Josh Berdine

Easier after #406

I don't have a simple enough case to make a good example. I can point to this [dune-workspace](https://github.com/facebook/infer/blob/d082f364488863169986b2b868915b86978b9460/sledge/dune-workspace) file that relies on "sledge" and "sledge+flambda" switch already being setup (manually)....

Just for info, we would appreciate such a full-speed plus debug info binary mode.

I think that an important aspect of `raise_notrace` is that, since it does not record the backtrace, it can be used to raise and handle exceptions within a handler that...

Yes, that seems good to me. Sorry for the delay, I missed this.

FWIW, the [manual](https://caml.inria.fr/pub/docs/manual-ocaml/libref/Stdlib.html) uses the term "structural equality": > e1 = e2 tests for structural equality of e1 and e2. I'd default to using the same terminology. I would hesitate...

FWIW, in the past I ran across so many bugs due to this that I added two type-incompatible modules, `HashMap` and `HashMultiMap`, both as thin wrappers of `Hashtbl`, and shadowed...

I encountered the same linking error, and https://github.com/antoinemine/apron/pull/28 fixed it for me.

For the record, I haven't needed to set environment variables to install using opam with gmp and mpfr installed via homebrew. Note though that e.g. the conf-mpfr package [expects](https://github.com/ocaml/opam-repository/blob/d8f637a9c8f939f85f62e386726599c0955e2c6d/packages/conf-mpfr/conf-mpfr.2/opam) homebrew...

In the projects I am involved in that use llvm, a vendored version of conf-llvm using this change is used with no trouble. As far as I understand, @mseri saw...