Perl Cookbook
by Tom Christiansen & Nathan Torkington

Here are the changes made in the 9/98 reprint:


<orthography, high>
ix.line5    run-up "Bytes to Read 254".  Line 4 now reads

7.15 Determing the Number of Bytes to Read.................254


<orthography, high>
x "9.11 Sysmlink" changed to "9.11 sysmirror" and
  "10.14 Redifining
	There is a spelling error in the 10.14 title!
	It's redefining, not redifining.

<orthography, medium>
xxvi.1.8	discussing how to make
		^^^^^^^^^^
    should read: "we discuss" instead of "discussing"

xxxii:	Add "Jason Ornstein" right before
	"Jason Stewart" in the second set of reviewers.  Add "Eric
	Watt Forste" right after "Eric Eisenhart".

<technical, high>
12.3.4          $seen{$1}++;
        replace $1 with $byte.  This is in the first code example
        on the page.  NOT the second.

<orthography, high>
15.2.2		the Switches'' section
    add quotes in front of  Switches

<technical, high>
19.4.-2		$big = "\u$little"
	    should be
		$big = "\U$little"
    make sure this is the first display, not the second

<technical, high>
19.4.-1		$little = "\l$big"
	    should be
		$little = "\L$big"
    also make sure this is the first display, not the second;
    the second one begins "To alter ..." and is correct as is.

<technical, high>
35.-1.-6	while (<DATA) {
			    ^
    should have a close angle bracket:
		while (<DATA>) {


<technical, high>
47.-2   The printf has spaces after the % and before the .1f in four
        places.  Those spaces must be removed, leaving "%.1f"
        instead of "% .1f" as it is now.


<technical, high>
52.7		@chars = ( "A" .. "Z", "a" .. "z", 0 .. 9, qw(% ! @ $ % ^ &
*));
	% appears twice inside the qw( ... ).  Delete the first one.
<orthography, high>
57.-1.1		The sin, cos, and atan2 functions in
	the "a" in atan2 is not in Courier.  Make it so.


<technical, high>
67.4.1          The ternary ("hook") operator discussed in
                    ^^^^^^^ wrongly in Courier
    should be
        The ternary conditional ("hook-colon") operator discussed in
            ^^^^^^^ make in Garamonde

<technical, medium>
68.-1   There for 3 instances of print here, the first two
	with wrong indents.  This:

        for $factor ( sort { $a <=> $b } keys %factors ) {
                print "$factor";
            if ($factors{$factor} > 1) {
            print "**$factors{$factor}";
            }
            print " ";

    Should change to this: (the changed lines are marked
    with a *;  do not include the *, please.)

        for $factor ( sort { $a <=> $b } keys %factors ) {
   *        print "$factor";
            if ($factors{$factor} > 1) {
   *            print "**$factors{$factor}";
            }
            print " ";


72 para. -2, line 3: change "(Before then" to "By 2038"
and line 4:  remove the close parenthesis.

<technical, high>
86.1.3      Change
                $time = gettimeofday-$t1;
            to
                $time = gettimeofday - $begin;

<technical, medium>
95.-4.-2	the ternary ("hook") operator
		    ^^^^^^^ courier
	should be
		the conditional operator ("?:")
		    ^^^^^^^^^^^ Garamond
					  ^^^^^^^ courier

<technical, medium>
98.-1.1	The <FM> should be <FH>.

<technical, high>
102.3.4	    This line
		unless ($seen{$item})
	    Should be
		unless ($seen{$item}) {

<technical, medium>
127.2.1	The "memorizing" word in the comment should be
	"memoizing" without the "r".  Yes, I'm serious.


<technical, low>
133.3.{1,4,5}  Change all occurences of $name to $size here

<technical, low>
133.5.1   Change $name to $size here also

133. the code line was moved to its own paragraph

	print "$food is $food_color{$food}.\n


<technical, medium>
136.4.-2  Change
	    save substantial time and memory
	  to
	    save substantial memory

<technical, low>
136.-3.5   Delete
		Banana is a food.
		Martini is a drink.

<technical, medium>
137.-4.1    The "while (<FILE>) {" line is indented 4 extra
	    blanks, and needs to come out flush with the margin.


