You can do this with upto and downto Fixnum's methods:
array = [0, 1, 2, 3, 4] last_index = array.size - 1 3.upto last_index do |i| puts array[i] end # => 3, 4 last_index.downto 3 do |i| puts array[i] end # => 4, 3
PS. as speed benchmark, iteration with rotation faster
array.rotate(3).each {|e| puts e}
benchmark:
require 'benchmark' array = Array.new(10000000) { rand(1...9) } last_index = array.size - 1 Benchmark.bm do |x| x.report 'upto' do 10000.upto last_index do |index| a = array[index] + 1; end end x.report 'downto' do last_index.downto 10000 do |index| a = array[index] + 1; end end x.report 'rotate' do array.rotate(10000).each {|e| a = e + 1 } end end # RESULTS: # user system total real # upto 0.680000 0.000000 0.680000 ( 0.681932) # downto 0.680000 0.000000 0.680000 ( 0.679752) # rotate 0.590000 0.040000 0.630000 ( 0.622901)
but, as memory benchmark, iteration by array indexes less memory hungry, especially on big array sizes:
require 'memory_profiler' array = Array.new(10000000) { rand(1...9) } last_index = array.size - 1 { upto: -> { 10000.upto last_index do |index| a = array[index] + 1; end }, downto: -> { last_index.downto 10000 do |index| a = array[index] + 1; end }, rotate: -> { array.rotate(10000).each {|e| a = e + 1 } }, reverse_rotate: -> { array.reverse.rotate(10000).each {|e| a = e + 1 } } }.each { |desc, code| puts "#{desc.to_s} => #{MemoryProfiler.report(&code).total_allocated_memsize.to_s}" } # RESULTS (in bytes): # upto => 0 # no additional memory allocation # downto => 0 # no additional memory allocation # rotate => 80000040 # implicitly copied array 1 time # reverse_rotate => 160000080 # implicitly copied array 2 times