Ruby Benchmark: String#bytesize vs String#size

benchmark.rb

# String#bytesize vs String#size

require 'benchmark'

N=20_000_000

short_utf8_string  = "ääää"
long_utf8_string   = "äääääääääääääääääääääääääääääääääääääääääääääääääääääääää"
short_ascii_string = "aaaa"
long_ascii_string  = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
short_mixed_string = "aäaä"
long_mixed_string  = "aäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäaäa"

Benchmark.bmbm do |x|
  x.report("utf8.short.size"){N.times{ short_utf8_string.size }}
  x.report("utf8.short.bytesize"){N.times{ short_utf8_string.bytesize }}
  x.report("utf8.long.size"){N.times{ long_utf8_string.size }}
  x.report("utf8.long.bytesize"){N.times{ long_utf8_string.bytesize }}

  x.report("ascii.short.size"){N.times{ short_ascii_string.size }}
  x.report("ascii.short.bytesize"){N.times{ short_ascii_string.bytesize }}
  x.report("ascii.long.size"){N.times{ long_ascii_string.size }}
  x.report("ascii.long.bytesize"){N.times{ long_ascii_string.bytesize }}

  x.report("mixed.short.size"){N.times{ short_mixed_string.size }}
  x.report("mixed.short.bytesize"){N.times{ short_mixed_string.bytesize }}
  x.report("mixed.long.size"){N.times{ long_mixed_string.size }}
  x.report("mixed.long.bytesize"){N.times{ long_mixed_string.bytesize }}
end

result.txt

Conclusion:
- Don't call size on a string which might contain utf8 characters.
- You'll be probably fine using bytesize on everything, but if you
  know there are only going to be ascii characters you may be even
  better off just using size.

Output:
                           user     system      total        real
utf8.short.size        2.910000   0.010000   2.920000 (  2.916396)
utf8.short.bytesize    2.530000   0.000000   2.530000 (  2.539148)
utf8.long.size         3.140000   0.010000   3.150000 (  3.138874)
utf8.long.bytesize     2.270000   0.000000   2.270000 (  2.267662)
ascii.short.size       2.030000   0.000000   2.030000 (  2.032265)
ascii.short.bytesize   2.460000   0.000000   2.460000 (  2.464350)
ascii.long.size        2.200000   0.000000   2.200000 (  2.201368)
ascii.long.bytesize    2.470000   0.010000   2.480000 (  2.480259)
mixed.short.size       2.780000   0.000000   2.780000 (  2.789172)
mixed.short.bytesize   2.920000   0.010000   2.930000 (  2.928942)
mixed.long.size        3.340000   0.000000   3.340000 (  3.347672)
mixed.long.bytesize    2.610000   0.000000   2.610000 (  2.609181)