Krish Agarwal

According to section 11.5.3 of the standard, "For decorations, comparison operations for equality = and its negation shall be provided, as well as comparisons >, dac > def > trv...

In the test suite we have, `DecoratedInterval(2, 1) == nai() == true` but `nai() == nai() == false` Aren't they contradicting?

Master ``` julia> @which Interval(3,4)/Float32(4) /(x::Number, y::Number) in Base at promotion.jl:314 ``` Branch ``` julia> @which Interval(3,4)/Float32(4) /(a::Interval{T}, x::S) where {T, S

Yes, even I think it should return the same type as the type of the input interval, I will add tests for it.

Yes, I noticed this too, will be fixed.

``` julia> x = (3f0..4f0) / 2.1 [1.42857f0, 1.90477f0] julia> x = (3f0..4f0) / big(2.1) [1.42857f0, 1.90477f0] julia> typeof((3..4)/2) Interval{Float64} julia> Interval{BigFloat}(2,4)/2 [1, 2]₂₅₆ ```

Yes!

Resolved the merge conflict, good to merge, I think.

Rounding is handled by the `@interval` macro, which generates correctly-rounded intervals. ``` julia> x = @interval(1//10) Interval(0.09999999999999999, 0.1) julia> x.lo == x.hi false ```

`interval(Inf, Inf)`, `interval(-Inf, -Inf)` and `Interval(NaN, NaN)` are returning argument errors now (according to the tests)