David Allsopp

Cross-referencing https://github.com/ocaml/opam/issues/3038#issuecomment-326979380 but just because `GIT_TERMINAL_PROMPT=0` may not work in every single instance doesn't mean we shouldn't be doing it!

It's been there for a long time, well before Windows was being thought (added in https://github.com/ocaml/opam/commit/f3f6465b1fa406573917d46bc02f261f0906590f)! We should probably be conservative about turning the warning off universally, but I agree...

Notes from dev meeting: - The fact that older opam would silently interpret this as `>=` means we don't feel this can be reliably used in `opam` files - However,...

> > Extend ocamltest to have tests marked as "not-to-be-run-in-parallel" > > I've been using "make parallel" for trunk testing without realizing that some tests should not be run in...

I don't think opam should get into the business of trying to kill sub-processes when an error has been detected (this is likely to be a stability disaster on Windows),...

> > I don't think opam should get into the business of trying to kill sub-processes when an error has been detected (this is likely to be a stability disaster...

As noted in the end of the description in #5176, if we got to do the upgrade process, we'd probably opt to skip have the `>=` invariant trick, but it...

From the requested native English perspective, `with_locked` only leaves me wondering what a "locked" is, I'm afraid. FWIW, I would say `with_mutex_locked` (which is unnecessarily longer, especially as `Mutex.with_mutex_locked`) or...

> Basically it seems that from an english perspective we can't play with the module name for `with_*` Indeed, even if the module name is always given, it just reads...

Checks for 2.2.0: - `--with-mccs` should cause an error in this case (that should be working?) - `make cold` should force this - i.e. mandate that it builds `opam` with...