--- Log opened Wed May 24 00:00:43 2017 | ||
-!- witness_ [uid10044@gateway/web/irccloud.com/x-nowpvqgxvtbfunme] has quit [Quit: Connection closed for inactivity] | 01:04 | |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-hqpdwwbckniyzchm] has joined #shogun | 01:40 | |
@sukey | Pull Request #3819 "make guassian get rid of LAPACK" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3819 | 03:55 |
---|---|---|
@sukey | Pull Request #3819 "make guassian get rid of LAPACK" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3819 | 04:02 |
-!- mikeling|afk is now known as mikeling | 04:04 | |
mikeling | wiking: ping | 04:11 |
-!- witness_ [uid10044@gateway/web/irccloud.com/x-hqpdwwbckniyzchm] has quit [Quit: Connection closed for inactivity] | 04:14 | |
mikeling | lisitsyn: ping | 04:35 |
@wiking | pong | 07:12 |
mikeling | wiking: Hi, do you have time to review this https://github.com/shogun-toolbox/shogun/pull/3819? | 07:51 |
@wiking | yeah let's ahve a chat in 5 minutes | 07:51 |
@wiking | *have | 07:51 |
mikeling | sure | 07:51 |
@wiking | just a sec | 08:03 |
mikeling | np :) | 08:03 |
@wiking | k here | 08:12 |
@wiking | so i'm just checking something | 08:12 |
mikeling | ok | 08:12 |
@wiking | so | 08:13 |
@wiking | we have this ancient wrapper for some lapack functionallities | 08:13 |
@wiking | ./src/shogun/mathematics/lapack.h | 08:13 |
mikeling | yep | 08:14 |
@wiking | so i'm just wondering whether we should add something there | 08:16 |
@wiking | we could do a big | 08:16 |
@wiking | #endif //HAVE_LAPACK | 08:16 |
@wiking | #else | 08:17 |
@wiking | and there add some functionalities | 08:17 |
@wiking | but of course | 08:17 |
@wiking | this would mean that if somebody doesn't have LAPACK | 08:17 |
@wiking | mmm | 08:17 |
@wiking | ok no this is not the best idea | 08:17 |
@wiking | sorry | 08:17 |
@wiking | ok them i'm just thinking about the naming of the | 08:17 |
@wiking | function | 08:17 |
@wiking | the problem with these 4 letter namings of a function | 08:18 |
mikeling | yep | 08:18 |
@wiking | is that it's completely jibberish | 08:18 |
@wiking | when you code | 08:18 |
mikeling | sorry, what's the jibberish? | 08:18 |
@wiking | cblas_dgemv | 08:19 |
@wiking | or vvmp | 08:19 |
@wiking | btw | 08:19 |
@wiking | lisitsyn, ping | 08:19 |
@wiking | so i mean | 08:20 |
@wiking | this thing with | 08:20 |
mikeling | ah I see. | 08:20 |
mikeling | the vvmp just means *vector produce vector then plus matrix* which that function does | 08:21 |
@wiking | yes | 08:22 |
@wiking | i know but again | 08:22 |
@wiking | think about when you are coding | 08:22 |
@wiking | this is not intuitive | 08:22 |
@wiking | just as dgemv | 08:22 |
@wiking | i mean lapack function has a rule as well | 08:22 |
@wiking | for naming | 08:22 |
@wiking | the first character is f = float or d = double | 08:22 |
mikeling | mmm I have no idea about the rule. | 08:23 |
mikeling | so you mean the dgemv is not some combination of random letters | 08:24 |
mikeling | it do have pertinent meaning for that function | 08:25 |
mikeling | maybe we should use 'dgemv'? Just remove the cblas_ prefix? | 08:26 |
@wiking | nono | 08:27 |
@wiking | it's horrible | 08:27 |
@wiking | i mean i hate these naming | 08:27 |
