Behold: The Janitor Machine



Hi there!

Let me tell you a story in which I use a €200 Chromebook laptop to check out the full Firefox project, and then compile it in under 12 minutes.

The Hardware

For this experiment, I bought the cheapest viable laptop I could find: An 11,6" Asus Chromebook C200MA-KX018 with an Intel Celeron N2840, 4GB RAM and 32GB SSD (it’s even cheaper on, but since I live in France I had to get it from

Note: Chromebooks run on Chrome OS, which is basically just Google Chrome with a desktop interface, so it’s not for you if you need apps that aren’t entirely web-based (like for example LibreOffice, Sublime Text, or most video games, unless you hack your Chromebook and use its Linux core directly).

Choosing right: I followed the Wirecutter’s amazingly useful guide on how to buy a good Chromebook. My instincts told me to buy their featured Dell Chromebook 13 (now discontinued), but that’s more like €600 in France, while I wanted something really cheap yet somewhat fast. Basically the guide says don’t buy anything with less than 4GB RAM, and make sure it has an Intel processor, but since their budget choices weren’t available or too expensive in France, I had to do my own research. I made the trade-off of a smaller screen (11" / 720p while the Wirecutter won’t go below 13" / 1080p) and looked for the cheapest option with still a decent processor (by checking scores on

The Deeds

On this inexpensive hardware, I was able to:

  • Quickly check out the full Firefox repository,
  • Comfortably hack on its source code using an IDE and its Terminal (with an actual Bash, colors, Vim, SSH, Git completions…),
  • And then compile the whole thing in less than 12 minutes.

Note: This also works with projects like Servo, Thunderbird and Chromium.

And I intend to continue using my €200 Chromebook for all my contributions to Open Source going forward.

The Catch

This sounds too good to be true, doesn’t it? Well, if you’re somewhat familiar with the Janitor project, you already know the trick: I’m not actually checking out or compiling things on my local hardware, but instead I’m using the Cloud9 IDE web service with custom development containers (based on specially crafted project Dockerfiles) that run on a scalable cluster of powerful dedicated servers.

TODO: Explain the Janitor better.
TODO: Conclusion, and say I’ll report back with updates on how well this experiment goes.


Nice, thx @Jan for the benchmark and the useful advice! :+1:t4: