Make and and or work with 0 arguments.

[mlucy]

Now that we have r.args it's extremely annoying that they don't with with no arguments. (Also, other languages like Common Lisp with arbitrary-arity and and or operators have 0-argument version.)

There's another issue for this somewhere but I couldn't find it searching.

[mlucy]

Bumping this to 2.2 not because it's particularly important but because I have a fix for #4132 that also changes this, so it would be nice to squeeze it into the discussion period.

[danielmewes]

What are the results of these terms when called with zero arguments?

[mlucy]

In SBCL (and this makes sense to me intuitively):

* (and)
T
* (or)
NIL

The reason this makes sense intuitively to me is that variadic and works if you define (and a b) <=> (and a b true) and or works if you define (or a b) <=> (or a b false).

[deontologician]

Just one minor data point, we default to true with 0-arg .contains, which could be seen as an or-ish operation

[deontologician]

also, I personally think 0-arg or returning null is yucky, but I don't have a strong opinion

[mlucy]

@deontologician --I think in ReQL 0-arg or should return false. It returns NIL in CL because that's the only false value.

[danielmewes]

I think contains is actually more similar to and, assuming the documentation is correct: "When called with values, returns true if a sequence contains all the specified values."

Allowing 0-argument or with r.or() -> false and 0-argument and with r.and() -> true sounds good to me.

[danielmewes]

Marking as settled with
r.or() -> false and r.and() -> true

[mlucy]

This is in next (the review was completed a while ago and I had just forgotten to merge it in).