coq-simple-io VS advent-of-coq-2018

My point about a proof assistant language, let's just say Coq, is focused on pure, referentially transparent functions. There are no IORef like escape hatches required to make things work, although you could probably argue that stuff like this is still pure as, there are certainly escape hatches. Generally speaking, in proof assistants, the program can be self-contained, and is thus "pure".

It's possible to do all that work still in Coq, so that the extracted code can directly be compiled into an executable. One way is to use the coq-simple-io library, which basically wraps the OCaml standard library (including functions for reading and writing files/stdin/stdout) as Coq axioms. For example, I did extraction that way in a previous iteration of AoC: https://github.com/Lysxia/advent-of-coq-2018/blob/master/sol/day01_1.v

It's possible to do all that work still in Coq, so that the extracted code can directly be compiled into an executable. One way is to use the coq-simple-io library, which basically wraps the OCaml standard library (including functions for reading and writing files/stdin/stdout) as Coq axioms. For example, I did extraction that way in a previous iteration of AoC: https://github.com/Lysxia/advent-of-coq-2018/blob/master/sol/day01_1.v