/explanations-hackathon-2020-concordia

Primary LanguageShell

The goal is to wrap your contribution into a Docker image in order to quickly and easily run your work, without having to set-up the environment. Imagine if you want to try contributions from someone and need to install 67,890,987,650 modules/binaries/packages, modify your path etc. And we have many contributions.

Nope. Nope. Nope.

The idea is to wrap everything in a dockerfile and have two scripts: one that builds the docker image from your dockerfile, the other that speciies what needs to be run when one runs your docker image.

Summary

Here is the overall point-of-view of what you will end up with :)

  • A folder named as the targeted language (in my case elixir and another folder for erlang).
  • The language folder contains two files (Dockerfile and build-and-run.sh) and a folder (data).
  • The folder data contains all the files that you need to run your example (script files, resources, whatever), and start.sh specify how to start your example. Some languages can have fancy way to run one of their program, and you want to hide this, encapsulate it.

You will find the hierarchy inside this repository. You can obviously use this repository as a base by forking/downloading it ;).

Write a dockerfile

The Dockerfile will copy, add, download, everything that needs to be done to have a properly set-up environment to run a program in your targeted language.

  • Easy way: Look for dockerhub XXX where XXX is the langage you want to work with. It is most likely that you will find a docker image in docker hub for your language.
  • Hard way: if you can't find such docker image, you will need to manually wrap the runtime and all that is needed to run your program. I can help with that :).

In my case, I had to use elixir, and found this docker image. Sweet \o/ To avoid to execute the command that the image elixir specified (here it starts the elixir shell), I wrote a dockerfile that :

    1. extends the elixir image, that does the black-magic to set-up a functionning elixir environment,
    1. adds the necessary files for my program to work,
    1. specifies the script that will be run, once one starts the image

Here is an example of a Dockerfile that uses elixir, copies whatever is inside the local data folder into a /data folder inside the final docker image, and specify the script ./start-elixir-script.sh that need to be run when one start the image. This is a generic dockerfile and can easily be tuned to match your needs.

FROM elixir

COPY data/ /data
WORKDIR /data

ENTRYPOINT ./start.sh
CMD /bin/bash

Write the start script

This is most likely straightforward and will depend on the language you are using. In my case, I just need to call the elixir executable that was added by the docker image elixir that I extended. Here is the content:

elixir hackhaton_2020_elixir.exs

Write a tiny helper

I wrote a build-and-run.sh script that build my image, and starts it. This is really basic, but by doing so, to reproduce my results, you just have to get my files, type ./build-and-run.sh, and that it is. The docker image is built, the environment set-up, the program started \o.

docker build . -t elixir-custom 
docker run elixir-custom