When an OPAM-CI build fails, you conveniently get a Dockerfile + script to reproduce the failure locally. Which is great! I used to use this all the time to repro failures. After you get the repro, it used-to-be, that you could look at the output, and there would be hash-ids of intermediate Docker images in the log. You could then use “docker exec” (I think it was) to fire up the image with a shell right there, so you could debug interactively.
But this last bit seems to be gone (I read that Docker changed their build-engine, so it no longer produces that), and after an hour of googling and reading various people’s posts, I don’t see anything that’s replaced this functionality. Does anybody here have any pointers to how to get into the midst of a failing opam-ci docker build?
I guess I can just convert the dockerfile into a shellscript and run that, but geez, the docker approach was so … convenient and self-contained, it’d be great if I could continue to use it.
Thanks in advance for any advice.