Thanks for your interest in contributing to the Opentrons platform! This Contributing Guide is intended to ensure best practices for both internal Opentrons contributors as well as any external contributors. We want to make sure you’re set up to contribute effectively, no matter if you’re helping us out with bug reports, code, documentation, feature suggestions, or anything else. This guide covers:
- Opening Issues
- Opening Pull Requests
- Commit Guidelines
- Project and Repository Structure
- Development Setup
- Robot Environment
Other important reading not included in this document:
This Contributing Guide was influenced by a lot of work done on existing Contributing Guides. They're great reads if you have the time!
Filing an issue is a great way to contribute to the project! Bug reports and feature requests are really useful for us as we plan our work. If you’d like to open an issue, please consider the following questions before opening:
- Is this issue for a bug, a feature request, or something else?
- Please make this is clear in your description so it’s easier to address
- Has this issue already been opened?
- Duplicate tickets slow things down, so make sure to search before you open!
- If there’s already a ticket, please comment on the existing thread!
- Is this a support request?
- If yes, you're better off checking out our support page rather than opening a GitHub issue
To ensure your issue can be addressed quickly, please fill out the sections in the existing issue template to the best of your ability!
If you’d like to contribute code to the Opentrons platform, pull requests (PR's) are the way to do it. Any code contributions are greatly appreciated! If you’re an external contributor, we’re going to assume you are familiar with the fork and pull request flow. If not, this blog post by Scott Lowe is a good introduction.
Please note that by contributing to the Opentrons platform, you agree to share those contributions under the terms of the Apache 2.0 license.
Before opening any PR, please run through the following questions:
- Does this PR address an already open issue?
- If not, please consider opening an issue first
- This is to ensure you don't end up duplicating work or wasting time on a PR that won't be accepted
- Does this PR incorporate many different changes?
- If yes, would the PR work better as a series of smaller PR's?
- Our team is more than happy to help you figure out an incremental plan
- Does this PR include code changes without test and/or documentation updates?
- If yes, your PR may not be ready to open
- Tests and documentation are a vital part of any code contribution
- Are there a reasonable number of commits and are they properly informative?
- The best kind of PR is a tiny PR with a single commit
- To avoid introducing problems into our Git history, we may have to ask you to squash or otherwise amend your commit(s)
- See Commit Guidelines below for tips on keeping a good Git history
To ensure your code is reviewed quickly and thoroughly, please fill out the sections in the existing pull request template best of your ability! If you’d like some recommended reading for writing good pull requests, check out:
- How to write the perfect pull request
- The (written) unwritten guide to pull requests
- The Art of a Pull Request
After your Pull Request is merged (or otherwise closed), you’ll want to make sure to delete the branch in GitHub. You probably want to delete your local branch, too, depending on your own personal organizational strategies / general paranoia.
If you're looking for something to work on, especially for a first contribution, check out our list of easy issues. Be sure to drop a comment in the thread before starting work to make sure nobody else has picked it up.
Before you're ready to make a commit, you should do you best to make sure that:
- All tests are passing
make test
- See Testing section for more details
- All code quality checks are passing
make format lint
- See Code quality section for more details
- Especially consider setting up your code editor to run formatting and quality checks automatically as you code
Good commit messages are essential to keeping an organized and readable Git history. A readable Git history makes our lives easier when doing necessary work like writing changelogs or tracking down regressions. Please read How to Write a Git Commit Message by Chris Beams and then come back here. These selected guidelines (copied and pasted from that article) are a very good starting point to think about when writing your commit message:
- Separate subject from body with a blank line
- Capitalize the subject line
- Do not end the subject line with a period
- Use the imperative mood in the subject line
- Use the body to explain what and why vs. how
When committing, we use commitizen to format our commit messages according to the Conventional Commits specification. This allows us to automatically generate CHANGELOGs based on commit messages. To commit files, first install commitizen
, then:
git add path/to/files
git cz
This will launch the commitizen wizard, which will ask you to:
- Select a commit type, which will be one of:
feat
- A new featurefix
- A bug fixdocs
- Documentation only changesstyle
- Changes that do not affect the meaning of the code (white-space, formatting, missing semi-colons, etcrefactor
- A code change that neither fixes a bug nor adds a featureperf
- A code change that improves performancetest
- Adding missing tests or correcting existing testsbuild
- Changes that affect the build system or external dependencies (example scopes: gulp, broccoli, npm)ci
- Changes to our CI configuration files and scripts (example scopes: Travis, Circle, BrowserStack, SauceLabs)chore
- Other changes that don't modify src or test files
- Select a scope
- For
feat
,fix
,refactor
, andperf
, this should a top-level project, e.g.app
orapi
- For other commit types, use your best judgement or omit
- For
- Write a short commit title
- Written according to the guidelines above
- Write a longer description if necessary
- Also written according to the guidelines above
- Mention any tickets addressed by the commit
- e.g.
