Use stringified expressions instead of left and right

[TimoFreiberg]

Hi and thanks for the helpful library!
I was annoyed a few times now by the panic message only showing (left == right) instead of my actual arguments to the assert_eq call and had to add a custom message describing the arguments.
This change also helps in remembering which side was the expected or the actual value (which happens to me...)

Slightly ugly things in this MR:

---

The following message is duplicated 4 times:

assertion failed: `({} == {})`

---

Some existing tests now also verify this new feature, I think it would be slightly prettier to make these tests also pass x and y as arguments (so they don't check two things at once) and make a new test verify this new feature

[TimoFreiberg]

Any interest in this?

[aquarhead]

I'm thinking about the same thing, but maybe just let caller provide different names for left and right is simpler (and more straightforward?)

[cormacrelf]

@aquarhead It would actually be more complicated. You would need to add macro arguments to what are purportedly drop-in replacements for assert_eq. This breaks switching back with a one-liner. I like the stringified expressions; you can just configure it by introducing a let binding.

[rivy]

@cormacrelf , can you give an example of what you mean by configuring with a let binding?

[cormacrelf]

@cormacrelf , can you give an example of what you mean by configuring with a let binding?

let one = ...;
let very_fancy_two = ...;
assert_eq!(one, very_fancy_two);

[rivy]

@cormacrelf , can you give an example of what you mean by configuring with a let binding?

let one = ...;
let very_fancy_two = ...;
assert_eq!(one, very_fancy_two);

I don't understand. Does that show anything different from "left" and "right"?

[cormacrelf]

@rivy you’re confused because I said “can” instead of “could”. It doesn’t now, but it would if the two expressions passed to assert_eq were stringified and those strings were used instead of left and right. It would say < one / very_fancy_two > instead.

[rivy]

@cormacrelf , thanks for the clarification.