Clarification of phase 3?

[emummel20]

Hey quick question regarding phase 3:

It's my hope to use Scryer Prolog as the logic engine of a low level (and ideally, very fast) Shen implementation.

Do I understand correctly that you are referring to simply replacing the Edinburgh Prolog engine that Shen currently uses in the S-series with Scryer?

After doing so, are there also any plans to create a full port of Shen for Scryer? Because if that were to happen I'm pretty sure it would be my ultimate language....

[mthom]

I plan to create a new port of Shen on top of Common Lisp with Scryer Prolog as its underlying Prolog engine, yes.

[emummel20]

Alright that makes sense, thanks for taking the time! Very much appreciate this project and looking forward to see how it evolves.

[infogulch]

What state do you think the scryer project should be in before its ready to serve as the prolog part of a shen implementation?

[mthom]

It needs GC. I think it's about fast enough now but I'm still making improvements there.

[emummel20]

By the way, @mthom just wanted to follow up on this point "Implementing destructive operations like writing to vectors wouldn't so easy or efficient due to Scryer's emphasis on purity."

I believe the Shentong project was yours as well? Doesn't Haskell also focus on purity and have limited support for destructive operations? Perhaps for future implementers, could you share how you made it work in Shentong?

[mthom]

I used a number of unsafe operations under the ST monad.

[emummel20]

@mthom Let me ask you one last thing on this if I may:

Since shen's type system is implemented in prolog, should it be possible to create a prolog package with shen's full type system, adapted for use directly with prolog, either via comment annotations similar to Mavis or custom type-predicates like this one? Do you happen to know if anyone has tried this or if there are any well constructed sequent calculus type systems for prolog? cc: @triska :) (big fan btw)