Rsymphony is one of solver packages supported by TestDesign. As of Rsymphony version 0.1.29, Mac users may encounter error messages as below when trying to install the package from R.
* installing *source* package ‘Rsymphony’ ...
** package ‘Rsymphony’ successfully unpacked and MD5 sums checked
** using staged installation
Cannot find SYMPHONY libraries and headers.
See <https://projects.coin-or.org/SYMPHONY>.
ERROR: configuration failed for package ‘Rsymphony’This document describes a potential fix to this problem. An admin account is required.
The cause of the error is that Rsymphony requires SYMPHONY libraries, but the libraries are not properly installed. The goal of this section is to install the libraries.
Step 1. Open the terminal from Launchpad. Switch to
the admin account on your system by running the following line in the
terminal prompt. Make sure to replace username with the
actual username of the admin account. Type in the account password once
prompted.
su usernameStep 2. Install the homebrew package
manager as needed by following the instructions on https://brew.sh/. The purpose of
this step is to allow installing SYMPHONY with it.
Step 3. Install SYMPHONY and its dependencies by running the following in terminal. Each line must be run separately.
brew tap coin-or-tools/coinorbrew install symphonybrew install pkg-configbrew cask install gfortranRsymphony solver package should now be able to locate the SYMPHONY libraries. Retry installing the Rsymphony solver package in R.
If the install fails, proceed to the next step.
Step 1. Install wget by running the
following line in terminal. The purpose of this step is to allow
downloading the package file directly from CRAN servers.
brew install wgetStep 2. Run the following line in terminal to download the Rsymphony solver package.
sudo wget https://cran.revolutionanalytics.com/src/contrib/Rsymphony_0.1-29.tar.gzStep 3. Unpack the downloaded file by running the following line.
sudo tar -xzvf Rsymphony_0.1-29.tar.gzStep 4. Open the file
Rsymphony/configure by running the following line. This
opens nano text editor, with -c option to make
it display the line numbers.
sudo nano -c ./Rsymphony/configureComment out lines 51-55 by inserting #s at the beginning
of each line, as shown below. Save it as the same file with Control + O,
and when prompted for file name to write press Enter. Exit
nano with Control + X.
#if test ${status} -ne 0; then
# echo "Cannot find SYMPHONY libraries and headers."
# echo "See https://projects.coin-or.org/SYMPHONY."
# exit 1
#fiStep 5. Open the file
Rsymphony/src/R_symphony.cc by running the following
line.
sudo nano -c ./Rsymphony/src/R_symphony.ccChange line 12 #include <coin/symphony.h> as shown
below. Save it as the same file with Control + O, and when prompted for
file name to write press Enter. Exit with Control + X.
#include <symphony.h>Step 6. Run the following line in terminal to install the modified Rsymphony package.
sudo R CMD install RsymphonyThis will make the Rsymphony solver package available in R.
Do NOT run install.packages('Rsymphony') within R after
Step 6. Doing so will trigger a re-install of the unmodified
package.
If Step 6 fails with --lgfortran error, proceed to the
next step.
The cause of the error is that R needs Fortran libraries to build the package, but it does not know where they are. They are already available on your system: when you installed SYMPHONY libraries in the previous step, it installed gcc library as a requirement, which contains Fortran libraries as well.
Step 1. Open the file ~/.R/Makevars by
running the following line.
sudo nano -c ~/.R/MakevarsThe file will be empty in most cases. Add the following line to point R to the Fortran libraries. Save it as the same file with Control + O, and when prompted for file name to write press Enter. Exit with Control + X.
FLIBS = -L/usr/local/gfortran/lib -lgfortran -lquadmath -lmStep 2. Run the following line in terminal to install the modified Rsymphony package.
sudo R CMD install RsymphonyThis will make the Rsymphony solver package available in R.
Do NOT run install.packages('Rsymphony') within R after
Step 2. Doing so will trigger a re-install of the unmodified
package.