13:02 < NippunSharmaGitt> Hi, I had a doubt in the [convolution.hpp](https://github.com/mlpack/mlpack/blob/master/src/mlpack/methods/ann/layer/convolution.hpp). Here `inputHeight` and `inputWidth` refer to the shape of input but what do we mean by `inSize` ? Does it refer to the number of feature maps ?

14:22 < AlexNguyen[m]> <iamarchit123Gitt "Hey folks small query here how "> I think if you are testing, you have to change the `ignoreInput` flag to False in the main test file. I.e, `Log::Info::IgnoreInput = False`

15:37 < tobelegitGitter[> Hey everyone. I was looking forward to contributing to mlpack, however, I am unable to get started. Any help?

15:38 < zoq> tobelegitGitter[: Hello there, have you seen the getting involved section - https://www.mlpack.org/community.html

15:38 < zoq> tobelegitGitter[: https://www.mlpack.org/gsoc.html has some additional information as well.

15:42 < zoq> NippunSharmaGitt: Yes, the parameter description uses the term "input maps" -> "@param inSize The number of input maps."

15:47 < tobelegitGitter[> Yes, I read it but am not sure of the steps of contributing exactly.

16:09 < NippunSharmaGitt> Thanks @zoq . So the total input shape for layer would be `inputHeight * inputWidth * inSize` ?

16:09 < NippunSharmaGitt> (edited) ... be `inputHeight * inputWidth * inSize` ? => ... be `inputHeight, inputWidth,inSize` ?

16:09 < NippunSharmaGitt> (edited) ... be `inputHeight, inputWidth,inSize` ? => ... be `(inputHeight, inputWidth,inSize)` ?

16:12 < zoq> NippunSharmaGitt: correct

16:13 < zoq> tobelegitGitter[: Well the first step is to build mlpack, and checkout the tutorials.

16:13 < zoq> tobelegitGitter[: After that you can checkout the issues on GitHub and see if there is something you can contribute.

16:14 < NippunSharmaGitt> thanks. whenever you have time can you please take a look at #2706 ?

16:15 < zoq> NippunSharmaGitt: Sure

19:40 < rcurtin> zoq: do I have to do anything special for PVS-Studio now?

19:40 < rcurtin> "Error: License was not entered. Please enter a valid PVS-Studio license or request a trial license"

19:40 < rcurtin> from http://ci.mlpack.org/job/pull-requests-mlpack-static-code-analysis/7145/console

19:41 < rcurtin> I'm not sure if you also got a license for pioneer and polar

19:46 < zoq> rcurtin: I think I saw the same issue on the other nodes as well, I take a closer look into the issue.

19:46 < rcurtin> maybe they changed their licensing or something like that

19:49 < zoq> rcurtin: yep, let use the offical repo, instead of my fork.

19:49 < rcurtin> ok, if you can let me know what you run on polar and pioneer to fix it, I will do the same on knife, pretzel, and bigglesworth :)

19:50 < rcurtin> I am still negotiating with Georgia Tech to get dealgood back online, by the way

19:50 < rcurtin> all I need them to let me do is come to the datacenter and then work on the system, but because of COVID restrictions the IT team barely comes to campus anymore, so it's hard to figure out a time

19:50 < zoq> I just updated the jenkins job config, to use the viva64 repo.

19:51 < rcurtin> oh, nice, I thought we had installed packages on the system through apt

19:51 < rcurtin> I guess I forgot how it was set up

19:51 < zoq> Maybe they like it and never go back?

19:52 < zoq> I guess you can do a lot from home, if there is no hardware failure.

19:52 < rcurtin> :)

19:52 < rcurtin> yeah, you can avoid atlanta traffic when working from home, which is nice

19:52 < rcurtin> but in this case dealgood actually does have a hardware failure so it needs an in-person visit :)

19:53 < MattGomes> rcurtin Managed to get the build working on Visual Studio. I had some really bad time trying to compile OpenBLAS/Lapack with mingw. The only solution that worked was to build both with `clang-cl` and `flang`, with a few other dependencies pulled with Conda. Aside from that, there were a few tweaks to the root CMake to link some libraries needed by

19:53 < MattGomes> the `OpenBLAS` lib (mainly fortran ones). After all that, Visual Studio could fully use the CMake integration to build the full project!

