which_is_easier_to_read.rb

which_is_easier_to_read.rb

%w(id foo under_score bar batz_score)

["id", "foo", "under_score", "bar", "batz_score"]

[:id, :foo, :under_score, :bar, :batz_score]

which_is_faster.txt

Author: Ken Collins  
Date: December 16, 2010  
Summary: Arrays  

System Information
------------------
    Operating System:    Mac OS X 10.6.5 (10H574)
    CPU:                 Quad-Core Intel Xeon 2.66 GHz
    Processor Count:     4
    Memory:              8 GB
    ruby 1.8.7 (2010-08-16 patchlevel 302) [i686-darwin10.4.0]

"literal w/symbols" is up to 28% faster over 1,000 repetitions
--------------------------------------------------------------

    literal w/symbols    0.00223708152770996 secs    Fastest
    literal w/strings    0.00305581092834473 secs    26% Slower
    %w()                 0.00312900543212891 secs    28% Slower

xbenchmark.rb

require 'rubygems'
require 'bench_press'

extend BenchPress

author 'Ken Collins'
summary 'Arrays'

reps 1_000


measure "%w()" do
  %w(id foo under_score bar batz_score).each { |attrib| }
end

measure "literal w/strings" do
  ["id", "foo", "under_score", "bar", "batz_score"].each { |attrib| }
end

measure "literal w/symbols" do
  [:id, :foo, :under_score, :bar, :batz_score].each { |attrib| }
end

I seem to recall a rant by Ara Howard about use of symbols being evil.

String#to_sym for arbitrary strings is evil because it adds to the symbol table, so it can cause your process to grow as it runs. If you're typing %w(foo bar), than these aren't arbitrary strings, they're important to your code. So you'll probably be using them again. Thus, this is probably not an evil use case.