Vincent Laviron

> Could it still be possible to optimize the compilation in cases where the extensible types are defined top-level? I imagine that this would cover a lot of standard cases...

Note that `"%reraise"` has some of the same issues as `get_backtrace`. This is basically what `reraise` does: ```ocaml let reraise exn = let bt, prev_exn = get_backtrace_and_exn () in let...

> Do we actually want to support this? I don't know if we want to, but we already do (this module definition is accepted without this PR)

> Edit: a [github search for "lang:OCaml /type .* = .* true/"](https://github.com/search?q=lang%3AOCaml+%2Ftype+.*+%3D+.*+true%2F&type=code&p=5) suggests that the only use of rebinding `true` in the wild is exactly to re-export the bool type...

Cross referencing ocaml/ocamlfind#69, which is dealing with the same issue.

Thanks ! I've taken your suggestions into account, and I plan to click merge on Monday if the CI doesn't find anything to complain about.

(I'm fine with merging immediately too)

I'm busy this week, but I'll try to look at your PR next week.

Maybe we could take that as a hint that `Fun.id` should be implemented as a regular value (`let id x = x`) rather than a `%`-primitive ? I don't think...

Yes, this is exactly what we want. This will allow us to revert the changes to the testsuite related to `Fun.id` to their previous state, which is likely the expected...