Installation for Mere Mortals

Pre-compiled

Windows

I used the Appveyor service to compile “wdmtoolbox” on Windows for Python 3.7, 3.8, and 3.9. These versions are available via pip.

On Windows, the best solution for Python is conda, however there is some conflict between conda install … and pip install … that should be noted. What works best is to install everything that is available with conda install … first.

  1. Download and install either “conda” or “miniconda” in 32 or 64 bit versions according to your machine. https://conda.io/docs/get-started.html

  2. Open “Anaconda Prompt”.

  3. Create a Python 3.7, 3.8, or 3.9 environment and activate it. For the example below I am going to create a Python 3.9 environment called py39.:

    conda env -n py39 python=3.9
activate py39

  4. Install everything that you can using “conda”. I made the following lines short so they would be less likely to wrap, but you can install in any order.:

    conda install pandas scipy matplotlib
conda install sphinx ruamel.yaml regex
conda install lockfile tabulate tzlocal

  5. Install “wdmtoolbox” using “pip”.:

    pip install wdmtoolbox

  6. Use “wdmtoolbox”!:

    wdmtoolbox --help

A convenient way to use “wdmtoolbox” is to put the command that you would type at the command prompt into a batch file. A batch file is a text file (editable by any text editor) with a “.bat” file name extension. Then you can run all of the commands in the batch file by running the batch file.

Next time you open the “Anaconda Prompt” to use “wdmtoolbox” you have to activate the environment where it is installed.

Compile From Source

Windows

I use http://appveyor.com to compile the 32-bit version and Github Actions to compile the 64-bit versions. Getting the “appveyor.yml” and Github workflow configuration files right is a nightmare, but finally figured it out.

You should be able to compile it in other ways if you know your way around Windows, Python, and FORTRAN development. The minimal environment that is needed is to have the same “C” compiler as was used to compile your target Python version (for Python 3.x it is the “C” compiler with Microsoft Visual Studio 2019), and a FORTRAN compiler (for example Mingw compilers that come with MSYS2).

Linux

Why I don’t use “conda” on Linux:

  • I really didn’t like how “conda” was overriding system libraries on Linux. Really?

  • Is “gdal” fixed yet?

  • The wild, wild west of anaconda.org

I went back to virtual environments using “pip” to install everything. Works very well since the big players (numpy, scipy, pandas, and matplotlib) now deliver wheels.

  1. Need to have “gcc” and “gfortran” installed.

  2. Make whatever 3.7, 3.8, or 3.9 Python environment you want and activate it.

  3. I have had to install the “wheel” library first and then the basic science stack individually:

    pip install wheel
pip install numpy
pip install pandas
pip install matplotlib
pip install sphinx

  4. Install “wdmtoolbox”:

    pip install wdmtoolbox