Closes #xyz
- e.g.
Most of Opentrons’ projects live in the Opentrons/opentrons repository. Having multiple projects in one repository (also known as a monorepo) is convenient for keeping various inter-project dependencies in sync, but does require workflow considerations to keep everything organized and trackable.
Generally, the directory / file structure of our monorepo looks something like this:
- [Project]
- [Another Project]
- etc.
scripts
- Repository level scripts (mostly for CI)Makefile
- Top level makefile for CI- Various repository level dotfiles (CI and git config)
README.md
,CONTRIBUTING.md
,LICENSE
, etc.
Our projects use a mix of languages, but mostly Python (backend + robotics) and JavaScript (frontend). Each project has its own README
+ Makefile
+ dependency management.
If you'd like to contribute (or maybe just run the very latest and greatest version), this section details what you need to do to get your computer and local repository set up.
Individual projects may have additional instructions, so be sure to check out the various project README
s, too.
You will need the following tools installed to develop on the Opentrons platform.
- make
- git
- curl
- ssh
- Python v3.10
- Node.js v16
- Yarn 1
See DEV_SETUP.md for our recommended development setup guides for macOS, Windows, and Linux.
We use:
- pytest to test Python
- Vitest to test JavaScript
- To run tests in watch mode, you should also install watchman
- Cypress.io for end to end UI testing
You can tests with:
# run all tests (except e2e)
make test
# run tests per language
make test-py
make test-js
# run cypress e2e tests
make test-e2e
You can pass some options to the JavaScript tests:
# run JavaScript tests in watch mode
make test-js watch=true
# disable test coverage
make test-js cover=false
# update snapshot tests
# https://vitest.dev/guide/snapshot.html
make test-js updateSnapshot=true
And you can run code linting / typechecking with:
# lint all code
make lint
# lint + typecheck specific languages
make lint-py
make lint-js
make lint-css
make check-js
To help with code quality and maintainability, we use a collection of tools that can be roughly sorted into the following categories (with some overlaps):
- Linters
- Typecheckers
- Verify that the code is type safe
- typescript
- mypy - Static type checker for Python
- Formatters
These tools can be run with the following commands:
# lint all code and run all typechecks
make lint
# lint by language
# note: Python linting also includes typechecking
make lint-py
make lint-js
make lint-json
make lint-css
# typecheck JavaScript code
make check-js
# format JavaScript, JSON, Markdown, and YAML
make format
Most, if not all, of the tools above have plugins available for your code editor that will run quality checks and formatting as you write and/or save. We highly recommend setting up your editor to format and check your code automatically.
- ESLint - https://eslint.org/docs/user-guide/integrations#editors
- stylelint - https://stylelint.io/user-guide/complementary-tools/#editor-plugins
- Flake8 - Search your editor's package manager
- mypy - Search your editor's package manager
- TypeScript - https://github.com/Microsoft/TypeScript/wiki/TypeScript-Editor-Support
- Prettier - https://prettier.io/docs/en/editors.html
JavaScript dependencies are installed by yarn. When calling yarn, you should do so from the repository level.
