Skip to content

Latest commit

 

History

History
158 lines (103 loc) · 5.3 KB

HACKING.md

File metadata and controls

158 lines (103 loc) · 5.3 KB

Hacking on Homebrew-cask Code

If you'd like to hack on the Ruby code that drives this project, please join us, we'd love to have you!

What Are the Project Goals?

Homebrew-cask is an attempt to make a Linux-style package manager for precompiled OS X software. Homebrew-cask is not yet as featureful as apt or yum, but we are trying to be as close as we can get to those tools from the user's point of view.

We manage installed files via the "symlink farm" method, like GNU Stow and Homebrew.

What Is the Design Philosophy?

Homebrew-cask is designed to work like a traditional Unix tool:

  • All functionality should be accessible from the CLI. The user should be freed (freed!) from interacting with a GUI.
  • homebrew-cask should itself be scriptable.

Homebrew-cask is designed to work like Homebrew:

  • Like Homebrew, we don't require the user to install with sudo. In fact, we recommend against it.

What Is the State of the Project?

This is a young project. We are just getting off the ground. We are still revising our goals and adding new ones.

What Needs to be Done?

Plenty. Start with open issues !

Are You Interested in New Features?

Yes, definitely! Bring your own expertise. Just remember that the user always comes first.

Are You Interested in New Types of Packages?

Yes! (We call them "artifacts"). If something is useful (and precompiled) then we'd like to enable users to install it via homebrew-cask.

Could Homebrew-cask Also Be Used to Manage Settings?

It's a neat idea! We have talked about it but nobody has worked on it:

What About Installing by Copying Instead of Linking?

It's a neat idea! We have talked about it but nobody has worked on it:

We would want to make sure that uninstall works equally well when copying.

What About Installing Multiple Versions of a Package?

Yes, definitely! We have started working on it, so please contact us directly if you want to help.

What About Dependencies?

Yes, definitely! We have started working on it, so please contact us directly if you want to help.

What Is Your Relationship to Homebrew?

We are independent of Homebrew as a project.

From the user's point of view, homebrew-cask is a subcommand of Homebrew, so we try to match Homebrew semantics and philosophy wherever possible.

From the programmer's point of view, very little code is shared with Homebrew. It turns out that Homebrew's code is tightly linked with the Homebrew Formula definition. Casks are defined differently than Formulae, which (unfortunately) is a barrier to re-using code.

How Should I Set Up a Development Environment?

Cask authors often work directly within the Homebrew directory under /usr/local. For coding, that is usually not sufficient.

We recommend the following:

  1. Fork our repo: https://github.com/phinze/homebrew-cask/fork

  2. Clone a private copy of the repo:

    git clone https://github.com/<username>/homebrew-cask.git
  3. Add the official repo as the upstream remote:

    cd homebrew-cask
    git remote add upstream https://github.com/phinze/homebrew-cask.git
  4. Now you have two copies of the homebrew-cask codebase on disk: the released version in /usr/local/Library/Taps/phinze-cask, and a development version in your private repo. To symlink the Casks and rubylib folders from /usr/local/... into your private repo, run the following script:

    /<path>/<to>/<private>/<repo>/developer/bin/develop_brew_cask

    Now you can hack on your private repo, and use the brew cask CLI like normal -- it will interact with your latest code.

  5. Important: while in development mode, you can't safely run Homebrew's brew update command. To switch back to production mode, run

    /<path>/<to>/<private>/<repo>/developer/bin/production_brew_cask

Hanging out on IRC

We're on IRC at #homebrew-cask on Freenode. If you are going to develop for homebrew-cask, it's a great idea to hang out with us there. Here's why:

  • discuss your thoughts before coding and maybe get new ideas
  • get feedback from the Travis-CI bot on build failures
  • talk to caskbot about checksums, version info, and releases
  • just to be social!

Mind the test suite!

If you're making changes - please write some tests for them! Also be sure to run the whole test suite using rake test before submitting (if you forget, Travis-CI will do that for you and embarrass you in front of all your friends). :)

Submitting Your Changes

See the relevant section in CONTRIBUTING.md: Submitting Your Changes

Squashing

If your pull request has multiple commits which revise the same lines of code, it is better to squash those commits together into one logical unit.

But you don't always have to squash -- it is fine for a pull request to contain multiple commits when there is a logical reason for the separation.

<3 THANK YOU! <3