<technical, medium>
141.4.-2    The clsoing curly that's directly below the
	    print is in the wrong column.  It should be
	    in the same column as the foreach 4 lines above it.


<orthography, medium>
162.3	This should be in a table with the right-hand side
	in all Garamonde, save for the variables and special
	characters which should be in Courier.

<wording, medium>
167.-1.-2  This:
    not at the edges.  For example, "thistle" would be found, but
	should be
    not at the edges.  And while, "thistle" would be found,

<orthography, high>
    The word vis-a-vis has the wrong accent over the a.  It's
    a grave, not an acute accent.  Therefore, it should properly be
    the correct vis-à-vis instead of the incorrect vis-á-vis.
		    ^                                  ^
		should be grave			   not acute

<orthography, low>
170.1.-1    The "unlike" here should be separated from the preceding
	    word "generation" by an em dash.

<orthography, medium>
170 and 171 The typesetting of Nth is inconsistent across these
	    two pages.  Sometimes the N is italic; at other times,
	    it is not.	I tend to think it shouldn't be italic at all on
	    170, making it the way they are on 171.  But an alternative
	    change is to italicize all the Nth occurrences on 171.
	    I think if you do, that the th superscript should also be
	    in italics.  Hm, the title of 6.5 has a non-superscripted
	    th -- that should be fixed.

177.-2.3    The comment on the second while is copied from the
	    first one, but no longer applies.
	# line is between BEGIN and END
	# inclusive.
	    Should be
	# operate only between first and last line, inclusive.

<technical, medium>
178.2.3	    The comment on the second while is copied from the
	    first one, but no longer applies.
	# line is between BEGIN and END, but not same line
	    Should be
	# operate only between first and last line, but not same
    (Yes, this is essentially the same issue on two successive pages)

<technical, low>
179.1.2	    The comment
	# previous command fails
	    should be
	# previous command FAILS

<orthography, medium>
179.6.3	    The eof() in the code should not be slant courier,
	    but rather just regular courier like the part all
	    around it.

<technical, medium>
184.1.1	    The /0 modifier should be /o -- that's a letter, not
	    a digit.

186.-3.2	"<I\s*[^>", "*** GET RICH ***", or "+5-i".
	These double quotes are all vertical Roman double quotes,
	not vertical Courier double quotes.


<technical, high>
193.-2.3    The second "<b><i>" on this line should be "</b></i>"
	    instead.

<technical, high>
193.-2.-2   The second "<b><i>" on this line should be "</b></i>"
	    instead.

<technical, low>
193.4.1     Change "from HTML correctly" to "from all possible HTML
	    correctly".


<wording, medium>
201.-2.-1   You keep removing my joke:
		having pity on those who come after you.
	    which really is supposed to read:
		having pity on those who come after you--before they do.
	    And I wish it were put back. :-(


<wording, medium>
211.3	    My original wording with "can" got altered to this:
		enhancements are that it ignores anything that's not
		a plain text file, automatically expands compressed or
		gzipped files, recurses down directories, searches
		complete paragraphs or user-defined records, looks in
		younger files before older ones, and adds underlining or
	    Which to my ear implies it should always to that.  But
	    it doesn't.
		enhancements are that it can ignore anything that's not
					 ^^^^^^^^^^
		a plain text file, automatically expand compressed or
					         ^^^^^^
		gzipped files, recurse down directories, search
			       ^^^^^^                    ^^^^^^
		complete paragraphs or user-defined records, look in
							     ^^^^
		younger files before older ones, and add underlining or

<technical, high>
219	The snippet on ``Remove directory components from program name
	and arguments'' has been hopelessly mangled.  It should read:
	    ($name = " $0 @ARGV") =~ s, /\S+/, ,g;

<technical, high>
219	    This is wrong:
		$initial = m/^\S+\s+(\S)+\s+\S+/ ? $1 : "";
	    It should be
	       ($initial) = m/^\S+\s+(\S)+\s+\S+/ ? $1 : "";
	       ^        ^


<orthography, low>
237.2.1	    Change "less than symbol" to "less-than symbol".

