I think for authors we should provide something like travis-racket to demonstrate multiple possible CI systems, including GitHub Actions, for exactly the reasons @sorawee mentions. I think GitHub Actions is likely to be popular among Racket programmers for exactly the reasons we’re using it, plus the fact that we’re using it, but @sorawee is right that we shouldn’t push people to it.

@samth you mean for their own packages - makes total sense.

For those following along, Racket might be soon accepted into oss-fuzz. https://github.com/google/oss-fuzz/pull/3054

@samth By “we should provide”, do you mean the racket core team should say something like “if you’re writing a racket package and want to use github actions as your CI system, use action XXX”?

Racket was merged into oss-fuzz , The work it remains is to implement a fuzzer in C/C++ for Racket3m and RacketCS.

I mean we = “us in this group” should make sure there’s an easy way to do this, as Greg has done for Travis

@samth I wrote an action that provides CI for a single racket package https://github.com/jackfirth/racket-package-ci-action

Also @popa.bogdanp wrote an action that installs racket, allowing later actions to do arbitrary racket stuff like run raco setup or raco pkg install

There’s some overlap between those two solutions that we should figure out how to handle

@notjack could you just use @popa.bogdanp’s to install?

@samth do you mean use that one within the implementation of my action, or use it instead of my action entirely?