a lack of invalidation for some cases of constant redefinition?

[nsajko]

These two both throw in show:

abstract type AbstractType end
foo(x::AbstractType) = x
const AbstractType = 1
methods(foo)

abstract type AbstractType end
struct Bar <: AbstractType end
const AbstractType = 1
supertype(Bar)

cc @Keno

Pointed out on Discourse by Benny:

• https://discourse.julialang.org/t/introduce-a-command-line-option-to-warn-on-constant-redefinition/126213/5?u=nsajko

[Keno]

The error in show is a bug of course, but foo not being re-evaluated is intentional on the julia semantics side and is a job for Revise.