Jan Midtgaard

I've gotten a bit further on this now. First of all building without parallelism seems to help reproducibility: `dune build -j1 test --instrument-with mutaml` I also had more luck with...

Thanks - and apologies for the delay! :pray: I completely agree that having to wait for, e.g., 10 tests to complete is sub-optimal if the error is already happening in...

I just thought of another thing: :thinking: - When running a list of tests without the `--verbose / -v` option, users still need enough context printed to understand, e.g., an...

> > Do we want labeled and numeric stats printed progressively as well, or should they continue to go at the bottom? > > (I personally think either is fine,...

This is probably going to be a bit more complex than I initially thought... 🤷 For example, the opam repo still offers a mix of `v`-prefixed versions of `apron` as...

CI is failing on - Alpine - OpenSuse - Oracle - x86_32 - FreeBSD - macOS Note: none of these were changed (so far) in the PR. - Oracle and...

Compared to the above status - FreeBSD is failing package install (as Oracle has been removed) - Alpine, OpenSuse, FreeBSD, and macOS are failing the `javac -version` check The former...

OK, the latest 3 commit attempt to do so.

The error message is actually being printed for all failing platforms, it is just a bit lost in the output. For example, here's a quote from an Alpine log: ```...

There's something strange going on with x86_32 - CI passed in 3e9c57b and we got a green light https://opam.ci.ocaml.org/github/ocaml/opam-repository/commit/3e9c57ba0e4b6738a9c324694b8deef36b2c7bd2/variant/extras,x86_32-ocaml-4.14,conf-openjdk.1 - CI then failed in aabaa4e with the missing `server` error...