mlpack IRC logs, 2020-02-05
Logs for the day 2020-02-05 (starts at 0:00 UTC) are shown below.
--- Log opened Wed Feb 05 00:00:48 2020
00:58 -!- surfza2004 [~firstname.lastname@example.org] has joined #mlpack
00:58 -!- surfza2004 [~email@example.com] has quit [Quit: Leaving]
01:54 < kartikdutt18Gitt> Hi @zoq, what changes should I implement in #2163 ?
01:56 < kartikdutt18Gitt> Hi everyone, I am seeking reviews for the PR #2163, any suggestions are welcome. Thanks.
03:00 < PrinceGuptaGitte> Hi everyone, I'm trying to contribute to mlpack. Are there any issues I can help with.
03:02 < kartikdutt18Gitt> Hi @prince776, this might be useful https://www.mlpack.org/gsoc.html to get started. You can also see issue page on mlpack repo and select "good-first-issues".
03:05 < PrinceGuptaGitte> Thanks, good-first-issues is what I was looking for. I have read gsoc and community page already.
03:05 < kartikdutt18Gitt> Great.
07:57 -!- Ojas [firstname.lastname@example.org] has joined #mlpack
07:58 -!- Ojas [email@example.com] has quit [Remote host closed the connection]
08:19 < jenkins-mlpack2> Yippee, build fixed!
08:19 < jenkins-mlpack2> Project docker mlpack nightly build build #604: FIXED in 3 hr 5 min: http://ci.mlpack.org/job/docker%20mlpack%20nightly%20build/604/
14:55 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has joined #mlpack
15:33 -!- mdjordjic [firstname.lastname@example.org] has joined #mlpack
15:33 < mdjordjic> hi
15:33 < mdjordjic> is there a valid way to install mlpack on W10
15:34 < mdjordjic> or to use it in another words to put it
15:34 < mdjordjic> no success so far with VS2017 and VS2019
15:35 < zoq> mdjordjic: Hello, interesting, I guess you followed the instructions as mentioned on the getstarted page?
15:35 < mdjordjic> yes
15:36 < mdjordjic> the problem is you go through the builds
15:36 < mdjordjic> and then you finally need to build mlpack
15:36 < mdjordjic> like the last step
15:36 < mdjordjic> and a bunch of fatal errors will emerge
15:36 < mdjordjic> cannot find this
15:36 < mdjordjic> cannot find that
15:36 < mdjordjic> eventually you end up with
15:37 < mdjordjic> 4 out 54 sucesfull builds
15:37 < mdjordjic> or some other number
15:37 < mdjordjic> i did not see an offical instruction
15:38 < mdjordjic> but its possible to 'install' via NuGet Package Manager
15:38 < mdjordjic> i dont know if that is officially supported
15:38 < mdjordjic> but this failed as well
15:38 < zoq> mdjordjic: Any chance you might be able to post those errors (pastebin)?
15:38 < mdjordjic> VS2015
15:38 < mdjordjic> lets try with this fwith the NuGet
15:39 < mdjordjic> im installing via NuGet
15:40 < mdjordjic> so i will have errors quickly
15:41 < PrinceGuptaGitte> Installing by vcpkg worked for me when I was using Windows 10, you might wanna try that
15:41 < mdjordjic> that failed also
15:41 < mdjordjic> ie. after you run it
15:41 < mdjordjic> i left computer over night
15:41 < mdjordjic> and it halted at some point
15:41 < mdjordjic> installing some dependancy did not work
15:42 < mdjordjic> in theory i have something now
15:43 < mdjordjic> Severity Code Description Project File Line Suppression State
15:43 < mdjordjic> std::basic_string<char,struct std::char_traits<char>,class std::allocator<char> > const &,class arma::Mat<double> const &,bool,bool)" (??$Save@N@data@mlpack@@YA_NABV?$basic_string@DU?$char_traits@D@std@@V?$allocator@D@2@@std@@ABV?$Mat@N@arma@@_N_N@Z) ConsoleApplication6 c:\Users\TMP2UFE\documents\visual studio
15:43 < mdjordjic> 2015\Projects\ConsoleApplication6\ConsoleApplication6\Source.obj 1
15:54 -!- mdjordjic [email@example.com] has quit [Remote host closed the connection]
16:08 < HimanshuPathakGi> I think you compiler was not able to link boost library
16:09 < HimanshuPathakGi> (edited) ... think you compiler ... => ... think your compiler ...
16:09 < HimanshuPathakGi> mdjordjic have you installed boost on vs
16:15 < kartikdutt18Gitt> Hi @zoq, could you help me out with this https://github.com/mlpack/mlpack/pull/2172/files#r375355745 .
16:43 < PrinceGuptaGitte> Hi, I am working on issue #2084 , I have done it for data::load() aspect but for that I had to build it 3-4 times and each time it takes a lot of time. Is there a better way to work? Like building specific files separately to main build ?
16:47 < HimanshuPathakGi> Use -DBUILD_CLI_EXECUTABLES=OFF -DBUILD_PYTHON_BINDINGS=OFF -DBUILD_JULIA_BINDINGS=OFF while using cmake
16:47 < PrinceGuptaGitte> what that does?
16:47 < PrinceGuptaGitte> doesn't build python bindings?
