The Trees The Fork Oak Day7 - Contributing to Open Source

Starting another bomb survival demake

Project Page


Today I'm taking a break on Pando and introducing a new project. But first some background. When I was in high school my friends and I spent a lot of time working on various small games. They ranged from a super simple pong game when we were first starting, to eventually a remake of a Little Big Planet game called Bomb Survival for the windows phone 7. At the time the windows phone had just come out and every employee at Microsoft was given the option to get one for an extreme discount (it may even have been free) so we had a good time making something we considered fun for our new phones. Some guy online made a YouTube review for those who are interested.

These days I use the project as a toy to implement whenever I find a new game engine to play with. The mechanics are just complicated enough to be interesting while still being easy enough to build without much infrastructure. An example of one of my implementations can be found on the Pico-8 forums.

Fast forward to a couple weeks ago when I learned about SCRIPT-8 a fantasy console in the same vein as Pico-8 but written in JavaScript and containing a fascinating editor with time travel and easy game sharing. The project seems not to have gotten a lot of love online yet, so I decided to try my hand at building a game with it. Naturally I decided to port my old standard Bomb Survival game to the platform.

There was a hiccup however. The physics engine architecture I have found to work most naturally for a resource constrained destructible terrain game is a pixel based system. I will go into further detail about how to implement such a system later. Suffice it to say that we need some way to query the engine for what color a given pixel is.

Unfortunately SCRIPT-8 doesn't currently have an API to do this. So for today's daily post I delved into the source code for SCRIPT-8 and made a PR to add the functions I needed. Along the way I will highlight some things I've learned about how to effectively work with open source projects.

First Contact

The first step was to create an issue on the GitHub page broaching the idea of some pixel API changes. I tried hard to communicate clearly exactly what I was thinking of changing, why I wanted to change it, and that I was completely willing to build and own the feature. That last piece is crucial for small projects with a friendly main developer. Gabriel, the author of SCRIPT-8 showed that he was very willing to merge PR's from people other than himself, so offering to contribute was very likely to open doors. The issue I created can be found here.

Useful Tip #1: Ask First

It is better to ask first then to potentially waste time building something the maintainers don't want.

Sure enough an hour or so after I made the issue Gabriel responded agreeing that the feature was a good idea and gave me a pointer to where to get started. We're in


GitHub Process

Second step to getting code merged even before writing a single line of code is to fork the repository. Most GitHub projects will not give push access to random people on the internet. Instead the path to getting a change in starts by forking the repo which creates a new copy of the project on your account. You then make your changes there and make a PR or Pull Request which the original project owners can then either accept, make comments on, or reject. Forking can be done easily on GitHub by clicking the Fork button in the top right corner of the project page.


After forking its generally a good idea to look for instructions for how to get a development build of the project working on your machine. Many projects have these instructions in the root level README however some hide them in the GitHub wiki or docs. In the case of SCRIPT-8 I found them at the bottom of the README.


Unfortunately after cloning, installing the correct dependencies, and running yarn iframe-start as per the instructions, I was greeted with this unhelpful error:

  yarn run v1.13.0
$ npm-run-all -p iframe-start-js iframe-watch-css
$ cd src/iframe; react-scripts start
$ cd src/iframe; stylus src/css/Iframe.styl -o src/css -w
The system cannot find the path specified.
The system cannot find the path specified.
error Command failed with exit code 1.
info Visit for documentation about this command.
error Command failed with exit code 1.
info Visit for documentation about this command.
ERROR: "iframe-start-js" exited with 1.
error Command failed with exit code 1.
info Visit for documentation about this command.

Pretty cryptic... So I did some initial googling. After nothing obvious turned up I then did the important next step of searching the project for similar keywords.

Useful Tip #2: Search the Project

This serves two purposes. First it ensures that nobody else has run into this problem and potentially found a fix. And second it gets me familiar with the types of questions being asked in the issues section. Some projects feel strongly about what goes into issues, so making sure there are similar issue reports is a good strategy. I can't count the number of times I have been able to figure out a problem by searching in the discussions in GitHub issues. Side note: I frequently find this strategy useful as a user as well not just for programming problems.

After not finding similar problems in the issues, and ensuring other people have asked similar questions with friendly responses, I created an issue. Crucially I included what machine I'm running on (this was the important factor in the end), the commands I used, and as much of the error info as possible. This gives the maintainers the highest chance of taking you seriously. If you don't provide enough details, maintainers can often blow you off.

Like clockwork, Gabriel responded an hour later suggesting I install stylus. I did as suggested, but still ran into the same issue. Counter intuitively I took his quick response as a good sign even though it was wrong in the end because it indicated Gabriel is very invested in the project and willing to provide useful suggestions. In the past I have tried contributing to projects with less helpful maintainers and been disappointed or turned away. Friendly maintainers make all the difference.

