
@wiking | OXPHOS: so ok for sgvect | 05:37 |
@wiking | so the thing is that we serialize the sgvec as part of a model right? | 05:37 |
OXPHOS | yep | 05:37 |
@wiking | say i have a LinearMachine that has a weight vector | 05:37 |
@wiking | SGVector<float64_t> | 05:37 |
@wiking | w | 05:37 |
@wiking | so you need to serialize the refcount structure | 05:38 |
@wiking | because in this case there would be refcount = 1 | 05:38 |
@wiking | but thin kabout a more complicated model | 05:38 |
OXPHOS | like? | 05:41 |
OXPHOS | btw I saw the leaks | 05:42 |
@wiking | what do yo umean lik? | 05:42 |
@wiking | *like | 05:42 |
OXPHOS | oh I was waiting for your example about the "more complicated model" | 05:43 |
OXPHOS | wiking^ | 05:43 |
@wiking | well more complicated can be | 05:45 |
@wiking | that you have some Label vector | 05:45 |
@wiking | and say those are referenced my the machines in the bagging machine | 05:46 |
@wiking | bagging machine like RF | 05:46 |
@wiking | so that label vector would have like | 05:47 |
@wiking | refcount = as many trees there are in RF | 05:47 |
OXPHOS | yeah.. | 05:48 |
@wiking | so yeah you have to set serialize the ref count | 05:48 |
@wiking | at least the count | 05:48 |
OXPHOS | okay. yeah i was thinking i have to | 05:51 |
@wiking | like i twas in the old serialization case right? | 05:51 |
OXPHOS | right | 05:52 |
OXPHOS | lemme see how to do this inheritance serialization | 05:52 |
@wiking | mmm i mean you dont need to serialize the SGRef object itself | 05:55 |
@wiking | but yeah | 05:56 |
@wiking | that's one way | 05:56 |
OXPHOS | wiking: I just need the m_refcount, i.e. RefCount class right? but m_refcount is private in SGRef. so I guess I have to go through SGRef | 06:02 |
@wiking | OXPHOS: i mean the best is if you actually serialize the inherited things automatically | 06:05 |
@wiking | if possible | 06:05 |
@wiking | because that way if later there's a change in SGReferencedData then we dont need to modify anything in the serialization code | 06:05 |
@wiking | and you need to go through SGReferencedData | 06:05 |
@wiking | "class SGVector : public SGReferencedData" | 06:06 |
@wiking | OXPHOS: http://uscilab.github.io/cereal/polymorphism.html | 06:07 |
@wiking | ^ that's your howto | 06:07 |
@wiking | mmm | 06:08 |
@wiking | wait maybe not enough | 06:08 |
@wiking | but yeah you can define releation like | 06:08 |
@wiking | CEREAL_REGISTER_POLYMORPHIC_RELATION(SGVector, SGReferencedData) | 06:08 |
@wiking | CEREAL_REGISTER_POLYMORPHIC_RELATION(SGMatrix, SGReferencedData) | 06:09 |
@wiking | RefCount is just a member of SGReferencedData | 06:09 |
@wiking | but yeah i guess that'd need a serialization method as well :) | 06:09 |
@wiking | btw | 06:10 |
OXPHOS | saw it . thanks! | 06:10 |
@wiking | OXPHOS: http://uscilab.github.io/cereal/inheritance.html | 06:10 |
@wiking | but yeah | 06:10 |
@wiking | i have a question | 06:10 |
@wiking | so the template<Archive> save/load_cereal(Archive& a) methods are currently implemented in cpp | 06:11 |
@wiking | right? | 06:11 |
OXPHOS | yes | 06:11 |
OXPHOS | for sgobject | 06:11 |
OXPHOS | it's in sgvector.h | 06:11 |
@wiking | OXPHOS: yep yep i mean sgobject | 06:12 |
@wiking | would it be hard to move it to sgobject.h? | 06:12 |
@wiking | because the problem is that even if you expose now the templated version of the save/load thingy | 06:12 |
@wiking | the user will have to recompile shogun | 06:12 |
@wiking | if she wants to serialize it into a new Archive format | 06:12 |
@wiking | right? | 06:12 |
OXPHOS | I think all existed/specified formats will be compiled in the begining? | 06:13 |
@wiking | ok so let me put it this way | 06:14 |
