This repository is structured as follows:
| File name | description |
|---|---|
| README.md | This file. |
| pipelines/ | Contains pipeline files |
| images/ | Dockerfile for respective images in their directory & corresponding test task & scripts. |
Let's say you are working on images/clang-toolchain, and you realize clang-tidy-10 was not installed.
Add a test in images/clang-toolchain/test.bash. Look around, there is a test
called it_has_executables and it runs type -p over a list of things that
we expect to be in the path. Append clang-tidy-10 to that list.
We use a [two-stage build][docker.multi-stage] where a simple "test" stage runs some assertions over the built image. Testing is integrated with building of the image:
docker build --pull --target test images/clang-toolchain
echo $?The flag --target test ensures that docker build doesn't optimize away the test stage because it's not used to produce the final output.
If the return status is 0, your test is wrong because it's not broken, otherwise you've written a test that successfully failed.
Fix the test by adding clang-tidy-10 executable to the image
-- in this case, we add clang-tidy-10 to the list of packages we apt-get install in the Dockerfile.
In other cases, the package name might differ from the executable name.