Jacques Garrigue
Jacques Garrigue
I have tried to answer the requests. For the different behaviors, I discriminate on the kind of exception expected, assuming a special role for `Equality`. It was not clear to...
@Octachron Do you mean that you would prefer going back to the original behavior of `mcomp` ? I suppose this is fine, since we are now able to distinguish between...
I have rebased, and re-enabled the check of univars in `mcomp`.
Merged into trunk and cherry-picked to 5.3.
This is clearly sound, but this is a hack. Note that the problem is a bit different that with `%raise`. There it could be seen as a either a consequence...
@yallop: This could already be done in two steps thanks to the relaxed value restiction: ```ocaml let r : 'weak list ref = ref [] let _r_unsafe : 'a. 'a...
> When we check a pattern `K (p1, p2)`, it may be possible to detect that the value of `p1` may in fact not influence the typability of `p2`. Intuitively,...
I see. In checking mode we are working on a restricted form of patterns, in particular without type annotations, so this could be doable. This is still tricky because dependencies...
The way you state it is a bit confusing, because for this statement to be correct `a1, a2...` must be seen as distinct from `'a1, 'a2, ...`, more specifically they...
I do want to look into this, in particular the principality part, but I see that I can be slow. So in the end you don't need to wait for...