16:48 < kartikdutt18Gitt> @prince776 you can build certain files rather building everython
16:48 < PrinceGuptaGitte> oh thanks I'll try it
16:48 < kartikdutt18Gitt> Everything, like make test -j2
16:49 < HimanshuPathakGi> Python binding helps us to use mlpack in python
16:49 < kartikdutt18Gitt> Yes as @himanshupathak21061998 pointed certain things aren't necessary for certain issues and we can use make to disable them.
16:50 < PrinceGuptaGitte> make: Nothing to be done for 'load_impl.hpp'. , it didn't build it
16:50 < kartikdutt18Gitt> Only certia
16:50 < kartikdutt18Gitt> only certain files/ folders can be built that way.
16:50 < HimanshuPathakGi> Hey zoq I think we should add this as warning in documentation for doing a faster build or anything else
16:51 < HimanshuPathakGi> @prince776
16:51 < PrinceGuptaGitte> Sorry if I'm asking too much questions(I'm kindda new to this env) but how do I build a certain file say load_impl.hpp after making changes to it
16:52 < HimanshuPathakGi> You need to specify target for this
16:53 < kartikdutt18Gitt> Not an issue @prince776 , Ask as many as needed, people here are happy to help 😊.
16:54 < PrinceGuptaGitte> When i specify target like : make load_impl , it says : *** No rule to make target 'load_impl'. Stop.
16:56 < kartikdutt18Gitt> @prince776 I don't think there is build file for that.
17:01 < PrinceGuptaGitte> Apparently whenever I make changes, if i call "sudo make install" in build directory it builds only the changed parts and changes are exhibited.
17:01 < PrinceGuptaGitte> I guess this makes sense
17:02 < rcurtin> oh no, I forgot about the video meet-up. the next one would be tomorrow at 1800 UTC (25 hours from now) though so I'll just send an email now :)
17:09 < rcurtin> mdjordjic: if you prefer, you can open an issue on Github; for the large number of errors you are seeing, it might be easier to manage like that :)
17:09 < rcurtin> up to you though of course
17:10 < rcurtin> also, just a note, it turns out that the datacenter hosting mlpack.org will be closing at the end of February. so I'll have to run it across town to a new datacenter; I think we can keep the downtime to only a couple hours, and maybe we can set up some simple backups to keep various mlpack services online
17:10 < rcurtin> I'll provide a bit more info as we get closer to the move... right now I am still calling different datacenters to see what's available
17:11 < rcurtin> (this is just for mlpack.org, none of the build servers or anything)
17:11 < kartikdutt18Gitt> Hi @rcurtin I finally fixed the benchmarks on #2178, if you get the chance kindly have a look. Thanks.
17:12 < rcurtin> when I have a chance I will, I'm pretty overloaded at the moment. it will take a little while to carefully review the benchmarking code, etc.; benchmarking is not easy to get right :)
17:13 < kartikdutt18Gitt> Sure Thanks, yes it is not that easy, I will try to come up with a graph for the same.Thanks
17:13 < rcurtin> sounds good, thanks
17:14 < rcurtin> I am *really* behind on PRs and mlpack right now, it's a bit overwhelming
17:14 < rcurtin> I've only been able to spare a minute here or there over the past week or two, and next week I'm on vacation... :(
17:15 < kartikdutt18Gitt> I understand, we need more reviewers.
17:16 < PrinceGuptaGitte> Just a while ago "sudo make install" only built the files which were changed(so build process was very fast) but now it started building all files again.
17:16 < kartikdutt18Gitt> Thanks for all help though, personally in pointing out errors in my benchmarking.
17:17 < rcurtin> kartikdutt18Gitt: no problem :)
17:17 < rcurtin> PrinceGuptaGitte: you should only need to do 'make' to rebuild, and CMake will build everything that depends on the files that you changed
17:18 < rcurtin> so, e.g., if you changed src/mlpack/core.hpp, it would need to rebuild just about everything
17:35 -!- UmarJ [~UmarJ@184.108.40.206] has quit [Quit: UmarJ]
17:35 -!- UmarJ [~UmarJ@220.127.116.11] has joined #mlpack
17:52 < PrinceGuptaGitte> For issue #2084 (Add manual type specification support to `data::Load()` and `data::Save()`)
17:52 < PrinceGuptaGitte> When we specify extension should the filename given to parameter be still like ("data.csv")?
17:53 < PrinceGuptaGitte> As in what should be the way it will be used :
17:53 < PrinceGuptaGitte> `data::load("data", armaMat , true , true, "csv")`
17:53 < PrinceGuptaGitte> or
17:54 < PrinceGuptaGitte> `data::load("data.csv", armaMat , true, true, "csv")`
18:01 < rcurtin> PrinceGuptaGitte: have you seen PR #2135? it's linked to in the issue comments
18:12 < PrinceGuptaGitte> Oh I didn't see that. I'm gonna see it to understand where to proceed and what to do. Thanks for pointing it out
18:51 < rcurtin> what I'm saying is that there is already a PR open, so perhaps it would be better to try and help review and improve that one instead of starting over
19:14 -!- ImQ009 [~ImQ009@unaffiliated/imq009] has quit [Quit: Leaving]
20:15 < birm[m]> How do we usually reassign stale PRs anyway? The new person probably won't have write access to the author's branch.
20:19 < zoq> birm[m]: No official process in place, if anybody likes to work on a stale PR, just clone the branch (or cherry pick the commits) and open a new PR.
--- Log closed Thu Feb 06 00:00:49 2020