After turning over the easy stones for some quick fixes I started to push a bit further and debug myself. After some fooling around and searching I eventually landed on this stack overflow page. 3 answers down gave me the crucial hint. Apparently npm, the package manager for javascript doesn't handle windows consoles very well when trying to use package.json scripts. The solution was to configure npm to use a bash implementation on windows instead of the default of cmd.exe. The core of the issue is that cmd couldn't figure out where the npm installed the dependencies for the build process however bash was used correctly and the path was hooked up properly.

Useful Tip #3: Use the Google Luke

Google foo is invaluable. Maintaining cross platform compatibility in a project is extremely hard. Especially if the maintainer only uses or tests on one platform themselves. Encouragingly there are more JavaScript developers on earth than any other kind. So Google is saturated with a near infinite set of monkey developers sticking wrenches where they don't belong. More often than not somebody has stuck a wrench where you have and figured out how to pull it out.

Immediately after figuring out the solution, I documented it in the issue and closed it. Gabriel commented a while later asking me to add a comment to the README describing the pitfall.

Quick Single Change PR

I lied a little above when I said the first step to making a change in an open source project is to fork it. The wonderful folks at GitHub have built quicker way if the change is simple enough. In the repository explorer, if you click on a file there is an edit button.


Clicking it will open up an in browser text editor for you to make small changes.

Making the changes, adding a title, and a description will do the intermediate steps of forking the project if it isn't already on your account, creating a new branch with your change in it, and making a Pull Request. I wouldn't recommend this for big functional changes since there isn't a clean way to test the change without cloning and such, however for documentation changes this strategy is invaluable.

The PR containing the documentation change can be found here

When in Rome Do As the Romans Do

With that issue out of the way and successfully building the dev server, I began the important process of understanding the project's coding style. The first pull request I ever tried to make on GitHub was a minor disaster because I naively thought that I knew how to format things better than the maintainer. The maintainer kindly pointed out my mistake and linked to some contributing guidelines, but at that point I was too embarrased to continue.

Useful Tip #4: Don't Force Your Style on the Maintainers

I wont make that mistake again, so now any time I consider making a change in a project I scan files trying to understand their patterns and formatting style. Programmers feel strongly about coding style. They are much more likely to merge a PR if the code feels like something they would have written.

In the case of SCRIPT-8, the author tends to omit semicolons (brave), use let for variable declarations (awesome), and use es6 module style (even more awesome). I tried to keep these things in mind while making my changes.

Finishing it Up

With all that out of the way, the last part was to actually complete and test the changes. This part took relatively little time since the actual changes were pretty small. I used the hint Gabriel gave in my first issue to figure out where to make the change and emulated the pattern in that part of the code to add my own functions to the API. After testing (and fixing my stupid bugs) I then decided to build a demo to show off the changes and make sure everything was working.

  let x = 20;
let y = 0;
let vx = 1;
let vy = 1;


draw = () => {
  for (let i = 0; i < 100; i++) {
    setPixel(x, y, getPixel(x, y) + 1)
    x = x + vx
    y = y + vy
    if (x >= 127) {
      vx = -1
    if (y >= 127)  {
      vy = -1
    if (x <= 0) {
      vx = 1
    if (y <= 0) {
      vy = 1
    if (random(0, 100) < 5) x += 1
    if (random(0, 100) < 5) x -= 1
    if (random(0, 100) < 5) y += 1
    if (random(0, 100) < 5) y -= 1

After a short while running the above demo yields this image:


Lastly I looked up and updated the documentation to show the new API.

Useful Tip #5: Do the Chores for Them

Nobody likes doing chores, and god knows project maintainers have many, so wherever possible, make their life easier. Updating documentation for public facing changes is a good way to gain good will from the project maintainer.

And that was that! The PR containing the actual changes can be found here. Currently I'm waiting for Gabriel to wake up and take a look at my PRs. I hope he finds them adequate, but I'm sure we can resolve any issues if they come up.

Hopefully this summary of my thought process when contributing to a project is useful to someone. Its the sort of thing I wish I had when I first tried.

What's Next

The Bomb Survival Demake project has to wait until the PR gets merged and the new version goes live, so I will probably be back to working on Pando next time.

This marks the completion of my first week of EveryDays!!! I still haven't found the right balance of post length, but it is encouraging that I tend to lean toward too much rather than not enough content. Heres to hoping the next week goes just as well. Maybe this time with fewer post-midnight publishing timestamps.

Till tomorrow,