[Bug] Not handling all cases during pattern-matching in lambdas

[AZMCode]

As of now, using a non-exhaustive pattern in a lambda returns the following error message

carp: src/Qualify.hs:220:55-108: Non-exhaustive patterns in lambda

With zero indication of where the offending lambda is located in code, unlike other compile time error messages.

[scolsen]

This is reporting a case where we actually aren't handling all possible cases in the compiler. Do you have some example code that triggered this?

[AZMCode]

Sure, let me whip up some code to reproduce it. A moment.

[AZMCode]

Crashes the carp interpreter with version 0.5.5

(reduce &(fn [(Pair x y) elm] (Pair (+ x 1) y)) (Pair 0 0) &(Array.repeat 16 &(fn[] ())))

Not sure if it's possible to reduce this example even further

[scolsen]

Thanks a ton, @AZMCode! I'll take a look and see if I can get a fix in the next day or two.

[AZMCode]

Thank you for taking the time.

[AZMCode]

After a long time not using the language, I came back to it for funsies, stumbled upon this bug again, and decided to reduce the code further. Here's an updated example that triggers the exact same bug:

(def f (fn [ (Unit) ] ()))

[AZMCode]

I don't know if this code is valid at all, but it surely shouldn't crash the interpreter/compiler

[scolsen]

This short repo case is great to have, thank you @AZMCode! I am hoping to spend some more time on Carp again soon and can try to fix this.

(wound up getting busy with other things as usual, after pledging to work on this before, but it is still on my radar!)