19:53 < rcurtin> dealgood has an attached Apple Xserve RAID (https://en.wikipedia.org/wiki/Xserve_RAID), and a drive in that seems to have failed

19:54 < rcurtin> MattGomes: awesome! if you have any ideas of how we can improve our documentation to make the process easier for others in the future, please feel free to open a PR

19:54 < rcurtin> very few of us are Windows experts (I am definitely not very knowledgeable about Windows builds), so I think our Windows documentation can be pretty imperfect at times

19:54 < zoq> rcurtin: And what is proposed fix, to remove the Xserver?

19:54 < MattGomes> I could write a visual studio guide, might create a PR this week when I have some time from work.

19:54 < rcurtin> zoq: yeah, I'll just remove it; there's no need for the Xserve storage when all dealgood is is a Jenkins slave

19:55 < rcurtin> MattGomes: sounds good, thank you!

19:57 < MattGomes> rcurtin I was only able to build with VS as the CMake is quite well written and simple. Most open source projects don't really write CMake that is compatible with Windows, unfortunately. Kudos to you guys.

19:57 < rcurtin> thanks! one good thing we have going is that we do test our builds on Windows systems via Azure and AppVeyor

19:57 < rcurtin> without those tools it would be so much harder

19:57 < rcurtin> so we were able to get things at least working on the setup that our build nodes have :)

19:58 < MattGomes> rcurtin indeed, having a multi platform pipeline helps with portability :)

19:58 < zoq> MattGomes: In the past I used mingw to build OpenBLAS without any issues - http://kurg.org/mlpack_mingw_build.html

19:58 < zoq> But I guess that changed.

19:59 < MattGomes> rcurtin the issue is that it seems that Mingw64 is not shipping `libgfortran-3.dll` which was needed for the OpenBLAS project (at least when building the shared lib version)

20:00 < MattGomes> maybe you guys have a different version of mingw to mine, I'll definitely try this again at some point to see if I can figure it out

20:01 < rcurtin> MattGomes: hmm, I'm not sure that we are building with MinGW64 at all, I think our full build in our test pipeline is on MSVC, and we download precompiled OpenBLAS

20:01 < rcurtin> but, I could be wrong

20:02 < zoq> The build instructions I referenced above use mingw, so at some point it worked :)

20:02 < MattGomes> Yeah I think you guys have the precompiled `openblas.dll.a` with the corresponding DLL from one of the `OpenBLAS` contributors. I tried that too but it still required `libfortran-3.dll` on my build :(

20:03 < rcurtin> hmm, strange... I don't fully understand linking on Windows so I'm not sure why it would be needed in your situation but not on Azure

20:03 < MattGomes> It's crazy how badly supported BLAS libraries are on windows, I thought we would have done better by now.

20:03 < rcurtin> yes, especially because everything in scientific computing is built on BLAS!

20:03 < rcurtin> I know that the Julia and Python folks get around this by basically hiding everything in their package managers

20:04 < MattGomes> yes, I had no issues with BLAS/LAPACK on linux before. But it's extra painful getting these on Windows haha

20:04 < rcurtin> so for instance, if you install numpy on Windows, it comes in a "wheel" which contains a carefully compiled OpenBLAS dll

20:04 < MattGomes> It does, I kind of did that with Anaconda to compile OpenBLAS

20:04 < MattGomes> I guess I could have pulled OpenBLAS from Anaconda as well

20:04 < MattGomes> but I was on a mission to compile it myself :D

20:05 < rcurtin> :)

20:05 < MattGomes> Anyway, I need to get some sleep now as I am working tomorrow. Will get that VS guide PR this week when I get some time. See ya

20:06 < rcurtin> sounds good, see you later!

21:11 < tobelegitGitter[> I just built the mlpack.

21:12 < tobelegitGitter[> What are the next steps?

21:25 < tobelegitGitter[> git init and setting origin ?

21:25 < tobelegitGitter[> git init and set origin ?