A development dependency is any dependency that is used only to help manage the project. Examples of development dependencies would be:
- Build tools (webpack, babel)
- Testing/linting/checking tools (vitest, typescript, eslint)
- Libraries used only in support scripts (aws, express)
To add a development dependency:
# with long option names
yarn add --dev --ignore-workspace-root-check <dependency_name>
# or, with less typing
yarn add -DW <dependency_name>
A project dependency is a dependency that an application or library will import
at run time. Examples of project dependencies would be:
- UI / state-management libraries (react, redux)
- General utility libraries (lodash)
Project dependencies should be added to the specific project that depends on them. To add one:
yarn workspace <project_name> add <dependency_name>
After you have installed a dependency (development or project), you may find that you need to also install typescript type definitions. Without type definitions for our external dependencies, we are unable to typecheck anything we import
. If you are using a dependency that does not come with type definitions, see the typescript consumption guide for how to find and use community-created type definitions.
Not every JavaScript package has an available TypeScript definition. If you find yourself using such a library, you may need to create your own ambient type declaration.
This section is a work in progress.
cd
into the project directory- Use
pipenv install [--dev] <package_name>
to add the dependency
Be sure to check out the API README
for additional instructions.
To run the Opentrons HTTP API in development mode:
# run API with virtual robot
make -C robot-server dev OT_ROBOT_SERVER_simulator_configuration_file_path=simulators/test.json
# run API with robot's motor driver connected via USB to UART cable
make -C robot-server dev
Generally to test your code on the robot, you will want to push the whole mono-repo to the robot in case there are large differences between your robot's server version and the code you are using from github. You can do this via:
# Use this command From the top level opentrons folder
make push host=${some_other_ip_address}
To put the robot server on a test robot, do:
# push the current contents of the api directory to robot for testing
# defaults to currently connected ethernet robot
make push-api
# takes optional host variable for other robots
make push-api host=${some_other_ip_address}
To SSH into the robot, do
# SSH into the currently connected ethernet robot
make term
# takes optional host variable for other robots
make term host=${some_other_ip_address}
If make term
complains about not having a key, you may need to install a public key on the robot. To do this, create an ssh key and install it:
ssh-keygen # note the path you save the key to
make -C robot-server install-key br_ssh_pubkey=/path/to/pubkey host=${some_other_ip_address}
and subsequently, when you do make term
, add the ssh_key=/path/to/key
option:
make term ssh_key=/path/to/privkey
If you create the key as ~/.ssh/robot_key
and ~/.ssh/robot_key.pub
then make term
and make install-key
will work without arguments.
OT-2 robots use systemd-journald for log management. This is a single log manager for everything on the system. It is administrated using the journalctl utility. You can view logs by just doing journalctl
(it may be better to do journalctl --no-pager | less
to get a better log viewer), or stream them by doing journalctl -f
. Any command that displays logs can be narrowed down by using a syslog identifier: journalctl -f SYSLOG_IDENTIFIER=opentrons-api
will only print logs from the api server's loggers, for instance. Our syslog identifiers are:
opentrons-api
: The API server - anything sent tologging
logs from the api server package, except the serial logsopentrons-update-server
: Anything sent tologging
logs from the update server packageopentrons-api-serial
: The serial logs
OT-2 robots use systemd
as their init system. Every process that we run has an associated systemd unit, which defines and configures its behavior when the robot starts. You can use the systemctl utility to mess around with or inspect the system state. For instance, if you do systemctl status opentrons-api-server
you will see whether the api server is running or not, and a dump of its logs. You can restart units with systemctl restart (unitname)
, start and stop them with systemctl start
and systemctl stop
, and so on. Note that if you make changes to unit files, you have to run systemctl daemon-reload
(no further arguments) for the init daemon to see the changes.
Our systemd units are:
opentrons-api-server
: The API serveropentrons-update-server
: The update server
An OT-2's filesystem is mounted from two separate locations. /data
, /var
, and /home
are from the "data" partition, and everything else is from the root partition (or generated by the system). The root partition is what gets updated, by being overwritten. To make this work, the root partition is mounted readonly, which causes writes to files in that partition to fail with the error "readonly filesystem". To prevent this, you can remount the partition:
mount -o remount,rw /