<technical, high>
237.2.-1    Change ""> $ARGV" to "< $ARGV".

<technical, low>
237.2.-1    Change "this use of open" to "this use of magic open".


<technical, high>
243.2.-1    Delete the line:
    % perl -i.old -ne 'print unless /^START$/ .. /^END$/' bigfile.text

250 frombooktech

<wording, low>
263.1.2     Change "data structures" to "data structure".

<orthography, high>
263.-2.-4   Change
    Perl wouldn't give a ``print on closed filehandle warning even though
	    to
    Perl wouldn't give a "print on closed filehandle" warning even though
			 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
			 note change of quotes to vertical primes,
			 and put that quoted string all in Courier,
			 as on the previous page.


<wording, low>
264.1.-2    Change "on the rare system" to "on the increasingly rare system".

<wording, medium>
264.2.-2    Change
		you lock filenames,
	    To
		you lock file names,
			      ^^^^^
			      place the word "names" in italics.

<wording, high>
264.3.1	    Change "use it lock" into "use it to lock".

<wording, low>
264.-4.-1   change "if various" into "should various".

<technical, medium-high>
276.-4.-2	Change:
    Perl also provides unbuffered I/O operations: sysopen, sysread,
    syswrite, sysseek, and close.  Buffering, sysopen, and close are
    discussed in Chapter 7.
		into
    Perl also provides unbuffered I/O operations: sysread,
    syswrite, and sysseek, all discussed in Chapter 7.

<technical, high>
279.3.1	    Change "/n" into "\n".  Oh my.

<technical, high>
283.5.3         You can also use POSIX::clearerr
        No you can't.  POSIX::clearerr says "use IO::Handle::clearerr
        instead".   Remove this sentence, and

284.6   also remove the
        See Also bit on page 284 that points people to the POSIX
        module.

<orthography, low>
285.5	    The italic instances of N here should not be slanted.

<wording, medium>
285.6.2     change "proportionate" to "proportional"

<wording, high>
285.6.2     "to size of the file" should be "to the size of the file"
	    or "to the file's size".


<wording, high>
293.1.2	    Change "read 90 just bytes" into "read just 90 bytes".

<orthography, high>
293.-4.-2   The em dash follwoing GUI should actually
	    be a close parenthesis, which is otherwise missing.

<technical, high>
311.-2.-1   The -C operator was forgotten!  Add below -A:
		-C    ctime   Same for inode change time (not creation)

<orthography, high>
312.1.1	   The word "stat' there should be in Courier, not Italics.

<technical, medium>
313.2.2	   If you possibly can, change
		list the same inode number,
	   Into
		list the same inode number (like vi and nvi in the
		earlier table)
	   And make sure vi and nvi are in italic.


<technical, medium>
316.-3	   The line with
		next if $! =~ /^Interrupted/;
	   could stand a comment:
		next if $! =~ /^Interrupted/;     # ^Z and fg

<technical, high>
316.-3	   The "\en" in the code example must be "\n" instead.

<orthography, medium>
329.-4.2    The sentence-ending period after ".bak" is in the wrong
	    font.  It should be Garamonde, not Courier.

<orthography, high>
329.-3.2    The comma is wrong here!!!!
		to be "lib/," whereas

	    It needs to be
		to be "lib/", whereas
	    And for that matter, the second vertical primes are in the
	    wrong font.  They are in non-courier, but should be courier.


<technical, high>
329.-1	    The name of the program in the A-header must be "symirror" not
	    "symlink".	Notice the name on the next page.


<wording, high>
350.2.1	    Delete the word "again" from "repeat again".

<technical, low>
350.middle  All three instances of "3 , 5" have an extra space before
	    comma that should not be there.

<orthography, medium>
350.-4.2    The function "time" is in garamonde, but needs to be in
	    courier.

<wording, high>
351.1.3     This sentence is twisted:
		The requirement of the backslash that the argument
		begin with the backslashed type character is sometimes
		restrictive.
	    Make it read:
		A backslash in a prototype requires that the argument
		actually begin with the literal type character, which
		can sometimes be annoying.

