unfortunately, the language I'm implementing calls functions "procedures", so things get hairy fast
Conversation
(the original language does not allow return values, but I added it on top)
1
1
Tbh if the functions are side-effecty, I'm completely on board with the 'procedure' naming!
I tried to get rid of all greek-letter-named stuff in Pikelet and just call things 'functions' and 'pairs' etc.
1
If I had it my way (tuples and non-primitive return types) there'd be no need for side effects, but alas...
1
1
(I don't have anywhere near enough time to implement these 🤣)
1
ugh - I feel you
1
Deadline is tomorrow evening. I think I shall have to accept that closures, tuples, and ADTs aren't gonna make the cut
1
but, but, what about dependent types, ; ____ ;
1
good luck implementing recursive types without a heap, that's why I gave up
1
1
ack, yeah that is terrifying - need to know the size somehow
1
kind of cool problem though! kind of would be cool to have something like that in pikelet… just not when it's due tomorrow 😢
It would be cool if you could inductively define fixed size slices perhaps. If you had unboxed types you'd need to know it was finite - which potentially is ok if you are in a total language? Figuring out if you can lay it out inside the stack size might be trickier though.
2
Show replies