From: Pat Eyler <pate@...>
Date: 2002-06-07T01:26:13+09:00
Subject: valgrind and ruby (long)

-----------------------------------------------------------------------
Disclaimer:  I'm not a strong enough programmer (Ruby or C) to be doing
development work on Ruby.  Take everything I say with a grain of salt.
-----------------------------------------------------------------------

I saw another announcement about valgrind (an open-source memory
debugger for x86-GNU/Linux) today, and decided to throw Ruby at it to see
what came up.  After installing valgrind (a painless process), I ran a
small ruby script with it.  Here are the important bits:

[pate@gnupate RWN]$ valgrind --version
valgrind-20020601
[pate@gnupate RWN]$ ruby -v
ruby 1.6.7 (2002-03-01) [i686-linux]
[pate@gnupate RWN]$ cat scanRURL.rb
#!/usr/local/bin/ruby -w

class Scanner
  def initialize(file)
    @file = File.new(file, "r")
    @announcements = Array.new
    @hotTopics	= Hash.new
  end

  HOT_TOPIC_THRESHOLD = 5

  def scan
    scanForAnnouncements
    @file.rewind
    scanForHotTopics
  end

  def outputAnnouncements
    puts "Announcements"
    puts "~~~~~~~~~~~~~~"

    @announcements.each do |announcement|
      puts announcement.subject
    end

    print "\n"
  end

  def outputHotTopics
    print "Hot Topics\n"
    print "~~~~~~~~~~~~\n"

    @hotTopics.each_pair do |topic, occurance|
      if occurance >= HOT_TOPIC_THRESHOLD
	puts topic.subject
	puts "\tOccured " + occurance.to_s + " times"
      end
    end
  end

  private
  #
  # Look for subjects that are anouncements - ANN denotes an announcement
  #
  def scanForAnnouncements
    @file.each do |line|
      if line =~ /^Subject:\s(\[?(ANN|Ann|ann).*$)/
	@announcements << MailSubject.new($1)
      end
    end
  end

  #
  # Look for topics that are discussed more than the HOT_TOPIC_THRESHOLD
  #
  def scanForHotTopics
    @file.each do |line|
      if line =~ /^Subject:\s(.*)$/
	mailSubject = MailSubject.new($1)

	if @hotTopics.has_key?(mailSubject)
	  occurances = @hotTopics[mailSubject]
	  @hotTopics[mailSubject] = occurances.next
	else
	  @hotTopics.store(mailSubject, 1)
	end
      end
    end
  end
end

class MailSubject
  include Comparable

  attr_reader :fullSubject
  attr_reader :subject

  def initialize(subject)
    @subject   = subject
  end

  def <=>(other)
    @subject <=> other.subject
  end

  def eql?(other)
    @subject.eql?(other.subject)
  end

  def hash
    subject.hash
  end

  def to_s
    print "Full subject is: " + @fullSubject + "\n"
    print "Subject is: " + @subject + "\n"
  end
end

if $0 == __FILE__
  if ARGV.size == 1
    file = ARGV[0]
    scanner = Scanner.new(file)
    scanner.scan
    scanner.outputAnnouncements
    scanner.outputHotTopics
  else
    p "USAGE: ./scanList.rb FILE"
  end
end

[pate@gnupate RWN]$ valgrind --leak-check=yes ./scanRURL.rb rubylang.0603
==9689== valgrind-20020601, a memory error detector for x86 GNU/Linux.
==9689== Copyright (C) 2000-2002, and GNU GPL'd, by Julian Seward.
==9689== Estimated CPU clock rate is 298 MHz
==9689== For more details, rerun with: -v
==9689==
==9689== Conditional jump or move depends on uninitialised value(s)
==9689==    at 0x80659D0: mark_locations_array (gc.c:359)
==9689==    by 0x8066523: rb_gc (gc.c:1008)
==9689==    by 0x8098629: load_file (ruby.c:857)
==9689==    by 0x809830B: proc_options (ruby.c:727)

[elided some errors]

==9689== Conditional jump or move depends on uninitialised value(s)
==9689==    at 0x80659F7: mark_locations_array (gc.c:364)
==9689==    by 0x8066523: rb_gc (gc.c:1008)
==9689==    by 0x8098629: load_file (ruby.c:857)
==9689==    by 0x809830B: proc_options (ruby.c:727)

[elided some errors]

==9689== Conditional jump or move depends on uninitialised value(s)
==9689==    at 0x8065B6C: rb_gc_mark (ruby.h:606)
==9689==    by 0x8065A37: mark_locations_array (gc.c:379)
==9689==    by 0x8066523: rb_gc (gc.c:1008)
==9689==    by 0x8098629: load_file (ruby.c:857)

[elided some errors]

==9689== Use of uninitialised value of size 4
==9689==    at 0x8065B81: rb_gc_mark (gc.c:452)
==9689==    by 0x8065A37: mark_locations_array (gc.c:379)
==9689==    by 0x8066523: rb_gc (gc.c:1008)
==9689==    by 0x8098629: load_file (ruby.c:857)

[elided some errors]

==9689== Use of uninitialised value of size 4
==9689==    at 0x8065B81: rb_gc_mark (gc.c:452)
==9689==    by 0x8065A37: mark_locations_array (gc.c:379)
==9689==    by 0x8066523: rb_gc (gc.c:1008)
==9689==    by 0x8098629: load_file (ruby.c:857)

[the errors continue ... skipping to the end]

==9689== More than 30000 total errors detected.  I'm not reporting any
more.
==9689== Final error counts will be inaccurate.  Go fix your program!
==9689==

[output of command elided]

==9689==
==9689== ERROR SUMMARY: 30000 errors from 37 contexts (suppressed: 0 from
0)
==9689== malloc/free: in use at exit: 379208 bytes in 8308 blocks.
==9689== malloc/free: 82108 allocs, 73800 frees, 3639387 bytes allocated.
==9689== For counts of detected errors, rerun with: -v
==9689== searching for pointers to 8308 not-freed blocks.
==9689== checked 4608320 bytes.
==9689==
==9689== definitely lost: 0 bytes in 0 blocks.
==9689== possibly lost:   20 bytes in 1 blocks.
==9689== still reachable: 379188 bytes in 8307 blocks.
==9689==
==9689== 20 bytes in 1 blocks are possibly lost in loss record 1 of 6
==9689==    at 0x40042D79: malloc (in /opt/lib/valgrind/valgrind.so)
==9689==    by 0x806550D: ruby_xmalloc (gc.c:97)
==9689==    by 0x80869AC: top_local_setup (parse.y:4800)
==9689==    by 0x807ACDC: rb_yyparse (parse.y:286)
==9689==
==9689== LEAK SUMMARY:
==9689==    possibly lost:   20 bytes in 1 blocks.
==9689==    definitely lost: 0 bytes in 0 blocks.
==9689==    still reachable: 379188 bytes in 8307 blocks.
==9689== Reachable blocks (those to which a pointer was found) are not
shown.
==9689== To see them, rerun with: --show-reachable=yes
==9689==