Results 1654 comments of Gabriel Scherer

So here is the explanation for the issue. A wrong implementation of the check would be that when we see an application `t u` (where `t` itself may be an...

This is a nice trick, should we consider including it in the standard library? (This could be exposed as a wrapper after the fact, or hidden as new variants of...

Introducing it in CamlinternalFormat and relying on type-directed lookup could be a way forward, by very intentionally not trying to provide a one-size-fits-all solution.

This sounds like a good L3/M1-level internship topic to me, with a useful goal and a large amount of GADT programming in perspective.

I would expect `Printf` to also get such an interface eventually. I think that it is reasonable to expect the same argument-list definition to work for both modules. This would...

I still see benefits in writing something like `scanf ibuf (fun [foo; bar; baz] -> ...)`, but I don't know if people actually use Scanf nowadays except for throwaway code,...

I believe that this is okay and we haven't had expression of interest of other potential reviewers, so if the CI is still okay I will go ahead and merge.

Yes, AppVeyor is currently broken on all PRs.

I implemented a refactoring that separates the lock types in two -- a first attempt: https://github.com/gadmm/ocaml/compare/caml_plat_lock_non_blocking3...gasche:ocaml:caml_plat_lock_non_blocking4 What do you think? My problem with the current approach is that it relies...