mlpack / mlpack

mlpack: a fast, header-only C++ machine learning library
https://www.mlpack.org/
Other
5.11k stars 1.61k forks source link

R package update for deprecated Armadillo load/save methods #3406

Closed eddelbuettel closed 1 year ago

eddelbuettel commented 1 year ago

Issue description

In Armadillo 11.4.4, @conradsnicta deprecated two load/save methods. RcppArmadillo 0.11.4.4.0 brings this to CRAN, now CRAN warns because of it. In the upload notes I promised mlpack would take care of it so I will look into a quick 'narrow' patch for a potential point release. @rcurtin and team can and will of course decide if a wider release is warranted instead.

Your environment

Steps to reproduce

Install the R package under a current (Rcpp)Armadillo.

Expected behavior

Quiet behavior

Actual behavior

We see (and here I quote from the email I got)

Package: mlpack
Check: whether package can be installed
New result: WARNING
  Found the following significant warnings:
    ./mlpack/core/data/save_impl.hpp:107:40: warning: ‘bool arma::Mat<eT>::quiet_save(std::ostream&, arma::file_type) const [with eT = long unsigned int; std::ostream = std::basic_ostream<char>]’ is deprecated [-Wdeprecated-declarations]
    ./mlpack/core/data/save_impl.hpp:128:43: warning: ‘bool arma::Mat<eT>::quiet_save(std::ostream&, arma::file_type) const [with eT = long unsigned int; std::ostream = std::basic_ostream<char>]’ is deprecated [-Wdeprecated-declarations]
  See ‘/home/hornik/tmp/CRAN/mlpack.Rcheck/00install.out’ for details.
  Used C++ compiler: ‘g++-12 (Debian 12.2.0-14) 12.2.0’
eddelbuettel commented 1 year ago

Doh. And syncing with the repo I saw that @conradsnicta actually beat my issue with the merge in #3405 by a good 40 minutes or so. I should have checked. Developers in different timezones for the win.