<technical, low>
351.2.1	    Move the # WRONG over left and align with the # RIGHT below

<technical, high>
351.4.1	    Unlike the one above it, the second mypush line needs both its
	    arrays backslashed.  Make it read:
		mypush( @{ $x > 10 ? \@a : \@b }, 3, 5);	# RIGHT
				     ^     ^                    ^
				     add these                  tab this

<technical, low>
356.-2.-2   Move the grow() over to the right to line up with the
	    word local right above it!

356	    There is a spelling error in the 10.14 title.
            It's "redefining", not "redifining".

<wording, medium>
365.5.1	    Change "it may look as though a reference were litte more" into
	    "a reference might look like little more", since the
	    same construct is already too prevalent.

<technical, low>
366.-4.1    Change
		$rv = \@array;
	    into
		$aref = \@array;

<technical, low>
367.-4.-1   Change
		It's best to use IO::Handle->new() for them
	    into
		It's best to use Symbol::gensym() or IO::Handle->new() to
				 ^^^^^^^^^^^^^^^^^^^
		generate them


<wording, medium>
375.3.-2    Change
		Symbolic references are usually a bad idea.
	    into
		Symbolic references to variables are usually a bad idea.
				     ^^^^^^^^^^^
<technical, low>
379.4.-1     Currently it says:

            ($a, $b, $c, $d) = (1 .. 4);        # initialize
            @array =  (\$a, \$b, \$c, \$d);     # refs to each scalar
            @array = \( $a,  $b,  $c,  $d);     # same thing!

	Add a line making it:

            ($a, $b, $c, $d) = (1 .. 4);        # initialize
            @array =  (\$a, \$b, \$c, \$d);     # refs to each scalar
            @array = \( $a,  $b,  $c,  $d);     # same thing!
            @array = map { \my $anon } 0 .. 3;  # allocate 4 anon scalarresf


<technical, low>
392.-1.-2   change "insert an empty node" to "insert into an empty node".


<wording, high>
406.4.-2    The non-word "barework" with a `k' should really be
	    "bareword" with a `d'.

<wording, low>
406.5.1     Change
		A better implementation would look like:
	    into
		A better implementation might look more like:


<orthography, high>
414.2.-3    "When you--or more likely your system administrator--install
this module, its components go"
	    changed to

"When you or (more likely) your system administrator installs this module,
its components go"


<orthography, medium>
414.4.1	    changed
		So sometimes none of the @INC directories works out
	    to
		So sometimes none of the @INC directories work out

<wording, high>
414.-3.2    Change "directories to @INC" into "directories", since
	    otherwise it won't parse.

<wording, high>
417.-1.-2   There's a duplicate word at "also also".  Remove either
	    of them, but not both. :-)


<orthography, high>
433.3.-1    	``for later,'' or ``for nobody'':
	    delete the comma.


<technical, high>
453.-2.2	Change
		    use vars qw($NAME $AGE $PEERS);
		to
		    use vars qw($NAME $AGE @PEERS);

<orthography, high>
454.2.1		Change "same name" to "same names" to accord
		with "globals" earlier in the sentence.


<technical, high>
504.-2.1    Change "DataDumper" to "Data::Dumper".


<technical, high>
505.1.-1    The output line is wrong.   It should be:
		Tom Toming: ARRAY(0x73048) ARRAY(0x73e4c)

508.3		$sth = DBI->prepare(SQL)
	should read
		$sth = $dbh->prepare($sql)

<technical, high>
523.3.1	    This:
		print colored("venom lack\n", "red on_black");
	    should be
		print colored("venom lack\n", "red", "on_black");


<orthography, high>
529.1.1     Put a space between "-1parameter" to make it "-1 parameter".

<technical, low>
529.-2.-2   The arguments here should be in parens. Make this:
		ReadMode 'noecho';
		$password = ReadLine 0;
	    Be this:
		ReadMode('noecho');
		$password = ReadLine(0);

<technical, high>
531.2.-1    The last line of the code reads
                $term->addhistory($seed_line);
            But it should read
                $term->addhistory($cmd);

<technical, medium>
555.1.1	        Prefix this page with:
	    A high-level workaround is given in Recipe 19.6.

