Re: [Toss-devel] Hierarchical Terms and Parametric Polymorphism

[Lukasz K. <luk...@gm...>]

Hi.

> (1) drop the proposal, go back and start again with the simplest type
> system with subtyping

Well - this may be the best option, but let's keep it as the last resort, right?

> (2) drop parametric polymorphism and the idea of a functional-like
> typed programming language, instead go for TFS-based language for
> expressing hypergraphs (still a programming language based on
> rewriting)

I do not like this idea. Sentences naturally have term-like structure when
first parsed, and types seem to be natural constraints (maybe to me, sure).
Of course - things later also have hypergraph structure, and if we can get
it earlier, then great. But I can also just parse sentences to normal terms
(= trees) and have another step to derive hypergraphs from sequences of
sentences = tree terms. Looks like quite a natural way to me, and easier.

> (3) have distinct TFS-variables (with GLB) and type-variables (with LUB)
>
> I understand that you oppose (2), just wanted to make sure. I'm not
> interested in (1). I'll work towards (3) then.

I guess that sums it up nicely. I hope we can still get something nice
from it - else we can revert as in (1) and think about making hypergraphs
from terms as the "next step" - after parsing and types and stuff. Maybe
that could interest you in case (3) does not work. But for now let's stay
optimistic and try out (3) - maybe it'll do just fine!

Lukasz