Setting up Pants
How to set up Pants for local development.
Step 1: Fork and clone pantsbuild/pants
We use the popular forking workflow typically used by open source projects. See https://guides.github.com/activities/forking/ for a guide on how to fork pantsbuild/pants, then clone it to your local machine.
openssl
Pants requires a more modern OpenSSL version than the one that comes with macOS. To get all dependencies to resolve correctly, run these commands:
$ brew install openssl
$ echo 'export PATH="/usr/local/opt/openssl/bin:$PATH"' >> ~/.bashrc
$ echo 'export LDFLAGS="-L/usr/local/opt/openssl/lib"' >> ~/.bashrc
$ echo 'export CPPFLAGS="-I/usr/local/opt/openssl/include"' >> ~/.bashrc
(If you don't have brew
installed, see https://brew.sh.)
Step 2: Set up Git hooks
We use two Git hooks:
- If you didn't touch any Rust code, we add
[ci skip-rust]
to your commit message. If you didn't touch any of our release-related code, we add[ci skip-build-wheels]
to your commit message. - Every time you run
git commit
, we run some checks and lints.
To install these, run:
$ build-support/bin/setup.sh
You can manually run the pre-commit check by running:
$ build-support/githooks/pre-commit
Use git commit --no-verify
to skip the checks.
We do not recommend this, though! The Python formatter Black frequently needs to make changes, and you'll need to do an entire new CI run if there are any formatting or linting issues.
Step 3: Bootstrap the Rust engine
Run ./pants
.
Pants will download Rust and compile the engine. This command will also set up the Python virtual environment.
Rust compilation is really slow. Fortunately, this step gets cached, so you will only need to wait the first time.
If this takes more than 25 minutes, please comment on https://github.com/pantsbuild/pants/issues/9822 with the time it took for you to compile. (It's extra helpful if you are willing to share your CPU's speed and # of cores.)
We default to compiling with Rust's release
mode, instead of its debug
mode, because this makes Pants substantially faster. However, this results in the compile taking 5-10x longer.
If you are okay with Pants running much slower when iterating, set the environment variable MODE=debug
and rerun ./pants
to compile in debug mode.
Downloading Rust and compiling the engine typically results in several gigabytes (~7-10 GB) of storage.
We have also not yet implemented automated garbage collection for building the engine because contributors are the only ones to need to compile Rust, not every-day users. Whenever we upgrade versions of Rust or make changes to the Rust code, the old files will stay around.
To free up space, run these commands:
rm -rf src/rust/engine/target
rm -rf ~/.cache/pants/bin
rm -rf ~/.cache/pants/rust
Warning: this will cause Rust to redownload and recompile everything.
Configure your IDE (optional)
Hooking up the Python virtual environment
Most IDEs allow you to configure a Python virtual environment so that the editor understands your Python import statements.
Pants sets up its virtualenv at build-support/virtualenvs/<arch>/pants_dev_deps.py{version}.venv
. Point your editor to the bin/python
file in this folder, e.g. build-support/virtualenvs/Darwin/pants_dev_deps.py36.venv/bin/python
.
Pycharm guide
- Use "New project" and click the option "Existing interpreter". Point the interpreter to the virtual environment location described above.
- In your project tree (the list of folders and files), secondary click the folder
src/python
. Click "Mark directory as" and choose "Source Roots".
VSCode guide
Add this to your settings.json
file inside the build root's .vscode
folder:
{
"python.analysis.extraPaths": ["src/python"],
"python.formatting.provider": "black",
"python.linting.enabled": true,
"python.linting.flake8Enabled": true,
"python.linting.flake8Args": ["--config=build-support/flake8/.flake8"]
}
python.analysis.extraPaths
lets VSCode know where to find Pants's source root. The other config enables black
and flake8
.
We'd love your help. You can either click "Suggest edits" in the top right corner of this page or message us on Slack.