<technical, medium>
555.3.-1       Add a See Also reference to Recipe 16.10

<technical, high>
561.1.-1	You can interpret its contents as described in Recipe 16.2.
	replace "16.2" with "16.19"

<technical, high>
567.4.1		# system("prog args 2>&1 1>tmpfile");
	this line should be aligned with the next, as it is in the
	first code chunk on the page


<technical, medium>
576.-4.-1	Add a See Also to Recipe 19.6

<technical, high>
577.4.1		open(FIFO, "> /path/to/named.pipe")     or die $1;
	replace "$1" with "$!"


<technical, high>
583.1.1		you also have to use @signame and %signo
	delete "also", "@", and "%"

<wording, high>
584.3.1        Change "systems ties" to "system ties".


<wording, high>
588.1.-4	Release 5.005 of Perl will support lightweight
	delete "will", replace "support" with "supports"

<wording, high>
588.1.-3	but it had not been released when this book went to print
	delete from "but it had" to "to print".  And change "5.005
	will support" to "5.005 supports".

<technical, high>
588.-4		# ...
	$SIG{INT} = \&catcher;
	swap the order of these two lines, so that $SIG{INT} appears
	before the #...

<technical, high>
591.4.5     This code
		while ($stiff = waitpid(-1, &WNOHANG) > 0) {
	    must be
		while (($stiff = waitpid(-1, &WNOHANG)) > 0) {

<technical, medium>
592,-2		use POSIX qw(:signal_h :errno_h);
	replace with
		use POSIX qw(:signal_h :errno_h :sys_wait_h);


<technical, high>
595.1    If an unexpected exception occurs, the alarm never
         got cleared!  Duplicate the alarm(0) line.
	 Make this:

	    } else {
		die;		# propagate unexpected exception
	    }

	Read this:

	    } else {
		alarm(0);       # clear the still-pending alarm
		die;		# propagate unexpected exception
	    }



602.3.-1    Add at the end of the paragraph, following "send and recv.",
	    this:
		    (Perl does not currently support sendmsg(2).)
	    And make sendmsg be italic.

<technical, medium>
610.3           setsockopt(SOCKET
                setsockopt(SOCKET
        in both cases, change SOCKET to SERVER

<orthography, high>
610.-1.3        # repeat calls to vec() for each socket to check
        put "vec()" in Courier not Italic Courier

<technical, high>
610.-1.-3       if (vec($rout, fileno(socket), 1)) {
        change "socket" to "SOCKET"

<technical, medium>
612.3           socket(Sockhandle, PF_INET,  ...
        change "SockHandle" to "SOCKET"

<technical, medium>
612.7           send(SockHandle, $MSG, 0, $portaddr) == length($MSG)
        change "SockHandle" to "SOCKET"

<technical, medium>
612.9           $portaddr = recv(Sockhandle, $MSG, $MAXLEN, 0) ...
        change "SockHandle" to "SOCKET"

<orthography, high>
620.5.2         ($kernel, $hostname, $release, $version, $hardware) = uname();
        put "uname()" in Courier not Italic Courier

<technical, high>
624.-1.2        use POSIX qw(: sys_wait_h);
        delete the space after the colon

<wording, high>
636.-3.1      "configurations file" should be "configuration file".

<technical, high>
638.-1.-2      The errant "Example 17-8" in the middle of the code must
	       be deleted.


<wording, high>
642.-1.1	Somehow these are in the wrong format.  This should
		reference Recipe 16.19 and Recipe 17.10, not by name.

<technical, medium>
645.1.-1        # @addresses is a list of IP addresses ("208.201.239.48",
"208.2
01.239.48")
        change the last "48" to "49"

<technical, high>
646.-1   #/usr

should be

	#!/usr

	or die "Can't find MX records for $host (".$res->errorstring,")\n";
        replace "," with "."



<wording, high>
681.3.1     Change
		Race even conditions crop up in apparently
	    to
		Race conditions crop up even in apparently



<technical, high>
688 code        while (@row = $sth->fetchrow()) {
        replace "fetchrow" with "fetchrow_array"

<technical, high>
692.-1.1        Recipe 14.10
        replace "14" with "19"

<orthography, low>
692.3.1        Put the word GET in italic.


