benchmark.rb

def array_unshift
  foo = [:foo, :bar, :bar]
  foo.unshift(:something)
end

def array_concat
  foo = [:something]
  foo.concat([:foo, :bar, :bar])
end

require 'benchmark'

n = 10_000_000
Benchmark.bmbm(15) do |x|
  x.report("array_unshift:")   { n.times do; array_unshift; end }
  x.report("array_concat:")    { n.times do; array_concat;  end }
end

# Rehearsal ---------------------------------------------------
# array_unshift:    2.890000   0.030000   2.920000 (  2.913567)
# array_concat:     3.380000   0.000000   3.380000 (  3.384986)
# ------------------------------------------ total: 6.300000sec

#                       user     system      total        real
# array_unshift:    2.870000   0.020000   2.890000 (  2.898327)
# array_concat:     3.420000   0.000000   3.420000 (  3.415004)

Interesting!

#!/usr/bin/env ruby
# encoding: utf-8

require 'benchmark'

headings = Array(1..25)
rows = [Array(1..25)] * 10_000

Benchmark.bm do |b|
  b.report '#unshift' do
    1000.times do
      r = rows.map(&:to_s)
      r.unshift(headings)
    end
  end

  b.report '#concat' do
    1000.times do
      r = []
      r << headings
      r.concat(rows.map(&:to_s))
    end
  end
end

Yielded

./which_way.rb    
       user     system      total        real
#unshift 63.780000   0.240000  64.020000 ( 72.364274)
#concat 62.340000   0.220000  62.560000 ( 70.409771)

Conclusion: It does not really matter...

👍

Very interesting results!