OXPHOS | the only solution i can see is to move the SGObejct::self from .cpp to .h. Not sure how bazdback like this | 06:14 |
@wiking | there are now 3 formats | 06:14 |
@wiking | right? | 06:14 |
OXPHOS | yes | 06:14 |
@wiking | so what if i have a new Archive format | 06:14 |
@wiking | a custom one | 06:14 |
@wiking | because i'm a user like that | 06:14 |
@wiking | and i wanna use that for serializing shogun | 06:15 |
@wiking | objects | 06:15 |
OXPHOS | Yeah the codes has to be modified and everything will be re-compiled because it's SGObejct | 06:15 |
@wiking | in that case the only option is that actually i make that part of shogun | 06:15 |
@wiking | (the new format) | 06:15 |
@wiking | and then i need to recompile | 06:16 |
@wiking | ok so but if the templated load/save would be implemented in header | 06:16 |
@wiking | this coudl be avoided right? | 06:16 |
OXPHOS | I think so | 06:17 |
OXPHOS | yeah right | 06:17 |
@wiking | ok so | 06:18 |
@wiking | the question is | 06:18 |
@wiking | what is preventing us to be able to do this switch? | 06:19 |
OXPHOS | I need to iterate through the map, which is defined in .cpp | 06:21 |
OXPHOS | it's in a pimpl | 06:21 |
OXPHOS | wiking^ | 06:22 |
@wiking | ParametersMap? | 06:24 |
OXPHOS | wiking^ yes..it's self::map[basetag, any] | 06:26 |
@wiking | ok i suppose we dont want to expose the self in .h because of map/unordered_map | 06:30 |
@wiking | but lemme ask about this bazdback | 06:30 |
@wiking | :) | 06:30 |
OXPHOS | kk | 06:31 |
@wiking | so lets not merge that until we can resolve that either way | 06:32 |
@wiking | :p | 06:32 |
@wiking | i guess 12 hours more max | 06:32 |
-!- sanuj [~sanuj@117.204.242.64] has joined #shogun | 06:42 | |
-!- sanuj [~sanuj@117.204.242.64] has quit [Ping timeout: 260 seconds] | 07:08 | |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 07:15 | |
shogun-notifier- | shogun: Wu Lin :develop * f49f873 / cmake/FindMetaExamples.cmake: https://github.com/shogun-toolbox/shogun/commit/f49f8731ef0c0256b79b104ea916f37138ea2d37 | 07:15 |
shogun-notifier- | shogun: fix a bug in cmake | 07:15 |
shogun-notifier- | shogun: Wu Lin :develop * f3485af / cmake/FindMetaExamples.cmake: https://github.com/shogun-toolbox/shogun/commit/f3485af3f9bbce846d0c6f1cd013b09c25e53d24 | 07:15 |
shogun-notifier- | shogun: Merge pull request #3394 from yorkerlin/bugfix | 07:15 |
shogun-notifier- | shogun: | 07:15 |
shogun-notifier- | shogun: fix a bug in cmake | 07:15 |
shogun-buildbot | build #1309 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1309 blamelist: Wu Lin <yorkerlin@users.noreply.github.com>, Wu Lin <yorker.lin@gmail.com> | 07:17 |
shogun-buildbot | build #2951 of bsd1 - libshogun is complete: Failure [failed configure] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2951 blamelist: Wu Lin <yorkerlin@users.noreply.github.com>, Wu Lin <yorker.lin@gmail.com> | 07:26 |
shogun-buildbot | build #88 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/88 blamelist: Wu Lin <yorkerlin@users.noreply.github.com>, Wu Lin <yorker.lin@gmail.com> | 07:56 |
shogun-buildbot | build #38 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/38 blamelist: Wu Lin <yorkerlin@users.noreply.github.com>, Wu Lin <yorker.lin@gmail.com> | 08:05 |
-!- sanuj [~sanuj@117.204.242.64] has joined #shogun | 08:06 | |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has quit [Ping timeout: 250 seconds] | 08:22 | |
@wiking | bazdback: ping ping pingoooo ping | 08:50 |
@wiking | :) | 08:50 |
sanuj | i need you too bazdback | 08:52 |
shogun-notifier- | shogun: Viktor Gal :develop * 6249e61 / cmake/FindCtags.cmake: https://github.com/shogun-toolbox/shogun/commit/6249e61b211c667f7ca02bc29f802644bc7cfa12 | 08:52 |
