Errata

On page 15, I encountered the following:

"(By the way, you should be careful to spell it usr and not user - the folks who invented Unix were lazy typists, so they omitted a lot of letters.)"

"usr" stands for "Unix System Resource."

the sub which_element_is is written to return -1 for non-matches, which,
if used when subscripting an array will return the last element of the
array, probably not what the caller intended. So not only would returning
undef be more "Perlish" it would be less dangerous.

The (excellent) book, on page 79, talks about what happens when you reverse
a hash. Cool. Since I understand what a hash table "really" is, I understand
that keys are unique, values are not guaranteed to be unique. You explain
this concept well.

However, you use the example of the IP address/ hostname hash as one that can be
reversed because it only contains unique values. This is not true, if you think
about it for a second. The example is:

%ip_address = reverse %host_name;

In fact, many hostnames may point to one IP address. Happens a lot, especially
with the craze of virtual hosting. I don't know of an example that actually
works for this concept, but this one does not.

Does read:

... the output will say "|before<match>after|", ...

Should read:

... the output will say "Matched: |before<match>after|", ...

In the program code an 'elsif' is used where an 'else' would suffice.

To remove empty directories, use the rmdir function in a manner
similar to the unlink function:

rmdir glob "fred/*"; # remove all empty directories below fred/

...

As with unlink, rmdir returns the number of directories removed, and if
invoked with a single name, sets $! in a reasonable manner on a failure.

According to the latest PERL to which I have access (5.6.0), this example
does not work. "rmdir" can only remove one directory.

Furthermore, the third edition of _Programming Perl_ states (on p. 777) the
following:

rmdir FILENAME

This function deletes the directory specified by FILENAME if the
directory is empty. If the function succeeds, it returns true;
otherwise, it returns false.

So I believe that both the sample code AND the description of what "rmdir"
returns is erroneous on p. 183 of _Learning Perl_.

$personal_name{$a} cmp $family_name{$b}
should probably read
$personal_name{$a} cmp $personal_name{$b}
^^^^^^^^
because you've alread sorted by family name in the previous line

The text reads:

..., giving a format string of "a40 C I5 L". Each record is
thus 55 bytes long. ...

On my system, and I assume most systems that readers of the book have access
to, this format comes out to 65 bytes long as 'I' is a four byte integer,
not a two byte integer as described in the text. I understand that Perl
sometimes relies on underlying 'C' types and that 'C' ints can be two bytes
on some systems. (Unfortunately, I can't find any Perl documentation that
states how these pack formats relate to the underlying architecture.) But
this should either be made clear in the text or an example that is likely to
work on the majority of reader's machines should be used. (e.g. use 'S'
format since we're trying to contain values up to 300.)

This pattern is surprisingly tricky to get right.
The pattern shown, /(w+)(s+1)+/, matches
|I though<t t>hat that that was the problem|.
It is supposed to match
|I thought <that that that> was the problem|.
The pattern requires a leading word anchor, thusly: /(w+)(s+1)+/.

The second exercice asks for an identifier name (made of letters, digits, and u
nderscores...), so we can suppose that the name as at least 2 characters.

The proposed answer for those names is [a-z_]w*, but that match names with on
ly one character. I understand that a better solution would be [a-z_]w+

Index: missing "$$" entry, it's covered on 183.