Errata

The arrows are missing, as in the wrong example.

The rewritten version of pad_str should use $opts_ref
with the dereferencing arrow ->. This is important because that's the whole point of this topic!

The perltidy option listed as -pt=1 should instead be -pt=2.

Examples of code throughout the book are mostly formatted with -pt=2
so this should be the one used in perltidy, yes?

-pt=1

should be:

-pt=2

because throughout the book if, while, for parenthesized expressions
have tightly bound parentheses.

if ($x == $y) {

instead of:

if ( $x == $y ) {

The way you format the multi-line return statement is inconsistent with the way a very similar return is formatted on page 179: on page 179 the string concatenation dots are lined up under the "n" of "return" so that each of the concatenated strings starts in the same column. However, on page 182 the dots are placed under the "$" of $arg_ref instead.

You break your own rules slightly with the line:

open my $pipe, '/cdrom/install |'

since this is neither three-argument open nor IO::File open (page 207)

It should be:

open my $pipe, '-|', '/cdrom/install'

(yes it's a constant string so it's not unsafe, but rules are rules. Allow that and you're half way to allowing q{open my $pipe, "$foo |"} so long as you "know" that $foo is safe. Ok, maybe only a quarter of the way, but it's the thin end of a slippery slope.)

The example for Loop Labels does not seem to fulfill its intention, to 'give up trying to contact clients if any of their numbers is marked "Do Not Call"'.

The code as written gives up for a client only if the FIRST phone number in the client's list has a "Do Not Call". Otherwise, it uses that number as the preferred number to send an ad even if a subsequent phone number has a "Do Not Call". The "last PHONE_NUMBER;" statement ensures that subsequent phone numbers are not checked.

As noted in the comments of the first (non-boldface) code example on the page, the
_ref variables should use dereferencing arrows to access values. Thus, the two
lines:

my $gap = $opts_ref{cols} - length $text;
my $left = $opts_ref{centred} ? int($gap/2) : 0;

should read instead:

my $gap = $opts_ref->{cols} - length $text;
my $left = $opts_ref->{centred} ? int($gap/2) : 0;

Shouldn't $tax_form be $tax_form_ref, especially this close to the Reference
Variables recommendation?

Function name is incorrect - it should be odd() (the function from the previous code snippet),
but the text refers to even() which DNE:

"... then a better solution is usually to "inline" the call to even(), replacing ..."

-Should Be-

"... then a better solution is usually to "inline" the call to odd(), replacing ..."

In the discussion of prototypes, the example code includes a subroutine called "clip_to_range"
that takes $min, $max, and @data as parameters. Based on that information, I expected the code
to return only those data elements that are between $min and $max.

Inspection of the code, however, revealed that through the use of List::Util's min and max
routines data elements that were less than $min were changed to $min and those greater than $max
were changed to $max. The code in question is map { max( $min, min( $max, $_ ) ) } @data.

For example, given $min = 1, $max = 3, and @data = ( 0 .. 4 ) I expected the subroutine to
return ( 1, 2, 3 ). It actually returns ( 1, 1, 2, 3, 3 ).

If this is the intended behavior of the subroutine, I suggest renaming it (as the use of "clip"
implies cutting and removing) and/or adding a comment that indicates that is the intended
behavior. If the code itself is in error, one simple fix might be something like grep { ($min <=
$_) && ($_ <= $max) } @data.

Replace
-onechar
with
-one_char

The last word, 'it', should have a footnote explaining that the unlinking process proposed does
not work under MS Windows, and that IO::InSitu should be used instead.

One of the most important paradigm shifts in this book is getting users to use
Class::Std module to build new perl modules (which ROCKS by the way). However,
Perl6::Export::Attrs is not compatible with Class::Std. This shoudl be pointed out in
the book. Or the modules need to be fixed.

(http://safari.oreilly.com/0596001738/9780596001735-CHP-19-SECT-14);
The code refactoring is mildly flawed because the original

$target_name = $1 and last

is not behaving the same way as the refactored

$target_name = $name;
last CANDIDATE;

for value "0", i.e. the candidate string is "Name: 0; "