@wiking | sanuj: what do you nneed? | 08:52 |
shogun-notifier- | shogun: Prefer Exuberant Ctags over GNU | 08:52 |
shogun-notifier- | shogun: this fixes the problem of FreeBSD's ctags | 08:52 |
@wiking | i might be able to help | 08:53 |
sanuj | wiking, he suggested me to use sfinae to solve the plugin bug | 08:53 |
@wiking | ah | 08:53 |
sanuj | i have read about sfinae and understand it a bit | 08:54 |
@wiking | yeah and wher ehave you got stuck with it? | 08:54 |
sanuj | but not sure how to solve that bug using sfinae | 08:54 |
@wiking | i mean sfinae is all about not erroring when there's an unsupported type | 08:54 |
sanuj | wiking, https://travis-ci.org/shogun-toolbox/shogun/jobs/147180067#L1394 | 08:54 |
@wiking | just simply silently NOP | 08:54 |
shogun-buildbot | build #1310 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1310 blamelist: Viktor Gal <viktor.gal@maeth.com> | 08:55 |
@wiking | sanuj: btw do you know of a good substitution for std::hash in none-c++11 case? | 08:55 |
@wiking | you mean this | 08:55 |
@wiking | /opt/shogun/src/shogun/lib/any.h:149:34: error: no match for 'operator==' (operand types are 'std::function<shogun::Some<shogun::MockBaseClass>()>' and 'std::function<shogun::Some<shogun::MockBaseClass>()>') | 08:55 |
@wiking | ? | 08:56 |
sanuj | yeah | 08:56 |
sanuj | but i'm not comparing std::function anywhere explicitly | 08:56 |
@wiking | what do you compare then? | 08:57 |
@wiking | i know that you do a | 08:57 |
@wiking | ASSERT_EQ() | 08:57 |
@wiking | right? | 08:57 |
sanuj | EXPECT_EQ | 08:58 |
@wiking | or that | 08:58 |
@wiking | yeah | 08:58 |
sanuj | wiking, but the compiler error is not from there | 08:58 |
@wiking | btw | 08:58 |
@wiking | in any.h | 08:58 |
@wiking | #include <string.h> | 08:58 |
sanuj | wiking, it happens while compiling this https://github.com/shogun-toolbox/shogun/pull/3370/files#diff-7e05729c8a117ab30510a77a08865f33R1 | 08:58 |
@wiking | shouldn't that be <cstring> | 08:58 |
@wiking | to be correct? | 08:59 |
sanuj | wiking, actually yes, i'll fix that | 08:59 |
@wiking | k | 08:59 |
sanuj | wiking, but how to go about debugging this | 09:00 |
sanuj | how to use sfinae | 09:00 |
sanuj | is it for debugging or preventing the bug from happening? | 09:00 |
@wiking | latter | 09:00 |
@wiking | i mean you wanna be able to compile this code | 09:00 |
@wiking | return typed_storage == typed_other_storage; | 09:01 |
@wiking | here typed_storage = std::function<shogun::Some<shogun::MockBaseClass>()> | 09:01 |
@wiking | right? | 09:01 |
sanuj | yeah | 09:01 |
sanuj | wiking, so i need to use sfinae in any.h so that it compiles | 09:03 |
shogun-buildbot | build #89 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/89 blamelist: Viktor Gal <viktor.gal@maeth.com> | 09:03 |
@wiking | so can you just provide an empty implementation for that? | 09:05 |
@wiking | i mean the specialized case of any template | 09:05 |
shogun-buildbot | build #2952 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2952 blamelist: Viktor Gal <viktor.gal@maeth.com> | 09:05 |
sanuj | wiking, okay, i see | 09:05 |
sanuj | wiking, so a specialized implementations for std::function? | 09:06 |
@wiking | i mean i guess anything that is std::function<?> | 09:07 |
sanuj | oh yeah | 09:07 |
sanuj | cool, i'll try patching any | 09:07 |
shogun-buildbot | build #39 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/39 blamelist: Viktor Gal <viktor.gal@maeth.com> | 09:42 |
-!- sanuj [~sanuj@117.204.242.64] has quit [Ping timeout: 258 seconds] | 09:48 | |
