IRC logs of #shogun for Wednesday, 2017-06-07

« Tuesday, 2017-06-06 Index Thursday, 2017-06-08 »
--- Log opened Wed Jun 07 00:00:02 2017
@sukeyPull Request #3836 "[ProgressBar] Replace SG_PROGRESS with the new progress bar."  merged by vigsterkr - https://github.com/shogun-toolbox/shogun/pull/383603:02
@sukeyNew Commit "Merge pull request #3836 from geektoni/replace_SG_PROGRESS03:02
@sukey[ProgressBar] Replace SG_PROGRESS with the new progress bar." to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/796f1db625c3301e955c41f23d2a91d3b098272103:03
-!- mikeling [uid89706@gateway/web/irccloud.com/x-mqhezswsrvtpsxpu] has joined #shogun03:43
@sukeyPull Request #3838 "Linalg library format update (clang-format-3.8)"  opened by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/383803:44
mikelingwiking: I found push_back wouldn't change the dim_size03:45
mikelinglike we push_back in here https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/mathematics/Integration.cpp#L334-L33503:46
mikelingand the display_element will only output    DynamicArray 'DynamicArray' of size:1x1x1 element(0,:,0) = [ -1.0, ]03:47
@wikingmmm03:50
@wikingi'm not so sure03:50
@wikingwhat you mean by not changing dim_size03:51
@wikingdo you change?03:51
@wikingi mean you just num_elements++ right?03:51
mikelingwiking: yep, it just push the element to the back of array https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/lib/DynamicArray.h#L33703:54
mikelingbut we display element as the dim_size https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/lib/DynamicArray.h#L556-L55703:55
@wikingyeah push_back on deque resizes as far as i remember right?03:55
mikelingyeah03:55
@wikingok but so if dim1 is not changed neither by you03:55
@wikingthen i dont get03:56
@wikinghow come it's different03:56
@wikingi mean it's clear to me03:56
@wikingjust a sec03:56
@wikingso in integrate03:56
@wikingthe first 3 arrays03:56
@wikingare03:56
@wikingDynamicArray 'DynamicArray' of size: 1x1x103:56
@wikingelement(0,:,0) = [ -1.0, ]03:57
mikelingok, I don't think it's the reason make them(use deque or don't),03:57
@wikingDynamicArray 'DynamicArray' of size: 1x1x103:57
@wikingelement(0,:,0) = [ -1.0, ]03:57
@wikingDynamicArray 'DynamicArray' of size: 1x1x103:57
@wikingelement(0,:,0) = [ 0.0, ]03:57
@wiking^ this is when you have the original code03:57
@wikingand this is with your patches03:57
@wikingDynamicArray 'DynamicArray' of size: 1x1x103:57
@wikingelement(0,:,0) = [ -1.0, ]03:57
@wikingDynamicArray 'DynamicArray' of size: 1x1x103:57
@wikingelement(0,:,0) = [ -1.0, ]03:57
@wikingDynamicArray 'DynamicArray' of size: 10x1x103:57
@wikingelement(0,:,0) = [ 0.0, ]03:57
@wiking...03:57
mikelingyep03:57
@wikingso03:57
mikelingI know03:57
@wikingthere's a difference right?03:57
mikelingyep03:57
@wikingis it coming from the constructor?03:57
mikelingI got same output locally03:58
mikelingyes, I guess the size changed in set_array()03:58
@wiking?03:59
@wikingbut i mean03:59
@wikinglets step back one step here03:59
@wikingwhat are we trying to do right?03:59
@wikingmake CDynamicArray to use std::deque03:59
@wikingright?03:59
mikelingyep03:59
@wikingi'm not so sure if i understand03:59
@wikingwhy is it so hard03:59
@wikingi mean you did change SO many things (constructors, setters etc)04:00
@wikingin your PR04:00
@wikingbut couldn't we first just change DynArray to be deque04:00
@wikingand just change minimally the api?04:00
@wikingi mean you are working on this04:00
@wikingfor at least 3 days right?04:00
