GitHub LightHouse

A Lighthouse you say?

Through the tempests of issues we've built a lighthouse to shed a light on projects sinking under the pressure of fame and not having enough crew to fix all the issues. If you feel kind-hearted enough to help a fellow dev, pass by our Lighthouse.

Does it run locally?

Yes it does!

But how do I start it?

Run in docker

• define the github token in the environment variable GITHUB_TOKEN in the docker-compose.yml file
• run docker-compose up --build -d

Run locally

• run npm install --production
• define the github token in the environment variable GITHUB_TOKEN (export GITHUB_TOKEN=YourRoken)
• run npm start
• open index.html in your favorite web browser

Development and production

Install the dependencies using npm install

Start the webpack watcher by running npm run webpack Changes on the files will regenerate the css and js resources.

As this project uses CSS Modules the stylesheets are injected in the html page and not compiled in a file. For production you might want to have a unique css file for catching purposes.

So when you are happy with your changes check if the linter is happy by using npm run linter no errors means that he's happy, this guy is not very chatty

Then generate your nice js bundle and css file with the command npm run build it will check if the code is linted and then build the frontend.

Frontend

The frontend is developed using React.js and CSS modules. Each component is styled with it's own or shared style.scss file.

We serve it using github pages because it's free

Backend

We use heroku to serve the logic of the app and a scheduled job run it for a list of famous frameworks to speedup the loading time. It generates a js files containing the lists of repos for these frameworks and stores it in the github repo. It's a hacky way to store data without paying but it works.

How does it works ?

Basically it uses black magic to determine if a repo is trendy/useful enough to worth your time and if there isn't a lot of brave dev helping out already. Then it sends you a handpicked selection of repos that you might want to help

Whaa black magic you said ?

Yeah well not exactly, we count the number of open issues weighted by the number of day since there last update and we divide this so called "open issues score" by the code frequency. It's the average number of commit per hour done by all the project's contributors. We obviously multiply this number by over nine thousand. Then we multiply it by it's "trendyness" wich is the max between the number of stars and the number of fork.