@wiking | mikeling, http://www.netlib.org/lapack/lug/node24.html | 08:27 |
@wiking | there's the rule | 08:27 |
@wiking | but it's horrible | 08:27 |
@wiking | just a sec | 08:30 |
@wiking | iin the meanwhile i have found some errors :) | 08:30 |
@wiking | minor one | 08:30 |
@wiking | just commenting it into the pr | 08:30 |
mikeling | alright, it's complex and I guess people won't get the point when they first look at the code | 08:31 |
mikeling | ok | 08:31 |
mikeling | thank you :) | 08:31 |
@wiking | ok i've sent you the changes you need for the doxygen for sure | 08:33 |
@wiking | ok so you fix that part | 08:42 |
@wiking | i'm just thinking now of the naming | 08:42 |
@wiking | for that those function | 08:42 |
-!- iglesiasg [~iglesiasg@217.119.234.214] has joined #shogun | 09:06 | |
-!- mode/#shogun [+o iglesiasg] by ChanServ | 09:06 | |
-!- geektoni [c1cdd248@gateway/web/freenode/ip.193.205.210.72] has joined #shogun | 09:28 | |
@wiking | geektoni, you've pinged me yesterday? | 09:32 |
geektoni | hey wiking :) yeah, but I've already solved the problem. | 09:32 |
@wiking | kk | 09:32 |
lisitsyn | wiking: | 10:01 |
lisitsyn | iglesiasg: hola senor | 10:01 |
@iglesiasg | lisitsyn: que pasa | 10:03 |
@wiking | hi | 10:07 |
@wiking | lisitsyn, btw | 10:24 |
@wiking | iglesiasg, as well | 10:24 |
@wiking | any ideas how we should name these functions internally | 10:24 |
@wiking | cblas_dgemv | 10:24 |
@wiking | cblas_dger | 10:24 |
@wiking | cblas_dgemm | 10:24 |
@wiking | cblas_dgemv is kind of a shift and scale of a Y vector | 10:24 |
@wiking | since one could take that A matrix is transforming the X vector into Y's space.... | 10:25 |
@wiking | but yeah need better naming than dgemv | 10:25 |
@wiking | :) | 10:25 |
lisitsyn | at least without cblas | 10:28 |
lisitsyn | because we call eigen there right? | 10:28 |
@wiking | linalg :) | 10:28 |
@wiking | yeah of course we'll drop cblas_ | 10:29 |
@wiking | but would be great if we could have a more intuitive naming | 10:29 |
@wiking | than zxvaseY | 10:29 |
@iglesiasg | eigen-like naming? | 10:38 |
@iglesiasg | ShiftAndScale | 10:39 |
@iglesiasg | AndDifferentiateAndNormalizeAndShiftAgain | 10:39 |
lisitsyn | FcukThenUnfcukItAgain | 10:40 |
lisitsyn | scnr | 10:40 |
@iglesiasg | hahaha | 10:41 |
@iglesiasg | jajaja | 10:41 |
@wiking | : | 10:46 |
@wiking | :> | 10:46 |
@wiking | but yeah we need something | 10:46 |
@wiking | now | 10:46 |
@wiking | :) | 10:46 |
lisitsyn | wiking: I am fine with dgemv | 10:47 |
lisitsyn | just without cblas | 10:47 |
lisitsyn | I think these guys had some wisdom | 10:48 |
lisitsyn | I guess they had these_long_names and then decided to go with random sequences | 10:48 |
lisitsyn | at least it is googlable they said in 1970 :P | 10:49 |
@wiking | :> | 10:53 |
@wiking | it's not random :) | 10:53 |
@wiking | mikeling, ok now you should have all the righ tinfo for that pr and you can finish it | 10:55 |
@wiking | and then we can go back to the original PR ) | 10:55 |
lisitsyn | wiking: youve been tricked | 10:56 |
@wiking | ? | 10:56 |
lisitsyn | it was random then they have come with reasonable explanation | 10:57 |
lisitsyn | djfsjfsfj ok double integer field …. | 10:57 |
lisitsyn | :P | 10:57 |
@wiking | :) | 10:57 |
mikeling | so we are going to use degmv? Just remove the cblas prefix | 11:01 |
@wiking | yes | 11:07 |
@wiking | indeed | 11:07 |
-!- geektoni [c1cdd248@gateway/web/freenode/ip.193.205.210.72] has quit [Quit: Page closed] | 11:58 | |
@iglesiasg | do you guys know of any toolset for gdb that can do stuff like giving the max of an array? | 12:12 |
@iglesiasg | I am debugging some code and have a long float array[len]. From gdb, I would like to do something like p max(array, len). | 12:13 |
lisitsyn | I don't | 12:18 |
@sukey | Pull Request #3819 "make guassian get rid of LAPACK" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3819 | 13:59 |
-!- geektoni [c1cdd24a@gateway/web/freenode/ip.193.205.210.74] has joined #shogun | 14:05 | |
@wiking | anybody around? | 14:46 |
@iglesiasg | .o/ | 15:04 |
@iglesiasg | wiking: tell me | 15:04 |
@wiking | already fixed | 15:04 |
@wiking | but thnx | 15:04 |
@iglesiasg | </3 | 15:04 |
mikeling | wiking: mmm, it seems like the appveyor went wrong https://ci.appveyor.com/project/vigsterkr/shogun/build/4.2.763/job/ae4304tcat74e037 | 16:04 |
mikeling | but no error comes out | 16:05 |
mikeling | :/ | 16:05 |
mikeling | seems like the ci just run out of time and stopped. Maybe we could assume the code is good to go? | 16:10 |
-!- iglesiasg [~iglesiasg@217.119.234.214] has quit [Quit: leaving] | 16:59 | |
mikeling | lisitsyn: How could I retrigger ci if it accidentally error out? e.g https://ci.appveyor.com/project/vigsterkr/shogun/build/4.2.763/job/ae4304tcat74e037 | 17:10 |
@wiking | ill do it now | 17:13 |
@sukey | Pull Request #3819 "make guassian get rid of LAPACK" merged by vigsterkr - https://github.com/shogun-toolbox/shogun/pull/3819 | 17:14 |
@sukey | New Commit "Merge pull request #3819 from MikeLing/get_rid_of_LAPACK | 17:14 |
@sukey | make guassian get rid of LAPACK" to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/380c06f6d8ac74635bcca712e46c34bdd2d9c80c | 17:14 |
@wiking | done | 17:14 |
mikeling | wiking: Thank you, I'm going to update related pr | 17:16 |
-!- geektoni [c1cdd24a@gateway/web/freenode/ip.193.205.210.74] has quit [Quit: Page closed] | 17:18 | |
@sukey | Pull Request #3821 "Add new constructor and get() method to Some<>." merged by vigsterkr - https://github.com/shogun-toolbox/shogun/pull/3821 | 18:21 |
@sukey | New Commit "Merge pull request #3821 from geektoni/patch-18 | 18:21 |
@sukey | Add new constructor and get() method to Some<>." to shogun-toolbox/shogun by vigsterkr: https://github.com/shogun-toolbox/shogun/commit/c2a9559dc2664c75649f913ab1bc72caa1a3e5a8 | 18:21 |
@sukey | Pull Request #3812 "Add global fixture with binary label data" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3812 | 18:24 |
@sukey | Pull Request #3812 "Add global fixture with binary label data" synchronized by MikeLing - https://github.com/shogun-toolbox/shogun/pull/3812 | 18:46 |
-!- mikeling [uid89706@gateway/web/irccloud.com/x-cohfsjksvtvnffvx] has quit [Quit: Connection closed for inactivity] | 21:10 | |
-!- micmn_ [~micmn@ec2-54-67-24-92.us-west-1.compute.amazonaws.com] has joined #shogun | 21:36 | |
-!- micmn [~micmn@ec2-54-67-24-92.us-west-1.compute.amazonaws.com] has quit [Ping timeout: 246 seconds] | 21:36 | |
--- Log closed Thu May 25 00:00:44 2017 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!