@wikingi really dont understand why is it so complicated04:00
mikelingyes :/04:00
@wikingi mean it's just using a different container04:01
@wikingnothing more right?04:01
mikelingyes, you're right.04:01
mikelingBut things just don't work as I expected :/04:02
@wikingi mean yes04:02
@wikingbecause you just change way too many things04:02
@wikingi mean you have significantly changed04:03
@wikingthe public API of DynamicArray04:03
@wikingand one example of that is the set_array04:03
@wikingyou dropped some of them04:03
@wikingyou changed the arguments04:03
@wikingadded default values04:03
@wikingi mean i dont understand why04:03
@wikingwhy do you need to do those things?04:03
@wikingwhy cannot we just keep the public api as is04:03
@wikingand just change the private members04:04
mikelingjust want to simplify it04:04
mikelingbut04:04
mikelingyou are right04:04
@wikinglike this04:05
@wikinghttps://github.com/shogun-toolbox/shogun/pull/3832/files#diff-1a43b9169d617ac582b2a21332230795R7004:05
@wikingi mean you remove some elements from the constructor04:05
@wikingthis will act totally different04:05
@wikingthen the original implementation04:06
@wikingso i mean what i would do is04:06
mikelingbecause we don't need p_free_array or p_copy_array anymore04:06
@wikingwhy?04:06
@wikingi mean04:06
@wikingi dont understand why?04:06
@wikingi mean if you get a reference to a memory area and you have p_copy_array = true04:06
@wikingthen you actually have to copy that memory area04:06
@wikingand make that memory to be wrapped by deque04:07
@wikingagain04:07
@wikingi would not touch any of the public api04:07
@wikingi would keep it as is04:07
@wikingjust change from DynArray to deque04:07
@wikingand make sure that using unit tests04:07
@wikingthey are still the same04:07
@wikingthere are many many04:07
@wikingunit tests missing04:07
@wikingfor CDynamicAray04:08
mikelingkk, I agree. I will reset it and change the things step by step04:08
@wikingsuper simple to write04:08
@wikingtest each and every ctor04:08
@wikingmake sure it copies04:08
@wikingif it needs copying04:08
@wikingetc etc etc etc04:08
@wikingand then when you have all the public api04:08
@wikingALL04:08
@wikingnot just some tested04:08
@wikingthen you change the private member of CDynamicArray04:08
@wikingand start seeing the errors04:08
@wikingthen you fix one by one04:09
@wikingbut not BY removing public api04:09
@wikingor changing it totally04:09
@wikingno04:09
@wikingjust under the hood04:09
@wikingto keep the functionality the same04:09
@sukeyPull Request #3838 "Linalg library format update (clang-format-3.8)"  closed by OXPHOS - https://github.com/shogun-toolbox/shogun/pull/383804:24
shogun-buildbotbuild #76 of nightly_all is complete: Failure [failed test]  Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/76  blamelist: Michele Mazzoni <mic.mazzoni@gmail.com>, Viktor Gal <vigsterkr@gmail.com>07:58
-!- geektoni [~geektoni@93-34-234-212.ip52.fastwebnet.it] has joined #shogun09:01
-!- geektoni [~geektoni@93-34-234-212.ip52.fastwebnet.it] has left #shogun []09:33
-!- mikeling [uid89706@gateway/web/irccloud.com/x-mqhezswsrvtpsxpu] has quit [Quit: Connection closed for inactivity]11:02
-!- mikeling [uid89706@gateway/web/irccloud.com/x-uzalzmesuwfseior] has joined #shogun12:30
-!- olinguyen [81615ad9@gateway/web/freenode/ip.129.97.90.217] has joined #shogun16:06
@wikingmikeling, sup?16:13
mikelingwiking: sorry, I'm rebuilding the project. I will update the pr16:23
mikelingafter I finish it16:23
mikelingwithin today :)16:23