shogun-notifier- | shogun: Viktor Gal :develop * 515b334 / src/shogun/regression/LeastAngleRegression.h: https://github.com/shogun-toolbox/shogun/commit/515b3349560275e959aa459d8c5fd1d36d9a1f78 | 10:12 |
shogun-notifier- | shogun: Remove dependency for lapack in LARS | 10:12 |
-!- sanuj [~sanuj@117.204.242.64] has joined #shogun | 10:13 | |
shogun-buildbot | build #1311 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1311 blamelist: Viktor Gal <viktor.gal@maeth.com> | 10:14 |
shogun-buildbot | build #90 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/90 blamelist: Viktor Gal <viktor.gal@maeth.com> | 10:24 |
shogun-buildbot | build #2953 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2953 blamelist: Viktor Gal <viktor.gal@maeth.com> | 10:26 |
shogun-notifier- | shogun: Viktor Gal :develop * 14355ea / tests/unit/ (2 files): https://github.com/shogun-toolbox/shogun/commit/14355ea1b0baffe6b1eb8356000033bf5fa81445 | 10:40 |
shogun-notifier- | shogun: Include <numeric> when using std::iota | 10:40 |
shogun-buildbot | build #1312 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1312 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:06 |
shogun-buildbot | build #40 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/40 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:11 |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has joined #shogun | 11:14 | |
shogun-buildbot | build #2954 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2954 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:14 |
shogun-buildbot | build #91 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/91 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:15 |
OXPHOS | Hey wiking, I was trying to #include <cereal/types/base_class.hpp>, but got the error halfway: http://pastebin.com/BYXbQVBN any idea why? | 11:16 |
@wiking | just a sec checking | 11:22 |
-!- OXPHOS_ [c0a906af@gateway/web/freenode/ip.192.169.6.175] has joined #shogun | 11:31 | |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has quit [Ping timeout: 250 seconds] | 11:33 | |
@wiking | shogun-buildbot: force build --branch=develop 'bsd1 - libshogun' | 11:51 |
shogun-buildbot | build #2955 forced | 11:52 |
shogun-buildbot | I'll give a shout when the build finishes | 11:52 |
shogun-buildbot | build #2955 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2955 | 11:54 |
shogun-buildbot | build #41 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/41 blamelist: Viktor Gal <viktor.gal@maeth.com> | 11:54 |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has joined #shogun | 12:00 | |
-!- mode/#shogun [+o lambday] by ChanServ | 12:00 | |
shogun-notifier- | shogun: Viktor Gal :develop * 9273672 / cmake/FindMetaExamples.cmake: https://github.com/shogun-toolbox/shogun/commit/9273672ad3afa6d4ead25fbdadf2660b6cc8aeb8 | 12:00 |
shogun-notifier- | shogun: Remove GMM meta example if LAPACK is not available | 12:00 |
@wiking | OXPHOS_: mmm | 12:02 |
@wiking | OXPHOS_: all you did is just to include the cereal/types/base_class | 12:02 |
@wiking | .hpp:33: | 12:02 |
@wiking | ? | 12:02 |
OXPHOS_ | wiking: YES | 12:02 |
shogun-buildbot | build #1313 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1313 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:03 |
@wiking | :) upset? :) | 12:03 |
@wiking | or why the capitals? :> | 12:03 |
OXPHOS_ | confused.. :P | 12:03 |
@wiking | u r asian | 12:04 |
@wiking | when u write with capitals | 12:04 |
@wiking | people are super angry | 12:04 |
@wiking | :) | 12:04 |
@wiking | ok so what's your cereal version? | 12:04 |
bazdback | BAZDMAAAGGGZZ | 12:06 |
@wiking | bazdback: y0 | 12:07 |
@wiking | OXPHOS_: ok so just to make sure | 12:07 |
