isless for DataFrameRow

[bkamins]

isless for DataFrameRow ignores column names, but == checks them.

For NamedTuple isless is undefined so there is no inconsistency there. But we have it for DataFrameRow.

We have two options I think:

1. disallow isless for DataFrameRow (I think this is the way to go)
2. change == to ignore column names

[nalimilan]

isless is defined for named tuples when they have identical names. I guess we could do the same and throw an error when names differ. Is that used in a common pattern?

[bkamins]

it is not that common, but possible. I was misguided by the error message for NamedTuples:

julia> (a=1,b=2) < (a=3,c=4)
ERROR: MethodError: no method matching isless(::NamedTuple{(:a, :b),Tuple{Int64,Int64}}, ::NamedTuple{(:a, :c),Tuple{Int64,Int64}})

but forgotten that NamedTuple has key names in type signature. I will make a fix as you suggest as it makes most sense!