Early in college I put open source software projects on a pedestal. I wanted to contribute, mainly for bragging rights and the thrill that comes with it. I didn’t feel I was qualified enough because almost all of the issues I encountered in products that I used regularly were well above my expertise and experience. There was also no way known to me at that time for finding easy issues apart from Mozilla’s “Good First Bugs”. I learnt about amazing projects like Your First PR and Up For Grabs much later. My method of finding simple issues was much more involved. I used to scourge trending projects on GitHub for issues and bookmark them for review. You don’t have to do that though, you can use the amazing tools I’ve mentioned above thanks to Scott Hanselman, Kent C. Dodd and Charlotte Spencer. This is how I started and where I am now. I hope this helps other people who feel overwhelmed or think that the community is not welcoming or just somebody who wants to be reminded about why they started contributing.
Getting a feel for Open Source
Before you can dive in deep and start submitting non-trivial patches you need to get a feel for the process. I started by opening issues, lots of them, in projects that I used myself as a way to get other people to scratch my itch. You can do the same. Make sure to follow the ensuing discussion and try to participate in it. Once somebody starts writing a patch for your issue make sure to also follow the patch and learn how the other person did things. This will help you in gaining both some general technical knowledge about the project and also introduce you to a first hand demonstration of the code review and patch submission process in the project.
Small and trivial changes
Sometime after opening issues and following them I had gained enough information about the process to start submitting trivial changes. I realised that since the projects were open-source I could scratch my itches myself and felt the empowerment that being literate (in this case, computer literate) brings. It was my Eureka moment.
I started by contributing a new challenge and solution to an interactive coding challenge (thanks a lot Donne Martin). They were really small and minimal changes that were low-hanging fruit.
These kinds of changes are good to get a feel because they are low risk and are not blocking the project from moving
forward. A good way to find such bugs is to find issues labelled
help needed. These will help you to get
familiar with the tools of the trade and build up confidence. Do not shy away from asking questions whenever you feel
inquisitive or feel stuck on a problem. You can even ask other people in their pull requests as to why and how they did
I then proceeded to open a lot of PRs, a lot of which were rejected while some were accepted. There were some PRs which some would call wasteful or stupid and this PR to Neofetch is a good example of that. I literally added ASCII art to a PR. There’s an invaluable thing I learnt from the rejection of my PRs:
The most common reason for a PR to get rejected is that you did not open an issue and ask for feedback whether the feature aligned with the project’s goals or ideas.
I’ve never had a pull request rejected if there was some discussion before starting work and sending one. What usually happens in such a case is that the maintainer usually drops hints or reading material to help you fix the problem yourself. A very good example is this and this.
Picking a project
I don’t think I paid much attention to what the project was in the early days. I just wanted to get a shiny badge that said I had contributed code to someone somewhere. But that was a mistake. That approach is not sustainable because to make good contributions and improve with time you need context. Context means a familiarity with the project, both it’s code and the development process it follows. So here is some criteria to chose a project:
- The size of the codebase: The less code the easier it is to understand how all of it’s pieces fit together.
- The language of the project: If the project uses a language you are not familiar with it’ll be comparatively difficult to get a good understand of the project.
- Do you use the project: If you are a user of the project you know that any changes you make will affect you. There’s also a higher chance of you being able to find out issues yourself instead of relying on bug reports.
Read the documentation
After selecting a project, the first step is to go read ANY and ALL documentation they have. This is beneficial for two purposes:
- You can learn about how the project works, how to contribute, the coding style and how code is organised etc.
- You can also find gaps in the documentation to fill. This is a very easy and much appreciated way to contribute to a project. Documentation is one of the most neglected parts of a project and fixing it pays it’s dividends by helping future contributors.
When going through the documentation keep a focus on finding out the tools they use, the way the project is organised, the process that they follow and the core ideas and goals of the project.
For gathering information about the social context:
- If the project has any documentation for new contributors read it. Most projects on GitHub/GitLab include a CONTRIBUTING.md file that deals with it. Larger projects generally have a dedicated wiki like Mozilla.
- Subscribe to the project’s issues list (see watching projects on GitHub) and commit feed.
- If they have a Gitter or IRC channel, join it and follow the conversation and go read up on something interesting they mention.
- If they have a mailing list subscribe to it. Larger projects generally tend to use mailing lists and IRC for communication.
For gathering information about the technical context:
- Read the source code if the project is small enough. You will definitely learn something or even find some places with scope for improvement.
- Read the test cases. Tests are also an ignored part of most projects and a great place to start contributing and learning a very important skill.
- Follow some other people’s ongoing contributions to gain an idea of what it’s like to contribute patches and what the things to take care of while doing so.
- Follow the main contributors’ blogs and/or Twitter. One of the best examples of this is Mike Conley’s weekly live-coding session called The Joy of Coding. It’s a great way to learn how people generally work and you can pick up a lot of debugging tips and tricks along the way.
