In Julia, why use a Pair over a two element Tuple?

Question

When should I use a Pair over a two element Tuple and vice versa?

Answer 1

Mainly when you want to visually highlight, that there exists a relation between the first and second elements. For example, consider the following Dict construction

Dict(:a => 1, :b => 2)

Despite you can use Tuple, but it is more obvious here, that there is a relation between keys (:a and :b) and values (1, 2).

Actually, it is more than just obvious, it gives you the possibility to use the following definition

julia> Dict(((:a, 1) => (:b, 2)))
Dict{Tuple{Symbol, Int64}, Tuple{Symbol, Int64}} with 1 entry:
  (:a, 1) => (:b, 2)

Compare it with

julia> Dict(((:a, 1), (:b, 2)))
Dict{Symbol, Int64} with 2 entries:
  :a => 1
  :b => 2

By using only tuples it's hard to build a dictionary that can have tuples as keys.

Another, slightly different example can be found in join functions from DataFrames.jl

innerjoin(a, b, on = [:City => :Location, :Job => :Work])

Here it is easy to understand, that you bind City column of a with Location column of b. Though, tuple syntax would be fine, but Pair notation looks more readable.

On the other hand, if you are building two-dimensional point there is no need in using Pair, usual Tuple is more than enough. Point(x => y) definition looks strange, compared to Point(x, y).

So, the rule of thumb, if there is a relation between the first and second element use Pair, if they are independent of each other use Tuple.