conda

Once per Linux user:

    cp -p .bashrc .bashrc-before-conda
    cd ~
    bash /home/dojo/conda/repo.continuum.io/miniconda/Miniconda-latest-Linux-x86.sh
    (follow defaults)
    
    Edit .bashrc to append conda's directory instead of prepending it. E.g.:

        dojo@pan:~$ diff .bashrc-before-miniconda .bashrc
        134a135,140
        > 
        > #!!!# added by Miniconda 3.10.1 installer
        > #!!!export PATH="/home/dojo/miniconda/bin:$PATH"
        > 
        > # modified by dojo to _append_ instead of prepend
        > export PATH="$PATH:/home/dojo/miniconda/bin"
        dojo@pan:~$ 

    Close terminal window and open new terminal window to use above changes.

        conda --version
        conda update conda
        conda --version

    Install anaconda-client tool.
    You will need it eventually. Might as well get it done now.

        conda install anaconda-client

once per project: create new environment

    # Must specify at least package for environment.
    # Can not create empty environment.
    # Try to install all packages at once to avoid version skew problems.

    # Specifying particular version of python satisfies conda.
    conda create --name myproject python==3.5.2

    # Specify package I don't care about, configobj, to satisfy conda.
    conda create --name myproject configobj

    # Create environment with newest version of Python 3.
    conda create --name snakes python=3

to begin using an environment:

    newer versions of conda:

        conda activate myproject

    older versions of conda:

        source activate myproject

to quit using an environment:

    newer versions of conda:

        conda  deactivate

    older versions of conda:

        source deactivate

miscellaneous

    --name specifies environment to mess with

    # to get help
    conda whatevercommand -h

    # lists all environment names and paths. '*' indicates active environment
    conda-env list

    # lists all environment names and paths. '*' indicates active environment
    conda info --envs

    # list versions of packages currently available in an environment
    # environment default to current environment?
    # package arguments are regular expressions

        # list all packages and versions in current environment
        conda list

        # list all packages and versions in "bunnies" environment
        # ("bunnies" environment does not have to be active)
        conda list --name "bunnies"

        # list packages and versions with "python" anywhere in package name
        # in current environment
        conda list python

        # list package and version of python in current environment
        conda list ^python$

        # list package and version of configobj in current environment
        conda list ^configobj$

        # list package and version of python in bunnies environment
        # ("bunnies" environment does not have to be active)
        conda list --name bunnies ^python$

        # list package and version of configobj in bunnies environment
        # ("bunnies" environment does not have to be active)
        conda list --name bunnies ^configobj$

    anaconda search -t conda pytest-xdist

    # Create environment with newest version of Python 3.
    conda create --name snakes python=3

    # show python and package versions installed in current environment
    conda list

    # check web for packages and versions with
    # "beautiful-soup" anywhere in package name
    conda search beautiful-soup

    # check web for packages and versions with
    # "python" anywhere in package name
    conda search python

    # check web for packages and versions for python package name
    conda search ^python$

    # install package in an environment (does not have to be active)
    # installs beautiful-soup in bunnies environment
    conda install --name bunnies beautiful-soup

    # without asking for confirmation, install package in active environment
    # installs beautiful-soup in active environment
    conda install --yes beautiful-soup

    # install packages listed in a file
    # into an environment (does not have to be active)
    conda install --name bunnies --file requirements.txt

    # update a package
    conda update --name bunnies beautiful-soup

    # remove package from an environment (does not have to be active)
    conda remove --name bunnies beautiful-soup

    # remove an environment
    conda remove --name bunnies --all

    # for packages that conda can not install,
    # use pip within a conda environment
    source activate bunnies
    pip install pytest-xdist

    source activate bunnies
    pip uninstall pytest-xdist

    # See also cheatsheet
    http://conda.pydata.org/docs/_downloads/conda-cheatsheet.pdf

    # uninstall miniconda
    # rm -rf ~/miniconda

    # some packages, such as pytest-xdist, are not available by conda itself,
    # but can be installed by pip within the conda environment.
    # If the only package one wants is one not available with conda,
    # create a conda environment with a sacrificial package that conda does
    # have, then use pip within the conda environment, then use conda to remove
    # the sacrificial package.
    conda create --name foo configobj
    source activate foo
    pip install pytest-xdist
    conda remove configobj

# environments/projects are subdirectories of /home/dojo/miniconda/envs
# One can go directly from one environment to another.