Simplify on show is kinda annoying

[MasonProtter]

Currently, if I write

julia> using SymbolicUtils, Test; @vars a b;

julia> a * b * a
((a ^ 2) * b)

julia> @test a * b * a == a^2 * b
Test Failed at REPL[33]:1
  Expression: a * b * a == a ^ 2 * b
   Evaluated: ((a ^ 2) * b) == ((a ^ 2) * b)
ERROR: There was an error during testing

Because we simplify the output of Base.show, errors testsets tricky to debug. notice it says Evaluated: ((a ^ 2) * b) == ((a ^ 2) * b)

One can use showraw to see what the actual representation is, but I think maybe we shouldn't be simplifying on show.

[shashi]

An alternative fix is to simplify both sides in ==. We kinda need some form of this to check equality of expressions (isequal compares structure, so we can keep it that way and have == compare the "value" (I forgot what the julia convention is though)).

scmutils simplifies on show, but this was kinda disappointing:

1 ]=> (= (+ 'x 1) (+ 'x 1))
#| #f |#

1 ]=> (eq? (+ 'x 1) (+ 'x 1))
#| #f |#

1 ]=> (eqv? (+ 'x 1) (+ 'x 1))
#| #f |#

1 ]=> (= (- (+ 'x 1) (+ 'x 1)) 0)
#| #f |#

1 ]=> (eq? (- (+ 'x 1) (+ 'x 1)) 0)
#| #f |#

1 ]=> (zero? (- (+ 'x 1) (+ 'x 1)))
#| #t |#

[shashi]

Haha I found this in my old code:

;;; Utility to compare equality of two expressions
(define (expr-eq? a b)
  (equal?
      (write-to-string (simplify a))
      (write-to-string (simplify b))))

[shashi]

Closing this since the printing is unsimplified in tests now.

[MasonProtter]

@shashi I feel that especially since this package is trying to be generic and more agnostic about the users' usecase, simplify in show is a bad idea.

I think that one major problem is that simplify makes interactively devising rules to manipulate a Term harder because the object's output at the REPL is not actually what you're trying to manipulate.

---

PS: sorry I haven't had much time for this package lately. I'm hoping to get back into it soon!

[shashi]

Ok sure, feel free to make the change by just changing the default value of simplified_show[]