David Allsopp

Only problem with that is that it doesn't cover `--root`. It's reasonable to assume that this only happens to a user who knows about roots (because they have set `OPAMROOT`...

(putting in 2.2 as the Windows shell stuff touches this message anyway, and it's not a big change!)

Just marking that it’s a regression which definitely needs fixing. I haven’t looked at the headers, but it should probably move back to address_class.h, with whatever knock-on changes that requires.

Agreed - please can this PR not acquire new features. PRs against your own fork are a better way to start previewing the next stage while this part gets reviewed.

This can be closed.

I'm not sure that an example can actually be concocted which would trigger this, but it reads slightly strangely that the `Effect` module initialises a runtime structure (in this case...

The use of `%HOME%` also makes it compatible with opam-repository-mingw

Looks like a tweak is needed to the runners. `opam init` is also incorrectly displaying `~\.opam` as the install location.

Thanks for the review, @AltGr! I've hopefully fixed the CI - beforehand, `OpamStd.Sys.home` was falling to the default of `.opam` in the CWD which was a dreadful default for Windows,...

> A side effect is that when you're in a directory with a local switch, there's no way to temporarily override the switch to a global one. `opam exec --switch="foo"...