@wiking:)16:31
@wikingkk16:31
@wikingjust asking if you need any help16:32
mikelingwiking: thank you, I only have SplittingStrategy.standard test failed right now16:37
mikelingI'm working on it16:37
@wikingsounds good16:37
@wikinggood luck16:37
mikelingthank you :)16:38
-!- OXPHOS [92bd305b@gateway/web/freenode/ip.146.189.48.91] has joined #shogun17:56
OXPHOSwiking: there?17:57
@wiking yes18:01
@wikingbarely18:01
@wikingbut eyes18:01
OXPHOSoh me too. I'll pour.18:02
OXPHOSlinalg is failing vs2017 compiling consistently now. The error is C1060: out of heap space18:03
OXPHOSI looked up and people seem to think that too many templates is the problem18:03
OXPHOSsomeone suggests to change the precompiled header memory allocation limit)18:04
OXPHOS-zm18:04
OXPHOSdo you think this is a good solution?18:04
OXPHOSoh sorry on appveyor18:05
OXPHOSBut Heiko mentioned he had some ideas about linalg lib last week. Haven't heard from him yet.18:05
OXPHOSNot sure what he wants to change18:06
OXPHOSwiking ^18:06
@wikingmmm18:06
@wiking"precompiled header memory allocation limit"18:06
@wikinghow? :)18:06
OXPHOShttps://msdn.microsoft.com/en-us/library/bdscwf1c.aspx18:07
OXPHOSzm 1018:07
OXPHOSdefault is 10018:08
OXPHOSi.e. 75MB18:08
OXPHOSWe should be able to change build config somewhere..18:08
OXPHOSon appveyor18:08
micmnwiking, OXPHOS: A.col(i) = B.col(j) with sgmatrix/linalg?18:33
OXPHOSmicmn: sorry I didn't get it. What are you trying to say??19:04
@sukeyPull Request #3832 "use std::deque instead of DynArray(on going)"  synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/383219:19
@sukeyPull Request #3832 "use std::deque instead of DynArray(on going)"  synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/383219:23
mikelingwiking: ping19:23
mikelingstill aroung? :)19:23
micmnOXPHOS: sorry :p the corresponding eigen3 operation, put the content of the j-th column of B into the i-th column of A19:33
OXPHOSmicmn: you want to have the method in linalg library? Because SGMatrix doens't have col(i) now19:38
OXPHOSalso I feel like it should be in sgmatrix, not linalg lib19:38
micmnyeah, I wanted to know it there is a way of doing that without eigen19:39
OXPHOSmicmn: only this19:47
OXPHOShttps://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/lib/SGMatrix.h#L14019:47
OXPHOSI didn't see the downside having a set_column_vector() in sgmatrix @wiking19:47
micmnI mean, the purpose of linalg is to abstract away eigen (or other backends), am I getting it wrong? :D19:50
OXPHOSoh. we used to have native backends and now the linalg lib SHOULD allow us to use whatever backends we want, CPU and GPU-wise. But currently eigen is the only available CPU backend, and it is a hard requirement19:53
OXPHOSComputation-wise, linalg ops SHOULD for into one of the backends, eigen or viennacl or others19:54
OXPHOSAm i getting u right?19:55
OXPHOSfall*19:55
micmnyep :)19:56
OXPHOSso either have the method in sgmatrix and use raw pointers, or realize it in linalg lib with eigen, and other backends in the future19:56
-!- mikeling [uid89706@gateway/web/irccloud.com/x-uzalzmesuwfseior] has quit [Quit: Connection closed for inactivity]21:39
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds]23:31
-!- besser82_ [~besser82@fedora/besser82] has joined #shogun23:31
-!- mode/#shogun [+o besser82_] by ChanServ23:32
-!- besser82_ is now known as besser8223:32
--- Log closed Thu Jun 08 00:00:04 2017
« Tuesday, 2017-06-06 Index Thursday, 2017-06-08 »

Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!