@wiking | if you take that example from http://uscilab.github.io/cereal/inheritance.html | 12:07 |
@wiking | and do | 12:07 |
OXPHOS_ | wiking: sorry..i was angry about the error. should be 1.2.0 but I need to confirm | 12:07 |
@wiking | clang++ -I/usr/local/include -std=c++11 -c test_cereal.cpp | 12:08 |
@wiking | what happens? | 12:08 |
@wiking | bazdback: morning | 12:08 |
OXPHOS_ | wiking: no error | 12:09 |
OXPHOS_ | wiking: compiled | 12:09 |
@wiking | OXPHOS_: super :) | 12:09 |
shogun-buildbot | build #92 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/92 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:11 |
@wiking | OXPHOS_: btw include it before #include <shogun/lib/config.h> | 12:12 |
@wiking | i mean the base clase | 12:12 |
@wiking | *"class | 12:12 |
shogun-buildbot | build #2956 of bsd1 - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2956 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:13 |
OXPHOS_ | lemme try | 12:14 |
OXPHOS_ | wiking: not helping | 12:16 |
OXPHOS_ | wiking: SVMLightOneClass.cpp is not the first .cpp includes base_class.hpp. I don't know why the previous ones were fine | 12:17 |
bazdback | wiking: hey | 12:35 |
-!- bazdback is now known as lisitsyn | 12:35 | |
@wiking | ho | 12:37 |
-!- travis-ci [~travis-ci@ec2-54-91-68-140.compute-1.amazonaws.com] has joined #shogun | 12:42 | |
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/147423516 | 12:42 |
-!- travis-ci [~travis-ci@ec2-54-91-68-140.compute-1.amazonaws.com] has left #shogun [] | 12:42 | |
shogun-buildbot | build #42 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/42 blamelist: Viktor Gal <viktor.gal@maeth.com> | 12:50 |
-!- HeikoS [~heiko@smb-amrs-02.hotspot.hub-one.net] has joined #shogun | 12:51 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 12:52 | |
-!- HeikoS [~heiko@smb-amrs-02.hotspot.hub-one.net] has quit [Client Quit] | 12:56 | |
OXPHOS_ | wiking: hey any updates? : ) | 12:57 |
@wiking | on my end? | 13:00 |
@wiking | i dont have th ecode :))) | 13:01 |
@wiking | OXPHOS_: i'm trying to fix some opaque shit in the mkl | 13:01 |
sanuj | wiking, you work on computer vision right? | 13:01 |
@wiking | sanuj: used to | 13:01 |
@wiking | not lately | 13:01 |
@wiking | lately i'm a fucking devops idiot | 13:01 |
@wiking | :D | 13:01 |
sanuj | haha | 13:02 |
sanuj | wiking, what's your phd on? | 13:02 |
OXPHOS_ | wiking: ohhhh any hints on how to debug? It literally requires nothing but #include <base_class.hpp> to cause the error | 13:04 |
sanuj | lisitsyn, there? | 13:11 |
@wiking | sanuj: medical image processing | 13:12 |
@wiking | OXPHOS_: what you can do | 13:12 |
@wiking | is do a clang++ -E | 13:12 |
@wiking | that'll basically generate the whole file | 13:12 |
sanuj | wiking, i got an intern at this lab camp.lcsr.jhu.edu/intern/ | 13:13 |
@wiking | jhu is cool | 13:13 |
sanuj | they do robotics and vision in medicine | 13:13 |
@wiking | OXPHOS_: so with that (g++ or clang++) doesn't matter | 13:14 |
@wiking | -E will just generate the file, i.e. resolve all the macros | 13:14 |
OXPHOS_ | wiking: thanks! lemme try | 13:16 |
OXPHOS_ | sanuj: congrats! | 13:16 |
sanuj | OXPHOS_, thanks :D | 13:16 |
sanuj | wiking, you know any online sources from where i can learn about vision stuff related to augmented reality? | 13:21 |
@wiking | mmm | 13:21 |
@wiking | scholar.google.com | 13:21 |
@wiking | :) | 13:21 |
sanuj | :D | 13:21 |
@wiking | you think i'm joking :) | 13:30 |
sanuj | nope :) | 13:34 |
@wiking | lisitsyn: bazdmeg here? | 13:50 |
lisitsyn | wiking: yeap | 13:50 |
sanuj | lisitsyn, in any.h, i did this | 13:52 |
sanuj | bool equals(void** storage, void** other_storage) const | 13:52 |
