Untitled

require 'nokogiri'

require 'benchmark'

require 'thread'

require 'open-uri'

 

# for JRUBY: export JRUBY_OPTS="--1.9 -J-Xmx2048m"

# JRuby 1.6.7 doesn't support File.write yet in 1.9 mode. Just doing some cache here - not important

File.open('test.html', 'w'){|f| f.write(open('http://www.sec.gov/Archives/edgar/data/1129623/000095012310064472/d73737ddefm14a.htm').read) } unless File.exists?('test.html')

content = File.read('test.html')

 

class Executor

  def initialize(pool_size)

    @available_threads = Queue.new

    pool_size.times { @available_threads << nil }

    @threads = []

  end

 

  def execute

    @available_threads.pop

    @threads << Thread.start {

      yield

      @available_threads << nil

    }

  end

 

  def shutdown

    @threads.each &:join

  end

end

 

Benchmark.bmbm do |x|

  x.report('Parse HTML') do

    executor = Executor.new(10)

    50.times { executor.execute { Nokogiri.HTML(content) } }

    executor.shutdown

  end  

end

 

# Results for JRuby 1.6.7:

# Rehearsal ----------------------------------------------

# Parse HTML   4.501000   0.000000   4.501000 (  4.501000)

# ------------------------------------- total: 4.501000sec

#

#                  user     system      total        real

# Parse HTML   0.927000   0.000000   0.927000 (  0.928000)

 

# Results for MRI 1.9.3-p

# Rehearsal ----------------------------------------------

# Parse HTML   3.800000   0.050000   3.850000 (  3.853381)

# ------------------------------------- total: 3.850000sec

#

#                  user     system      total        real

# Parse HTML   3.780000   0.010000   3.790000 (  3.788918)

 

# They perform about the same if I change to Executor.new(1), ie, using a single thread

 

# Am I doing any mistake in this code?