Reputation: 980
I have the usual setup: first I define some types, then some functions from those types. However, since there are many ways to formalize the thing I do, I will do it in 3 versions. For simplicity (and to maintain an overview), I want my code in one file. I also want to minimize repetitive code. To this end, a setup w/ 3 Module
s for specific stuff and general definitions in front of them might work -- but not in the type of situation described below:
A general Definition
of function f: A -> B
, accessible in all
sections (or modules)
Module- (or section-) specific definitions of A
f
must be computable in all sections (or modules)
What setup do you recommend me to use?
Upvotes: 1
Views: 100
Reputation: 5811
Require Import Arith.
(* Create a module type for some type A with some general properties. *)
Module Type ModA.
Parameter A: Type.
Axiom a_dec: forall a b:A, {a=b}+{a<>b}.
End ModA.
(* Define the function that uses the A type in another module
that imports a ModA type module *)
Module FMod (AM: (ModA)).
Import AM.
Definition f (a1 a2:A) := if a_dec a1 a2 then 1 else 2.
End FMod.
(* Here's how to use f in another module *)
Module FTheory (AM:ModA).
Module M := FMod AM.
Import M.
Import AM.
Theorem f_theorem: forall a, f a a = 1.
intros. compute. destruct (a_dec _ _).
auto. congruence.
Qed.
End FTheory.
(* Eventually, instatiate the type A in some way,
using subtyping '<:'. *)
Module ModANat <: ModA.
Definition A := nat.
Theorem a_dec: forall a b:A, {a=b}+{a<>b}.
apply eq_nat_dec.
Qed.
End ModANat.
(* Here we use f for your particular type A *)
Module FModNat := FMod ModANat.
Compute (FModNat.f 3 4).
Recursive Extraction FModNat.f.
Goal FModNat.f 3 3 = 1.
Module M := FTheory ModANat.
apply M.f_theorem.
Qed.
Upvotes: 3