sanuj | { | 13:52 |
sanuj | if(util::EqualExists<T>::value) { | 13:52 |
sanuj | T typed_storage = *(reinterpret_cast<T*>(*storage)); | 13:52 |
sanuj | T typed_other_storage = *(reinterpret_cast<T*>(*other_storage)); | 13:52 |
sanuj | return typed_storage == typed_other_storage; | 13:52 |
sanuj | } | 13:52 |
sanuj | return true; | 13:52 |
sanuj | } | 13:52 |
sanuj | where EqualExists checks if operator== is present | 13:52 |
lisitsyn | does it work? | 13:52 |
sanuj | no | 13:52 |
lisitsyn | I'd guess so | 13:52 |
lisitsyn | :D | 13:52 |
sanuj | yeah | 13:52 |
sanuj | gives same compiler error | 13:53 |
@wiking | lisitsyn: so question | 13:53 |
lisitsyn | quite obvious | 13:53 |
sanuj | lisitsyn, what exactly were you suggesting yesterday | 13:53 |
lisitsyn | wiking: what q? | 13:53 |
lisitsyn | sanuj: sfinae! | 13:54 |
lisitsyn | bool equals(void** storage, void** other_storage) const | 13:54 |
lisitsyn | this should be different when it comes to T | 13:55 |
lisitsyn | that doesn't support | 13:55 |
lisitsyn | == | 13:55 |
sanuj | lisitsyn, so i just make another "equals" implementation which just returns true ? | 13:55 |
sanuj | yeah | 13:55 |
lisitsyn | sanuj: oh this is quite dangerous | 13:55 |
lisitsyn | I'd prefer keeping it unimplemented | 13:55 |
sanuj | lisitsyn, you don't want to implement operator== for any? | 13:56 |
lisitsyn | sanuj: you'd need two declarations of operator== | 13:57 |
sanuj | or do i just declare a specialization of equals for std::function and not implement it? | 13:58 |
lisitsyn | sanuj: yeah rather that | 13:58 |
sanuj | but is this sfinae? | 13:58 |
lisitsyn | bool equals(void** storage, void** other_storage, bool supports_equal=util::EqualExists<T>::value) const | 13:59 |
lisitsyn | I'd suggest something like that | 13:59 |
sanuj | lisitsyn, how will i use the supports_equal value in the function equals? | 14:00 |
lisitsyn | sanuj: you don't use it | 14:01 |
lisitsyn | you just put default value there | 14:01 |
lisitsyn | so compiler chooses the right implementation | 14:01 |
sanuj | okay, let me try that | 14:02 |
sanuj | lisitsyn, is this function specialization based on its argument "supports_equal" rather than its template argument T? | 14:05 |
lisitsyn | yeap | 14:05 |
lisitsyn | or you can go with some other approach | 14:05 |
lisitsyn | not sure which is best here | 14:05 |
sanuj | lisitsyn, i think it's good to check for the existence for operator== | 14:06 |
sanuj | lisitsyn, but if we leave it undefined for the cases when operator== doesn't exist then how will it decide whether it's equal or not? | 14:07 |
@wiking | lisitsyn: so in order to be able to put the template implementation of the serialization function of SGObject into the header (that way no need to recompile the whole library if one comes up with a new Archive type | 14:08 |
@wiking | this requires self, i.e. the parameterMap to be defined in the header right? | 14:12 |
@wiking | :) | 14:12 |
@wiking | lisitsyn: ^ | 14:17 |
@wiking | lisitsyn: pingooooo | 14:37 |
sanuj | wiking, is it possible to do function specialization based on its argument "supports_equal" rather than its template argument T? | 14:41 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 15:00 | |
lisitsyn | wiking: sry | 15:04 |
-!- sanuj [~sanuj@117.204.242.64] has quit [Ping timeout: 264 seconds] | 15:14 | |
@wiking | lisitsyn: i stole your ::Self idea for MKL | 15:16 |
@wiking | care to look at it? | 15:16 |
@wiking | :D | 15:16 |
@wiking | lisitsyn: but yeah what do u say on the above | 15:16 |
-!- OXPHOS_ [c0a906af@gateway/web/freenode/ip.192.169.6.175] has quit [Ping timeout: 250 seconds] | 15:24 | |
