Currently we run 7 different jobs:
- tests linux
- tests macos
- tests windows
- rustfmt
- clippy
- examples
- documentation
With this change I reduced them to 4 and hopefully sped them up.
The total execution time is limited by the tests, especially linux that calculates coverage. Having separate jobs for clippy and rustfmt (which take a very small amount of time) is a waste of energy.
With this PR:
Introduced a new cargo profile, `ci`, that should create smaller sized binaries and reduce the our cache usage.
I changed the test runner for macos and windows to [nextest](https://nexte.st/), which should be faster and is specifically designed for CI.
I merged all smaller tasks in a single job, misc, the steps clearly identify what is being tested so it shouldn't affect clarity.
Switched to using the [rust-cache](https://github.com/Swatinem/rust-cache) GH action, this simplifies our work by no longer having to worry about which directories to cache, rust-cache handles all that for us.
~~The bors task should also be modified, I'll get to it as soon as I have time. I believe it should be possible for us to have a single workflow described and have it both be the normal CI and the bors test.~~
João Borges