Armin Friedl
5a5b9d9e09
All checks were successful
continuous-integration/drone/push Build is passing
Same issue as in
|
||
---|---|---|
.drone.yml | ||
cool.flex.patch | ||
cool.y.patch | ||
Dockerfile | ||
entrypoint.sh | ||
luke.png | ||
README.md |
Lukewarm and the Cool
One chilly day Cool was busy compiling yet another fact, when suddenly it became very sad. It was about its buddy Bodhi. Bodhi 1.4.0 (!!) was ancient and tired. Its mirrors broke a long time ago and it just wanted to retire.
Cool on the other hand was still full of verve and vim, so it went looking for a new companion. Soon enough Lukewarm entered its life. Lukewarm was a slick, quick-witted, up-to-date container and they immediately got along well.
They lived happily ever after. The End.
Totally unrelated image of a cowboy from pngwing.com
Uhm?!
In short: If you are working through the edX Compilers course you probably want Lukewarm instead of the Bodhi VM. It provides all the tools you need for developing your Cool compiler in an up-to-date container image.
Features
Lukewarm is tailored for working through the Cool assignments. It also has several improvements over the provided Bhodi VM:
- All dependencies are pre-installed
- All
Cool
tools, scripts and binaries - Grading scripts are pre-installed and can be executed directly (no workarounds like in the VM)
- Work locally with your favourite tools, then compile and submit from within the container
- Predictable and lightweight
- Use it with CI/CD
- It's based on an up-to-date debian image so you can
apt-get
what you need - Tested with podman too
Where to get it
Lukewarm is published as a container image on https://hub.docker.com/. You can simply run it by:
docker run -it -v $PWD:/class arminfriedl/lukewarm
# or
podman run -it -v $PWD:/class arminfriedl/lukewarm
If you are using SELinux you may need to relabel the host folder so it is accessible from the container:
docker run -it -v $PWD:/class:Z arminfriedl/lukewarm
# or
podman run -it -v $PWD:/class:Z arminfriedl/lukewarm
If you want to debug with gdb you will need to enable:
docker run it -v $PWD:/class --cap-add=SYS_PTRACE --security-opt seccomp=unconfined arminfriedl/lukewarm
# or
podman run -it -v $PWD:/class --cap-add=SYS_PTRACE --security-opt seccomp=unconfined arminfriedl/lukewarm
You can find the repository the container image is built from on https://git.friedl.net/container/lukewarm. Drone does the heavy lifting and the build history can be found at https://drone.friedl.net/container/lukewarm.
How to use it
At first start this will set up all assignments in /class
from the course
distribution
tar.
If you mounted a host folder to /class
you can modify the assignments on your
local machine and build them in the container.
The directory structure looks like this:
.
├── assignments
│ ├── ...
│ ├── PA3
│ ├── PA3J
│ ├── ...
├── bin
├── etc
├── examples
├── handouts
├── include
├── lib
└── src
Edit your assignments in the ./assignments/PA[0-9]J?
folder depending on
whether you want to work on the C++ or the Java version. Each PA[0-9]J?
folder
contains a grade.pl
script. For grading and retrieving your submission code
just execute perl grade.pl
in the current assignment folder from within the
Lukewarm container.
Contribute
If you want to contribute to Lukewarm feel free to send patches to dev[at]friedl[dot]net. Alternatviely, you can issue a pull request on GitHub which will be cherry picked into my tree. If you plan significant long-term contributions drop me a mail for contributor access on the main tree at https://git.friedl.net/container/lukewarm.
Github Users
If you are visiting this repository on GitHub, you are on a mirror of https://git.friedl.net/container/lukewarm. This mirror is regularily updated with my other GitHub mirrors.
Currently I do not intend to move the main tree to GitHub. In case there is popular demand however this will change.