std: lessThan and greaterThan between signed and unsigned

[shawnl]

It is a deviation from C, but I think we should consider making this the behavior
of the operators. See #2133

With vectors this behavior can be kinda expensive if the type sizes are 8, 16, 32, 64. (this does not implement vectors, but it is easy to do so with my other patch set). But premature optimization is the root of all evil, and I think its better to prefer readability of the code, and then catch minor performance stuff later.

This PR is partially to spark discussion, and demonstrate well-defined behavior.

[andrewrk]

I'm OK to merge this right now, but I want to propose an alternate API:

• make Compare enum public and rename to CompareOperator
• add EQ and NE to CompareOperator enum
• pub fn compare(a: var, comptime op: CompareOperator, b: var) bool
• delete lessThan, lessThanOrEqual, greaterThan, and greaterThanOrEqual

[andrewrk]

I'll move this back to 0.5.0 if feedback is addressed

[data-man]

Maybe

pub const CompareOperator = enum {
    lessThan,
    lessThanOrEqual,
    equal,
    greaterThan,
    greaterThanOrEqual,
};

?

[shawnl]

@data-man Yes, .equal is also useful, I didn't think of that.

[data-man]

Oh! I missed notEqual :-)

[data-man]

@shawnl Please also add notEqual.
It's will be useful in many cases. E.g. DB ORMs, calculators, parsers, etc.
So developer will use this std enum instead of reinventing a wheel.

[shawnl]

@data-man notEqual is just !compare(a, .equal, b)

[data-man]

Sure.
I propose new element in enum.

So developer will use this std enum instead of reinventing a wheel.

[shawnl]

We could even support these comparisons between floats and integers, and floats of different bit widths. The only case I'm not sure about is when you have an integer that would round to a specific float (with a positive exponent) but is a different value, I think that should return not equal.

[andrewrk]

Current plan for the language specification (#75) is to allow comparisons between floats & ints.