@wiking | Saurabh7: ping | 15:30 |
lisitsyn | wiking: sry! | 15:41 |
lisitsyn | :D | 15:41 |
lisitsyn | interrupted all the time | 15:41 |
@wiking | lisitsyn: okok so read those when u can and react | 15:41 |
@wiking | (and comment on my pr) | 15:41 |
@wiking | :D | 15:41 |
@wiking | https://github.com/shogun-toolbox/shogun/pull/3396 | 15:42 |
@wiking | i dont remember when was my last time | 15:42 |
@wiking | that i sent a pr | 15:42 |
lisitsyn | wiking: SELF! | 15:48 |
lisitsyn | all the way | 15:48 |
lisitsyn | wiking: well this looks good | 15:49 |
-!- sanuj [~sanuj@117.204.242.64] has joined #shogun | 15:49 | |
lisitsyn | exactly what this is for | 15:49 |
@wiking | ok cool | 15:49 |
@wiking | i'll wait for travis | 15:49 |
@wiking | but then i'll merge | 15:49 |
@wiking | lisitsyn: but yeah any ideas about template<Archive>...? | 15:50 |
lisitsyn | wiking: yeah just read it but sorry don't really get the problem | 15:50 |
lisitsyn | is there some code? | 15:50 |
@wiking | ok so | 15:50 |
@wiking | lemme find the pr | 15:50 |
@wiking | ok look here | 15:52 |
@wiking | https://github.com/shogun-toolbox/shogun/pull/3375/files#diff-9c3599c0d2090e493be261b079e9b63eR523 | 15:52 |
@wiking | see the template<class Archive> void CSGObject::cereal_save(Archive& ar) const; is implemented in .cpp | 15:52 |
@wiking | so now this is a shit | 15:52 |
@wiking | if somebody comes with a new Archive format | 15:52 |
@wiking | not the 3 default ones | 15:53 |
@wiking | cereal::BinaryOutputArchive, cereal::JSONOutputArchive, cereal::XMLOutputArchive | 15:53 |
@wiking | because if you specify a new serialization format | 15:53 |
@wiking | then you have to recompile shogun | 15:53 |
@wiking | while if the templated save/load would be implemented in SGObject.h | 15:54 |
@wiking | see what i mean? | 15:54 |
lisitsyn | uh | 15:54 |
lisitsyn | so we want to support just any possible Archive? | 15:54 |
lisitsyn | I didn't know we do want it, now I get it | 15:54 |
lisitsyn | wiking: do we really want to support arbitrary Archives? | 15:55 |
@wiking | i mean why not? :) | 15:56 |
@wiking | i mean i dont care into what format the user wants to serialize | 15:56 |
@wiking | if he want to serialize directly into memcached or redis | 15:56 |
@wiking | i dont care | 15:56 |
@wiking | if he implements the Archive interface he can do it | 15:57 |
lisitsyn | ok I get it | 15:57 |
lisitsyn | wiking: ok we need to bridge the gap between templated archive | 15:57 |
lisitsyn | and hidden .cpp things | 15:57 |
lisitsyn | wiking: argh | 16:00 |
lisitsyn | this is tough | 16:00 |
lisitsyn | wiking: I am not really a fan of what kind of rape happened to 'any' | 16:02 |
lisitsyn | :D | 16:02 |
-!- shogun-notifier- [~irker@7nn.de] has joined #shogun | 16:05 | |
shogun-notifier- | shogun: OXPHOS :feature/cereal * b002f1e / src/shogun/base/SGObject.h,src/shogun/lib/any.h: https://github.com/shogun-toolbox/shogun/commit/b002f1ecffde34f87eff4657e391c289321927ee | 16:05 |
shogun-notifier- | shogun: update indentations and whitespace | 16:05 |
shogun-notifier- | shogun: Sergey Lisitsyn :feature/cereal * f95a0d7 / src/shogun/base/SGObject.h,src/shogun/lib/any.h: https://github.com/shogun-toolbox/shogun/commit/f95a0d73a2160d18ef0a8a6bed307972eda984a2 | 16:05 |
shogun-notifier- | shogun: Merge pull request #3393 from OXPHOS/cereal_style | 16:05 |
shogun-notifier- | shogun: | 16:05 |
-!- sanuj [~sanuj@117.204.242.64] has quit [Ping timeout: 240 seconds] | 16:05 | |
shogun-notifier- | shogun: update indentations and whitespace in SGObject and Any | 16:05 |
@wiking | :>>>> | 16:07 |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has joined #shogun | 16:08 | |
