From: Yohanes Santoso <ysantoso@...>
Date: 2002-12-06T06:18:46+09:00
Subject: Re: speeding up ...

--=-=-=

Mark Probert <probertm@nortelnetworks.com> writes:

> f = File.readlines(ARGV[0])

I think if you use #sysread, it will improve the time too.

jenny:~> ruby tmp.rb rfc2571.txt
                   user     system      total        real
using readlines  5.460000   0.140000   5.600000 (  5.720842)
using sysread  4.720000   0.180000   4.900000 (  5.006272)

The time difference is small since most of the time is spent in
manipulating Strings and Array. However, with larger files, the
difference will become obvious, especially in simple operations like
file copying, etc.

YS. <code below>


--=-=-=
Content-Disposition: attachment; filename=tmp.rb
Content-Description: test code

#! /usr/bin/ruby
#
def use_readlines(fname)
  f = File.readlines(fname)
  
  nl = f.length
  nw = nc = 0
  f.each { |line|
    nw += line.split.size
     nc += line.length
  }
  #puts "    #{nl}    #{nw}    #{nc} #{ARGV[0]}"
end

def use_sysread(fname)
  lines = ""
  begin
    File.open(fname) {|f|
      while buf = f.sysread(4096)
	lines << buf
      end
    }
  rescue
  ensure
    lines = lines.split("\n")
  end

  nl = lines.length
  nw = nc = 0
  lines.each{|line|
    nw += line.split.size
    nc += line.length+1 # for the stripped \n
  }
  #puts "    #{nl}    #{nw}    #{nc} #{ARGV[0]}"
end

require 'benchmark'
include Benchmark
    
if ARGV.size != 1
    puts "usage: #{$0} <filename>"
    exit(1)
end

n=100
bm{|x|
  x.report("using readlines") {n.times{use_readlines(ARGV[0])}}
  x.report("using sysread") {n.times{use_sysread(ARGV[0])}}
}

--=-=-=--