Pattern matching directly in arguments list

[calebh]

Currently it is not possible to pattern match arguments directly in the function or lambda heads. It would be nice to be able to do this, specifically for tuples and records. This is a very common feature in functional programming languages.

[Qata]

I'd like to suggest adding tuple syntactic sugaring as well so that with this, one could write (fst, snd) rather than ((fst, snd)) when the argument is a tuple.

[calebh]

Functions in Juniper are not automatically curried. The main difficulty is the creation of unnecessary closures on the C++ side. If you have a function f x y = ... and apply f 0 for instance, the return value will be another function that has to capture x. This would be an expensive operation in Juniper as x would have to me moved into a closure (on the stack) and bundled with a new function. At that point the function would immediately be applied.