Is there a way to safely redeclare a symbol?

Question

I often find myself experimenting in the REPL and I will say something like:

subset Bar of Int where * %% 57;

Then I play around with checks on the Bar-ness for things for a bit.

Everything is happy, until I realize that I want to change the definition of Bar.

If I just redefine Bar, I get a Redeclaration of symbol exception.

I tried using MONKEY-TYPING and augment like this:

use MONKEY-TYPING;
augment subset Bar of Int where * %% 37;

But that netted me the same error.

Why do I want this? So I can iterate on my subset (or class, or other symbol) definitions, while reusing the tests I've already typed that are in my history.

Answer 1

The REPL has its shortcomings. It is an elaborate construction of EVAL statements that try to work together. Sometimes that doesn't work out.

I guess the best we could do, is to introduce a REPL command that would make it forget everything it has done before. Patches welcome! :-)

Answer 2

I think the REPL does part of its magic by EVAL-ing each new input in a new nested lexical scope. So, if you declare things with my then you can shadow them with declarations entered later:

my subset Bar of Int where * %% 57;
sub take-Bar(Bar $n) { say "$n is Bar" }
take-Bar 57;

my subset Bar of Int where * %% 42;
sub take-Bar(Bar $n) { say "$n is Bar" }
take-Bar 42;

If you omit my, then for subset and class declarations, our will be used, and since our is actually my + adding the symbol to the enclosing package...; turns out if you delete the symbol from the package, you can then shadow it again later:

subset Bar of Int where * %% 57;
GLOBAL::<Bar>:delete;
subset Bar of Int where * %% 42;
42 ~~ Bar;

NOTE: These results are just from my experiments in the REPL. I'm not sure if there are other unknown side effects.