Based on my experience looking/installing/running/modifying open-source projects at my own computer, and my experience working on code at my employer, I came up with the following expectations for a README file for any project:
- Project name, obviously.
- Project purpose.
- What is the problem it tries to solve?
- Who should use this project?
- Why would people use this project?
- Project overview.
- What does this project do?
- How does this project work?
- What are the expected inputs and outputs?
- If this project is part of a larger pipeline or a larger system, where does this project fit in the overall picture?
- Setup instructions.
- How do I setup the project to get it up and running?
- What are the dependencies of this project?
- What are the assumptions about the environment this project is running?
- If it is a web service or site, where can I view it? In production? In dev?
- If it is a script, how should I run it? I hope it has built-in support for
--help.
- Code overview.
- Where is the code?
- This is simpler on external/isolated projects, because everything inside the same repository belongs to the project, and everything outside doesn’t.
- Inside my employer repositories, this is not straightforward, and needs to be explained. What directories inside which repositories are relevant to this project?
- How is the code organized?
- What is the purpose of each directory or file?
- Just an overview, so that I know what and where is the front-end code, the back-end code, the HTTP endpoints, the database modules, the configuration files…
- How does the different pieces of code interact?
- An brief overview of code interaction can help building a mental picture of the execution and data flows.
- What are the relevant databases and tables?
- What are the relevant metrics and dashboards?
- Where to find the documentation?
- The documentation should be close to the code. That way, the documentation is easy to find, and can be kept up-to-date together with the code. That’s why a README file is a good solution.
- If a project is spread over many directories, having a README on each one that points to the main documentation can be very helpful.
- If the main documentation is somewhere else, there should be a README (or code comments) pointing to the main documentation.
- Having comments on database tables and columns can also be very helpful.
- Where is the code?
- Authors.
- Who created it?
- How to contact them?
- History.
- When was it created?
- What were the major changes? (AKA change log, what’s new, release notes…)
- A brief history of the project can help understanding some legacy code and some technical decisions.
Why all of this?
Imagine someone who never heard about your project finds either your project or some piece of code. Either by accident, or because they need to fix/maintain something. Or maybe the project changed ownership.
How would such person get to know what to do? What if that person is you?
Too high expectations?
Not all answers have to be written, some can be implied.
I just want to be able to answer those questions after finding a project.
Some technical details can be understood by just reading the code… But that is only possible if the developer knows what code to read, and where it is.
But maybe my expectations are too high? Do you have a different opinion than me? I can compromise and settle to being able to answer the majority of those questions.
I think I should review the documentation of my side projects, I’m pretty sure many of them are lacking enough answers. However, smaller projects are easier to understand; the bigger the project is, the more it needs good documentation.