ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
< yashwants19[m]> <rcurtin "R bindings are merged :-D awesom"> Thank you rcurtin
< yashwants19[m]> :)
d1 is now known as [DONG]DrGibby
ImQ009 has joined #mlpack
yuvraj2701 has joined #mlpack
yuvraj_2701[m] has joined #mlpack
yuvraj2701 has quit [Quit: Leaving]
travis-ci has joined #mlpack
< travis-ci> UtR491/ensmallen#8 (moead_exp - 6db96d4 : UtR491): The build is still failing.
travis-ci has left #mlpack []
travis-ci has joined #mlpack
< travis-ci> UtR491/ensmallen#12 (moead_exp - ba973cf : UtR491): The build was fixed.
travis-ci has left #mlpack []
< rcurtin> shrit[m]: how is the cereal work going? it looks like there is a pretty simple syntax error in the build right now (see cover_tree_impl.hpp line 1753), but maybe you've already fixed it locally :)
< shrit[m]> Yes I have already fix it locally
< shrit[m]> Also I have added several fixes, But I did not push
< shrit[m]> I was afraid reverting all these commits if we are not going to support 16.04
< rcurtin> sounds good, fair point
< rcurtin> I just check the PR build in the mornings to see if the build is passing, and if not, if it's something I can quickly provide a pointer on :)
< shrit[m]> Good we are having an email now if the build is failing
< shrit[m]> rcurtin: The support for map container was easy, I had to include two cereal files properly in mlpack
[DONG]DrGibby is now known as d1
< rcurtin> shrit[m]: awesome, glad to hear it was easy
< rcurtin> hopefully there are not too many other compatibility issues :)
< rcurtin> actually, I guess, hopefully there are no other ones at all :)
< shrit[m]> The only remaining issues I am working on are the pointers for all the `const dataset ` we have left
< shrit[m]> Otherwise until now I am not seeing any new errors.
< rcurtin> awesome!
< rcurtin> if I remember right a quick grep only showed like four or five const pointers being serialized
< rcurtin> but maybe my grep missed something :-D we'll find out I guess
< shrit[m]> rcurtin: On my docker ubuntu 16 mlpack is building on perfectly. Hope that this will be the case on the build farm
< rcurtin> sweet! let's find out :)
< shrit[m]> We need to cereal on macOS and for Windows
< shrit[m]> add*
< rcurtin> for OS X you can install it with `brew`... I'm not sure exactly what to do with windows, whatever the same thing we are doing for armadillo?
< shrit[m]> I have never used macOS or Windows, my knowledge is very limited
< zoq> shrit[m]: brew install cereal will install 1.3.0 on macOS, so adding it to - https://github.com/mlpack/mlpack/blob/master/.ci/macos-steps.yaml#L17 should do the trick
< shrit[m]> Ok, perfect. I will add it once the linux build is finished to be sure that there is no future error from ubunut side.
< zoq> shrit[m]: On windows you have two options either download the header file and extract some known directory or install it via nuget.
< zoq> shrit[m]: There are two version in nuget: https://www.nuget.org/packages?q=cereal
< zoq> shrit[m]: Sounds like you would want at least 1.2?
< shrit[m]> I think 1.1.2 is the minumum support. but 1.2 should be fine
< zoq> shrit[m]: In that case adding: nuget unofficial-flayan-cereal -o $(Agent.ToolsDirectory) right after - https://github.com/mlpack/mlpack/blob/c3af5397472a360da8a791340170eb242925e314/.ci/windows-steps.yaml#L18 should do the trick for windows
< zoq> shrit[m]: Or download the header file, we do the same for armadillo - https://github.com/mlpack/mlpack/blob/c3af5397472a360da8a791340170eb242925e314/.ci/windows-steps.yaml#L32-L33
< rcurtin> zoq: nice find with nuget, I only saw that someone packaged 1.0.0, which is way too old
< AakashkaushikGit> Hey, I am a bit confused as to where the mean absolute error function should go ?
< AakashkaushikGit> should it go under methods/ann/loss_functions ?
< anjishnu[m]> Yep. The 2 hpp files will go into that folder.
< AakashkaushikGit> won't it be a single hpp file ?
< anjishnu[m]> No. One hpp file will have the class definition and the other will have the implementation.
< anjishnu[m]> See any of the other loss functions in the folder for an example of this.
< AakashkaushikGit> oh okay thanks, because before this i have done the declarations in hpp files and definitions in cpp files.
ImQ009 has quit [Quit: Leaving]
< AakashkaushikGit> https://www.mlpack.org/doc/mlpack-git/doxygen/build.html#build_config here in the Dependencies of mlpack there is a package mentioned as python-pandas but that is no more available even on the universe repository on ubuntu 20.04.1 but python3-pandas is available, so should that be updated there ?
< rcurtin> AakashkaushikGit: yeah, we should link to the python3 packages instead---want to open a PR for the documentation changing all the python packages to python3?
< AakashkaushikGit> Sure,
< AakashkaushikGit> (edited) Sure, => Sure.
< rcurtin> thanks!
< AakashkaushikGit> hey @rcurtin I have also taken up the mean abs percentage error so is it okay if i do the PR for documentation change after that?
< rcurtin> sure, no problem
< AakashkaushikGit> Great.
< shrit[m]> rcurtin: This build is failing because of files that are not on my local git, neither on my branch neither on my master of mlpack https://dev.azure.com/mlpack/mlpack/_build/results?buildId=4212&view=logs&j=24d3abe3-ef0b-5deb-3aab-64d839de2c3c&t=8be94158-0791-5881-12e7-451b62b18296
< shrit[m]> There is something strange in dev.azure I feel that it is building the wrong commit somehow. Otherwise, this error is impossible to get.
< rcurtin> shrit[m]: I wonder if azure is merging master in before building
< rcurtin> that may be the case, and I can see that spatial_dropout_impl.hpp was just merged, so maybe if you just merge master and adapt the new files, it will work
< shrit[m]> Of course, Is this a normal behavior from Azure?
< rcurtin> I'm not sure
< rcurtin> you can see what it is doing in the "checkout mlpack" stage of the job
< rcurtin> I think it is merging your commits onto master, then building
< shrit[m]> 👍️ R bindings seems to bring boost serialization
< rcurtin> shrit[m]: they do for now; Yashwant made a handwritten PR to show that cereal serialization can work: https://github.com/Yashwants19/RcppMLPACK/pull/13
< rcurtin> I think, for now, if you build with -DBUILD_R_BINDINGS=OFF, there should be no issue
< rcurtin> and before we merge cereal, we can get the cereal changes integrated into the R bindings
< rcurtin> (as part of your PR)
< rcurtin> but, Yashwant and I can worry about that part, so don't worry if the R build fails on CI for now
< shrit[m]> Perfect, I got the idea. 👍️
< abernauer[m]> That's exciting news on the R bindings congrats.
< rcurtin> thanks abernauer[m]! hopefully they will see some usage soon :)
< abernauer[m]> Wasn't up on the development lately. Quite busy right now between getting around a 100 job apps out a week and trying to fit improving my tech/programming skills in on top of that.
< rcurtin> no worries---hopefully some of the job applications turn into interviews :) good luck!
< zoq> Looking into the Linux Python build.
< rcurtin> thanks!
< rcurtin> jeffin143[m]: zoq: I use mutt for email so I don't see images unless I open the link... I am looking through some of these catch PR gifs and laughing really hard
< rcurtin> I think my favorite is this one: https://github.com/mlpack/mlpack/pull/2588