Static type system

[edolstra]

Nix won't be complete until it has static typing.

[shlevy]

Can you expand on this a bit? I think some efforts toward a better type system can be valuable, I'm just not sure whether the static/dynamic distinction even makes sense for nix.

[tel]

It's interesting to think what kind of typing scheme would be needed for Nix. There's apparently a pretty heavy need for anonymous isorecursive records, something like

Mu self: { override: {rs} -> { self, rs } }

[Ericson2314]

"Nix won't be complete until it has static typing." What a beautiful way to put it :).

@tel row polymorphism works well with mu types for this. mu types + row polymorphism sometimes severely limit the ways the attribute set can be functionally extended / unioned / etc, but if that becomes an issue, I think/hope there are sound ways to "break" the mu types and gain more expressiveness.

[domenkozar]

http://www.haskellforall.com/2016/12/dhall-non-turing-complete-configuration.html?m=1

[thufschmitt]

In case someone follows this issue but not the mailing-list, I opened a funding campaign for this issue at https://www.gofundme.com/typing-nix. You are all welcome to donate :)

[alexeymuranov]

@regnat, weren't the any other way to sponsor Nix development without sponsoring GoFundMe with 7.9% + $.30 (at least $269 currently)?

[Wizek]

Huh? How come this is closed @edolstra?

[edolstra]

Because it's not realistically going to happen (and I'm cleaning up some backlog issues).

[Fuuzetsu]

🚫 dropped