I think I learned an extremely important life hack today
when you want to consume information from a white paper from PL research academia and you never went through a PHD program / can't understand their greek grammars, check to see if they gave a talk on the paper instead!
Conversation
This tweet is brought to you by the rage I experienced when I first read this white paper and wasn't able to properly understand it:
cs.cornell.edu/andru/papers/u
Followed by my delight when I realized there is a talk for the same research!
5
3
70
Oh nice, this is literally the exact challenge I've been fighting with in my own lang over the last few days. Thanks for the link!
1
3
I'm still at the stage of figuring out the exact semantics, but effects in Tao (github.com/zesterer/tao/) *should* generalise just fine to bi-directional effects. Might be a few weeks until I have something worth showing the world though.
1
2
A somewhat problematic thing is that I'm trying to implement effect handlers on top of pure universe mapping (Haskell style), and this ends up needing a few odd extensions to typical effect systems to work properly without resorting to 'dirty' stuff like Haskell's unsafePerformIO
1
1
What is “pure universe mapping”?
1
1
main having type World -> World. The pure core of monadic IO. I don't want to be hiding behind impurity at some level of the language just for the sake of ease of implementation.
1
1
Ohhh. Have you seen Runners? I dunno if they would help with the dodgeyness you are talking about: math.andrej.com/2019/10/28/run
(full disclosure, have not had the time to learn them)
Interesting, this seems like I've so far been implementing monadic IO: effectful operations are intrinsics that take the world as an arg. & return it again, never copying or dropping. My difficulty had been strapping this to effect handlers, but I think I solved that last night.
1