Email or username:

Password:

Forgot your password?
Top-level
Devil Lu Linvega

@CapitalEx OOOoooh, I think you're right. I hadn't made the connection with the halting problem, but that's true, I guess a specific type is indeed the solution here.

4 comments
Capital

@neauoire :: (oops broke that one reply by accident)

Interested in what ideas you explore within Tal's limits. FP languages have the advantage of a strong type system and pattern matching. If your type is finite and each step reduces the given inputs to their base cases, you can say confidently that "this loop terminates".

Idris does this, but it also means some finite functions can't be statically resolved and are marked as partial.

Devil Lu Linvega

@CapitalEx I'm an absolute beginner when it comes to building type systems, so this is all fresh to me.

I've been going through the handlful of papers and blog posts on types/concat languages. This morning @toxi gave me a bunch of very interesting links on the topic: merveilles.town/@toxi@mastodon

I'll keep reporting my findings, I've started a stub on the topic here: wiki.xxiivv.com/site/type_syst

Karsten Schmidt

@neauoire @CapitalEx I just re-watched Jon Purdy's talk and it seems even 6 years later the number of people _actively_ researching/experimenting in this field is still extremely small! But this also means you're automatically kind of at the forefront and I'm looking forward to see what you'll come up with...

I'm still somewhat ambivalent about how far down the rabbit hole with respect to types in a Forth-style lang one should go before starting to compromise other core aspects like overall simplicity and actual utility. It's the same balance I'm striving to find in my TypeScript work. I do like structural typing and finding it very useful in general, but there're also a number of cases where I just learned to stay away from spending days/weeks on defining/refining/maintaining super complicated type constructs for very little practical gain. Types and type systems can be total nerd snipes and (IMHO) sometimes just spending more time on validation (where needed), documentation and explanations/reasoning of design decisions is time better spent than wasting it on super elaborate types. Sometimes it's just swapping one type of footgun for another and YAGNI... :)

@neauoire @CapitalEx I just re-watched Jon Purdy's talk and it seems even 6 years later the number of people _actively_ researching/experimenting in this field is still extremely small! But this also means you're automatically kind of at the forefront and I'm looking forward to see what you'll come up with...

Devil Lu Linvega

@toxi @CapitalEx I promise I'll try to not get sucked in too deeply that I can never re-emerge. :moomin_hello:

Go Up