lisitsyn | wiking: there must be prettier way to do that :D | 16:08 |
@wiking | lisitsyn: mmmm | 16:10 |
shogun-notifier- | shogun: Viktor Gal :develop * efd291f / src/shogun/classifier/mkl/MKL.cpp,src/shogun/classifier/mkl/MKL.h: https://github.com/shogun-toolbox/shogun/commit/efd291f4ac41ac4324dafd336694527ad6831004 | 16:23 |
shogun-notifier- | shogun: Hide GPLK or CPLEX parameters | 16:23 |
shogun-notifier- | shogun: this way one will not require to supply the compiler flags | 16:23 |
shogun-notifier- | shogun: for GLPK or CPLEX when using MKL. | 16:23 |
shogun-notifier- | shogun: Viktor Gal :develop * 9ef9147 / src/shogun/classifier/mkl/MKL.cpp,src/shogun/classifier/mkl/MKL.h: https://github.com/shogun-toolbox/shogun/commit/9ef91473de6cb005bcdd0e824bd1f5b9547f66bb | 16:23 |
shogun-notifier- | shogun: Merge pull request #3396 from vigsterkr/MKLparams | 16:23 |
shogun-notifier- | shogun: | 16:23 |
shogun-notifier- | shogun: Hide GPLK or CPLEX parameters | 16:23 |
shogun-buildbot | build #1314 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1314 blamelist: Viktor Gal <viktor.gal@maeth.com> | 16:25 |
shogun-buildbot | build #1315 of precise - libshogun is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/precise%20-%20libshogun/builds/1315 blamelist: Viktor Gal <vigsterkr@gmail.com> | 16:32 |
shogun-buildbot | build #2957 of bsd1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/bsd1%20-%20libshogun/builds/2957 blamelist: Viktor Gal <vigsterkr@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 16:36 |
-!- sanuj [~sanuj@117.204.242.64] has joined #shogun | 16:59 | |
shogun-buildbot | build #43 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/43 blamelist: Viktor Gal <vigsterkr@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 17:19 |
-!- shogun-buildbot [~shogun-bu@7nn.de] has quit [Quit: buildmaster reconfigured: bot disconnecting] | 17:22 | |
-!- shogun-buildbot [~shogun-bu@7nn.de] has joined #shogun | 17:22 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has joined #shogun | 17:28 | |
shogun-buildbot | build #93 of xenial - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/xenial%20-%20libshogun/builds/93 blamelist: Viktor Gal <vigsterkr@gmail.com>, Viktor Gal <viktor.gal@maeth.com> | 17:29 |
-!- travis-ci [~travis-ci@ec2-54-158-129-129.compute-1.amazonaws.com] has joined #shogun | 17:36 | |
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/147486486 | 17:36 |
-!- travis-ci [~travis-ci@ec2-54-158-129-129.compute-1.amazonaws.com] has left #shogun [] | 17:36 | |
-!- OXPHOS [c0a906af@gateway/web/freenode/ip.192.169.6.175] has quit [Quit: Page closed] | 17:46 | |
c4goldsw | ping wiking | 18:14 |
sanuj | lisitsyn, there? | 18:39 |
-!- lambday [8028b10a@gateway/web/freenode/ip.128.40.177.10] has quit [Quit: Page closed] | 19:02 | |
-!- 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] | 19:04 | |
-!- c4goldsw [5da420e6@gateway/web/cgi-irc/kiwiirc.com/ip.93.164.32.230] has joined #shogun | 19:07 | |
CaBa | is there an implementation of any kind of weighted training in shogun? i.e. if i want to add a sort of confidence to a binary label? | 19:16 |
-!- shogun-notifier- [~irker@7nn.de] has quit [Quit: transmission timeout] | 19:23 | |
c4goldsw | Hey CaBa, try asking sanuj or Saurabh7 | 19:56 |
-!- sanuj [~sanuj@117.204.242.64] has quit [Ping timeout: 258 seconds] | 20:04 | |
-!- sanuj [~sanuj@59.97.245.21] has joined #shogun | 20:17 | |
-!- 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] | 20:24 | |
-!- sanuj [~sanuj@59.97.245.21] has quit [Remote host closed the connection] | 20:42 | |
--- Log closed Wed Jul 27 00:00:39 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!