Actually the source_url link was broken. Here you go 
adamu
DaAnalyst
Congrats! An impressive effort.
Btw, looking at the docs..
`def next(fsm: FSM(TrafficLight)): FSM(TrafficLight) = `
` fsm_send(fsm, pair(:timer, unit()))`
Why FSM( TrafficLight) in the header if TrafficLight is already defined as a FSM?
Also why let for assignment? Is it for symmetry with function’s def then =? If so, then why the function’s = in the first place? record definition requires a do/end, why not stick to it elsewhere (like in Elixir :)?
IMO, the only thing more annoying than let is := (which you luckily don’t use)
0
mudasobwa
Creator of Cure
This is a dependent type, if I correctly understood the question.
Well, in BEAM languages = means match. People get accustomed to this and they do expect 1 = x to compile (assuming x has been assigned to 1 before.) Cure does not support in-place matches (yet?).
Also, it made way simpler to experiment with in-place matches without breaking things, because x = 1 is currently would raise at compile stage (thanks let.)
THe syntax, though, is not yet 102% settled. = in function declarations might have changed later too. Parser is isolated and might be easily modified.
2
mudasobwa
Creator of Cure
slouchpie
Looks amazing! The examples link on the website returns 404. https://cure-lang.org/examples/
Where can I see some examples?
2
mudasobwa
Creator of Cure
I promise I’ll fix the site sooner or later 
Also, the standard library is all written in Cure itself cure-lang/lib at main · am-kantox/cure-lang · GitHub
2
Onor.io
I like the idea of seeing Dependent Types getting more adoption. I think they’re a feature that should be adopted into more languages.
2
mudasobwa
Creator of Cure
Easier to say than to implement
The issue is probably almost nobody actually wants to sacrifice the development speed to get a proof of code correctness, like the one SMT-solver provides. In 99% of cases just let it crash works.
2
eileennoonan
This may be true but I’m excited about the prospect of being able to e.g. safely code a medical device on the BEAM.
My big question would be what problems does this solve that Gleam doesn’t?
Gleam also uses strict GADTs and exhaustive pattern matching. In theory if it compiles, then correctness has been proved.
But I don’t know enough about dependent types to know how they differ. Also I love the idea of FSMs being a first class citizen.
1