Skip to content

Latest commit

 

History

History
136 lines (97 loc) · 5.14 KB

mac_homebrew.md

File metadata and controls

136 lines (97 loc) · 5.14 KB
Raw

How to install - 🍎 OS

Homebrew: A Mac's package manager 📦

Most of the Unix-based systems have a package manager to install libraries, or even GUI software. Nonetheless, Mac does not have a native package manager, Homebrew fills this gap and will make your life easier, even for other tasks beyond library installation.

To install Homebrew, run this line in your terminal:

/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

With Homebrew installed you can use its different formulas (in Homebrew parlance) to install programs and libraries, like: brew install [library-or-package]:

  • Git (installed by default in Mac, but run xcode-select --install to update git, install compilers and other GNU tools).
  • psql: brew install psql
  • Python: brew install python3 (MacOS includes Python, but that is the system Python and it's better not to mess with it).
  • R: brew install r

But Homebrew is not limited to libraries, it also handles other common programs, like Chrome, or Spotify. You can use it to install other useful tools:

  • Sublime Text: brew cask install sublime-text
  • RStudio: brew cask install rstudio
  • DBeaver: brew tap caskroom/versions; brew cask install dbeaver-community

Homebrew does not install this libraries directly to the Applications folder, it rather symlink them from the /opt/homebrew-cask/Caskroom path to the /Applications directory. Same happens with the installed libraries, but these are symlink'd from /usr/local/Cellar to /usr/local/bin. This is helpful because we can alternate between different library versions (see brew switch)

Python 🐍 environment tools

Different python versions imply different library availability and python new functionalities. Working in different projects not only involves different versions of Python, but also several different libraries versions. Not being careful with Python installations can lead to ugly scenarios where we cannot control the python version we are running.

To avoid this, we have a myriad of options depending on the user expertise level.

Another reason of why having environments is a good idea is for reproducibility. We want your code to run successfully in any machine, and having the right libraries is pivotal to this.

1. Anaconda (for beginners)

Anaconda's Python distribution includes a handy environment manager called conda. Conda is able to take care of different aspects of our environment, like Python version and the libraries that the environment should have.

Once installed, you can create a new virtual environment (my_new_env) with all the requested libraries in the requirements.txt:

conda env install -n my_new_env -f requirements.txt

Another neat feature of Conda is the existence of its own libraries manager that serves as an alternative to pip. Just as Python's builtin libraries manager, conda can install any package in the conda distribution:

conda install <your package name here>.

This package manager has some advantages over pip. First, conda relies on its own compilers, hence it facilitates the installation of packages that need C compilers (although, you will need to install gcc via Brew) and with other ugly dependencies. Second, it has faster mathematical libraries like numpy since uses BLAS for compilling (faster linear algebra operation in the processor for Intel processors). Third, it also gives a more intuitive use of package installation.

Nonetheless, not all packages available via pip are in conda, so you will have you use pip and probably face some small conflicts (this is rare).

2. Miniconda (for beginners without much space disk left)

The Anaconda suite can be a heavy package for some space-limited users. One available option is Miniconda, a lighter version of Anaconda which also includes the environment manager conda.

3. pyenv + virtualenv (only if you live dangerously)

If you do not want to use Condas's Python distribution, you can manage your environments using pyenv and virtualenv. To install both you have to use Homebrew:

brew install pyenv
brew install pyenv-virtualenv

Once both libraries are installed, the next step is to link both libraries to let the computer know where are the pyenv python versions installed. This can be done by running this (assuming you use bash):

echo 'eval "$(pyenv init -)"' >> ~/.bash_profile
echo 'eval "$(pyenv virtualenv-init -)"' >> ~/.bash-profile

pyenv is a powerful library that has a lot of options, it is versatile and allow a smooth deployment. Although, sometimes is hard to use, and is not very friendly during development. pip + virtualenv, both Python libraries is other available option.