As it looks only a rather small part of QuickChick ever worked on Windows - large parts require ocamlc, perl, you name it and even in the cygwin environment, where this is there, it doesn't find it. The test file chosen for smoke test "PluginTest.v" was obviously not very effective.
So I decided to disable QuickChick on Windows - even for older picks - in order to avoid confusion. Are there objections against this?
See: (https://github.com/QuickChick/QuickChick/issues/269) where I also informed the QuickChick team about this.
fine by me. This sounds like something to pass on to Inria & the OCaml foundation though, since they want OCaml and friends to work on Windows...
one could argue QuickChick is a prestige project of sorts for Coq (and it's being used in industry IIRC)
I think I can fix it for the next release (end of February) at least for the compiled from sources variant, where the only issue is that QuickChick must find it. For the installed version it won't happen soon (one would need to make it easier to relocate an ocaml installation).
Btw.: I asked a few people for an "OCaml on Windows Hackaton" with the goal to make Coq Platform work without hacks, but it didn't happen as yet.
you could request a "OCaml on Windows" session here: https://github.com/coq/coq/wiki/CoqWG-2022-02
Even if you're busy with other things during the hackathon, flagging up the problem might rally other people to the cause.
The main effort would be to merge a sufficient part from the opam_mingw repo into the main repo and to discuss the ocaml changes in this repo. I would think this would require attendance of people with write rights in the main opam repo.
Last updated: Jan 30 2023 at 10:03 UTC