File.read 的 Ruby 性能

Question

给定以下脚本：

require 'rubygems'
require 'open-uri'
require 'benchmark'

response = open('http://gdata.youtube.com/feeds/api/videos?q=skateboarding+dog')

outside = Benchmark.measure do
  response_body = response.read
  10000.times do
    response_body.scan(/dog/)
  end
end

inside = Benchmark.measure do
  10000.times do
    response.read.scan(/dog/)
  end
end

puts [outside, inside].map(&:utime).inspect

我得到以下结果：

[1.25, 0.06000000000000005]

为什么每次读取文件的性能都会提高大约 20 倍？

如果我的系统信息很重要：

ruby 2.0.0p247 (2013-06-27 revision 41674) [x86_64-darwin12.4.0]

Answer 1

这是因为在你第一次测试之后，response 被读到最后，而在第二次测试的每次迭代中，read 的结果是微不足道的，这样节省了时间，而且它也只是返回空字符串。所以scan 也很快结束。

irb> response.read.scan(/dog/)
=> ["dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog", "dog"]
irb> response.read.scan(/dog/)
=> []