Re: [Caml-list] Constraining abstract type to be of a given subtype

[Dario Teixeira <darioteixeira@yahoo.com>]

Hi Daniel,

> Ok last message I promise.

No problem -- I appreciate the help, thanks!

> I don't know exactly what you want but you may prefer the following
> FOO (see the signature of kind_to_kind)  :

I decided to take a step back and forgo the constraint.  That is, I'm
making the type really abstract, even if in practice all the concrete
implementations of signature FOO will have a kind_t that is a subtype
of Kind.t.

Though it's nice to use the type system to ensure correctness by
construction, in situations (like this one) where one is pushing
against the limits of the type system, the solution can end up
being too cumbersome to use in practice.  Cf, per example, the
pre-3.13 type system tricks to encode GADTs that have been presented
in this list: though brilliant, they haven't gained that much
traction in actual usage because of the associated overhead.

Best regards,
Dario Teixeira