Implement == about dataarrays.jl HOT 5 CLOSED

If the two arrays are different sizes, I guess we return false even if they contain NAs?

from dataarrays.jl.

All of this sounds reasonable to me. Might be interesting to hear @StefanKarpinski's view on this.

from dataarrays.jl.

Both of @simonster's suggestions seem reasonable. DataFrames has already modified the "contract" of == to include returning NA as a viable option, so this seems like the logical extension of that.

from dataarrays.jl.

Thinking this through a little more, maybe we should return false if any non-NA values are unequal (the arrays would be unequal even if there weren't missing data), NA if all non-NA values are equal but NAs are present (we can't tell if the arrays are equal because of the missing data), or true if all values are equal and there are no NAs.

from dataarrays.jl.

That makes sense and would be consistent with interpreting x == y as equivalent to all(x .== y) under the current Boolean rules for NA.

from dataarrays.jl.