Results 397 comments of Florian Angeletti

At first glance, the current code is too syntactic. The compiler version (which may also fail in corner cases(?)) looks like this: ```ocaml let extract_option_type env ty = match get_desc...

That sounds sensible. I am wondering if text expressions should be computed lazily or not, but that properly doesn't matter in practice.

The project is lightly maintained I fear, I added a support for `1--5` in the parser in the github version. I will cut a new release by the end of...

Synchronizing the opam file with the one used in opam is probably always a good idea, thanks !

Any non-trivial work, even on the internals of the compiler, should be credited. In case of doubts, it is easier to add a Changes entry.

I am not sure what is your issue with the LGPL license? As far as as I can see, DkML does not modify the source code of the compiler, does...

I missed those patches, sorry. Then the distribution of DkML is covered by the second part: > or a modified version of the OCaml Core System that is distributed under...

The trunk parsetree `constant` is a record with la ocation information, thus the error message complains about seeing a variant constructor when a record was expected.

Symmetrizing the error message sounds like a good idea, and I expect the information to be easily available.

With type-directed disambiguation, a syntactic constructor doesn't have a type before name resolution, but it can be still be classified inside the variant kinds. The expected type is still present...