Set up a local development environment
This is one of the most difficult tasks if you have never built a large project yourself or built software from source before. Most projects generally have a README file that will give you enough information to run the project locally. Some larger projects have a dedicated category in their wiki that deals with this and is often called “Developer documentation”. Mozilla has a very comprehensive set of “getting started” documentation that deals with each step of the process:
- Getting the source code. Most projects use some sort of version control tool like Git or Mercurial. Go install those tools and “clone” the project.
- Learn about the various build configurations if there are any. Most project have different build settings that let you get a debug build so that you can attatch a debugger and experiment or an optimized released build which is what you normally run.
- Build the project. It can be as simple as running
makeor can involve using the project’s own build system.
- Run the tests and any linters to get an idea of how the entire process works.
Finding your first issue
Your goal should be to find something that is small and easy. Documentation and tests are great examples. Stylistic fixes also fit the criteria but are discouraged as patches on themselves since they tend to pollute the project’s history. You can also look through the project’s issue tracker for bugs labelled “need help” or “beginners” or something similar.
Working on your first issue
Learn to gain information about the code from the code itself.
git blame/hg blame. There’s a lot of information in the version control system to help you learn. You can look at the blame information and the commit messages to find out why something was written the way it is written. It also helps you to identify the people who have worked on that file before and you can reach out to them for more information.
- Pull request history. There is a lot of discussion and information available on the issue that was tracking a bug and also on the actual pull request. They can come in handy for gaining information.
Now you can move on to writing the actual patch.
Writing a patch
There’s a process I’ve developed for myself where I first try to hack together “proof of concept” to demonstrate that the task at hand can be done and how it looks like and behaves. It also helps to think about the edge cases.
Now you can start fresh and write your most elegant code. The code should be clear, avoid magic numbers and should be self-documenting. Your code should be such that the next person to come along should be able to intuitively understand what the code is doing.
Conform to the styleguide
If the project does have a styleguide, conform to it. There are generally some commit-hooks that run linters before allowing you to commit code to ensure that all code meets guidelines. If there’s no formal style guide or linter, keep your style consistent with what is already written in the file. The entire project should ideally look like it was written by a single person. Attention should be paid to variable naming conventions, spacing and layout and code organisation.
If you are submitting a patch relating to some code that is not covered by tests, write some tests for it (in a seperate commit generally). This benefits both you and the project because then the person reviewing your code doesn’t have to pull your changes, build the project with the changes and manually test everything. They can simply look at the tests to determine if the code is correct and will have intended effects. This will make their work easier and you will gain some points too.
The commit message
In open source you cannot assume that you will be working on the same project or be available for contact a year later. This means that the next person who has to work on that code should have all the required information without ever needing to contact you. This information is mostly passed down through self-documenting concise code, documentation and commit messages.
A good commit should satisfy the following:
- Should be wrapped within 72 characters.
- Should have a single line message written in imperative language giving a summary of the change. Mozilla uses a
Bug 1224225: Test for punycode/unicode in CSP source matching code r=ckerscb. It includes the bug number so that you can go and see exactly what was fixed and the related discussion. The summary is concise and very specific and written in imperative. The line ends with a convention of
xis one of
+which mean requested review, accepted review and changes suggested respectively. It also provides information about the person who reviewed the patch so that you know who the expert on that part of code is.
- The body of the commit message in a few lines should explain things in more detail and give some background about the issue and maybe link to related issues or discussion.
- The body can also be several paragraphs and should explain the “why” and “what” of your patch instead of “how”. It means that the patch doesn’t need to focus on the implementation detail. It should focus on why the approach was chosen among other alternatives and why a certain decision was made. If the fix is doing something complicated you may give an overview of what it is doing.
Opening a pull request
This is your chance to make a good first impression. You should go in expecting the standards to be very high. Expect your choices to be challenged, nits to be picked and suggestions to come. Be ready to defend your choices with informed decisions and be willing to incorporate feedback. Almost always a pull request has to go through several iterations before it is accepted. Remember that a critical feedback is not meant to attack you, it’s so you can improve your code and benefit yourself and the project. Even if your PR is rejected this is not a failure. It’s rather an opportunity to improve and internalize the feedback and try again!
If you follow this advice and the advice of the people running the project you will certainly get a PR merged within no time. From personal experience I can tell you that it is illogically thrilling knowing that your code was merged and other people will be using it and you’ve improved their lives.
The process above works like a feedback loop and the more you do it the more familiar you get with the project and rise in the ranks and it becomes much more easier for you to write your next patch. Over time you will begin to “own” some parts of the code in the sense that you know how it all works and maybe you even wrote most of it. When that happens remember to give back to people who are struggling to contribute just like you were once.
There’ll be another accompanying blog post for project owners and maintainers on how to attract and retain good quality contributors to your project. You can find the post here.