<technical, high>
708.-4.-1       lines of content
        delete "lines" replace with "bytes"

<orthography, medium>
709             code
        align the comments beginning with # and   or die "usage: $0 url\n";
        the final } in the program should be at the start of a new line

<technical,high>
709.1		LWP::Heuristic
	Should read:
		URI::Heuristic

<orthography, medium>
711.2.2         script.cgi?field=...
        indent the line some more to make it obvious that it is a
        continuation.

<technical, high>
711.-1.1        Use the HTML::LinkExtor module from LWP
        delete "LWP" replace with "CPAN"

<orthography, high>
719.2           Here's an example of a program run:
        This should be Garamond, not courier.

<technical, medium>
721.-2          open(F, "< $filename")
        Replace with  open(F, "< $filename\0")

<orthography, high>
722.1.1         simple.template for internal template() function
        "template()" should be in Courier not Italic Courier

<technical, medium>
723.1           "dbi:mysql:connections:mysql.domain.com:3306"
        delete ":3306"

<technical, high>
723.1           if (@row = $sth->fetchrow()) {
        replace "fetchrow" with "fetchrow_array"

<technical, medium>
723.1           print header();
        put parentheses in Courier not Italic Courier

<wording, medium>
723.4           but they can also include full code expressions
        delete "code"

<technical, medium>
723.-1          "dbi:mysql:connections:mysql.domain.com:3306"
        delete ":3306"

<technical, high>
724.1           if (@row = $sth->fetchrow()) {
        replace "fetchrow" with "fetchrow_array"

727.1.4         /^(\S+) (\S+)  (\S+)  \[([^:]+):(\d+:\d+:\d+) ([^\]]+) "
        Should be:
                /^(\S+) (\S+)  (\S+)  \[([^:]+):(\d+:\d+:\d+) ([^\]]+)\] "
        (that is, add the \] before the space and doublequote.

<technical, medium>
727.1           (\S+) (\S+)$/;
        replace ";" with " or next";

<wording, medium>
727.4.2         If IDENT (RFC 1413) was used, what it returned
        add ", otherwise C<->" (courier dash)

<wording, medium>
727.5.2         , whom they logged in as
        add ", otherwise C<->" (courier dash)

<wording, medium>
727.6.2         Date of request (01/Mar/1997)
        insert "e.g.," after open parenthesis.

<wording, medium>
727.7.2         Time of request (12:55:36)
        insert "e.g.," after open parenthesis.

<wording, medium>
727.8.2         Iime zone (-0700)
        replace "Iime" with "Time"
        insert "e.g.," after open parenthesis.

<wording, medium>
727.9.2         Method of request (GET POST PUT)
        insert "e.g.," after open parenthesis.
        replace "GET POST PUT" with "GET, POST, or PUT"

<wording, medium>
727.10.2        URL in request (/~user/index.html)
        insert "e.g.," after open parenthesis.

<technical,high>
729.-3	Used to read:

    format STDOUT =
    @>>>>>>>>>> @>>>>>> @>>>>>>> @>>>>>>> @>>>>>> @>>>>>> @>>>>>>>>>>>>>
    $lastdate, scalar(keys %hosts), 
    $count, scalar(keys %what), 
		      $posts, $home, $bytesum
    .

Now reads:

    format STDOUT =
    @>>>>>>>>>> @>>>>>> @>>>>>>> @>>>>>>> @>>>>>> @>>>>>> @>>>>>>>>>>>>>
    {
    $lastdate,  scalar(keys %hosts), 
		       $count, scalar(keys %what), 
					  $posts, $home, $bytesum
    }
    .

<orthography, high>
731.5.2         ment text to read ``photo'.' It prints the new document
        replace "'.'" with "''."

