Jacques Garrigue
Jacques Garrigue
Actually, shouldn't it be just ``. From a type theoretical point of view, there is not much difference between `` and ``, but in ocaml existential type variables are encoded...
> It looks like currently the typing of `(e : ty)` is careful to _not_ unify `ty` with the expected type before `e` is itself type-checked. I'm not sure what...
Actually, there is no need to lower the level here, as this is already done by the call to `rue`. However, there is no need to lower the level in...
Yes, we already checked the history, but I'm still not clear why this would be needed. It is only useful if the original type of `funct.exp_type` escapes in the type...
Thanks for trying. Some changes are good, some are less so: * adding missing labels is probably a good thing, even if this can be a pain * prefixing non-exported...
I'm a bit confused here. I indeed forgot to add this explicitly to the README, but the opam files are updated with the new dependency. IIUK there is a camlp-streams...
> @garrigue the bug here is not about opam, you need to tell the build system (dune in this case) to pick the new library up too. But I believe...
Actually, a trivial solution is to replace `highest_scope` by `highest_scope - 1` in `Ident.scope`. It could be even less. It should just be higher than any normal level. I agree...
That was just a suggestion in case you expected the current code to work. Otherwise, I suppose you need to apply a substitution that replaces the local scopes by the...