<orthography, high>
733.3.1         This version of hrefsub will always lowercase the <a>
        put <a> in Courier

<technical, REALLY HIGH!>
733 add to the end of the page:

Example 20-12. hrefsub

#!/usr/bin/perl -w
# hrefsub - make substitutions in <A HREF="..."> fields of HTML files
# from Gisle Aas <gisle@aas.no>

sub usage { die "Usage: $0 <from> <to> <file>...\n" }

my $from = shift or usage;
my $to   = shift or usage;
usage unless @ARGV;

# The HTML::Filter subclass to do the substitution.

package MyFilter;
require HTML::Filter;
@ISA=qw(HTML::Filter);
use HTML::Entities qw(encode_entities);

sub start {
   my($self, $tag, $attr, $attrseq, $orig) = @_;
   if ($tag eq 'a' && exists $attr->{href}) {
           if ($attr->{href} =~ s/\Q$from/$to/g) {
               # must reconstruct the start tag based on $tag and $attr.
               # wish we instead were told the extent of the 'href' value
               # in $orig.
               my $tmp = "<$tag";
               for (@$attrseq) {
                   my $encoded = encode_entities($attr->{$_});
                   $tmp .= qq( $_="$encoded ");
               }
               $tmp .= ">";
               $self->output($tmp);
               return;
           }
   }
   $self->output($orig);
}

# Now use the class.

package main;
foreach (@ARGV) {
        MyFilter->new->parse_file($_);
}


735:   Add a new sub-entry under $_ of:
	    [ $_ variable, ] accidental clobbering, 99

735:   Change the name of the <> operator from "circumfix" to "diamond".
       Circumfix is its class, not its name.

735:   Insert in the numbers section :
	    "1 while", 16

735:   Insert in the numbers section :
	    "0 but true", 348, 608

736:   Insert:
	Alias module, 453

736:   Insert as a sub-entry under aliases:
	    [ aliases, ] for functions, 357

736:   Insert:
	    autovivification, 367-368

737:   Move the entry on circumfix operator and make it be diamond
       operator instead.

738:   Add to the entry on closures: 37, 357, 360, 424

739:   Add to the entry on daemon servers: 596

739:   Add to the entry on DBI module: 688, 723

740:   Add a new entry for "directory handles, 319" either
       as its own entry or as a sub-entry to the "directory" entry.

740:   changed the entry reading
	    /e pattern matching modifier, 169
       to
	    /e substitution modifier, 18, 163, 169

740:   Add to the /ee entry:
	    /ee substitution modifier, 169

741:   Add a new entry (perhaps doing something with the now two evals)
	    eval, in substitution, 18, 163, 169

742:   Add a new sub-entry under filehandles:
	    [ filehandles, ] local, 257, 354,

742:   Add a new sub-entry under filehandles:
	    [ filehandles, ] passing, 255, 257, 354, 411

742:   And a new entry for
	    fmt program, 28


743:   Add to the sub-entry on hashes, adding elements to: 351

744:   Add to the entry on importing from modules: 357

744:   Add to the entry on indirect filehandles: 256-258

745:   Add to the entry on ioctl: 427

745:   Add a new entry on
	    LINE: (implicit loop label) 239

747:   Add to the entry on non-blocking I/O: 608

747:   Change the
	    /o pattern-matching operator
	to
	    /o pattern-matching modifier

749:    Add a new sub-entry on pattern-matching
	    [pattern matching, ] progressive (with /g), 170

749:    Add a new sub-entry on POSIX
	    [POSIX, ] setsid function, 635

750:    Add a new sub-entry under processes
	    [processes, ] groups, 268, 584

751:    Change
	    [references, ] as hash elements, 148-149
	to
	    [references, ] as hash values, 148-149

751:    Add a new subentry under references:
	    [references, ] as hash keys, 486

751:   Add a new sub-entry under references:
	    symbolic references, 17-18, 423-424

752:	Add to the sub-entry on
	select, reading from multiple filehandles: 610-611

752:    Add a new entry:
	    shopping cart, 696

752:    Add a new subentry under signals:
	    process groups, 268, 584

753:   Insert under sockets:
	    sockets, non-blocking 608

754:   Add a new entry on
	    Sys:Syslog module, 637

755:   Add a new entry:
	    typeglobs, 40, 256-258, 262, 355-358

755:   Add to the entry on unbuffered input/output: 524-531

757:   Add to the entry on window size, determining: 427

757:   Add a new entry:
	    Year-2000 problem, 72