Errata

My question is about the sentence below:

The following characters: - _ . ! ~ * ' ( )

In the book, tilde "~" is said to be an exception for URL encoding, but
according to RFC1738 (Uniform Resource Locators) section 2.2, tilde should be
encoded.

my $fh = $q->upload( $file );

On Win32 systems (I'm not sure about Unix) it appears that the "upload" method
should be passed the name of the "upload field" from the HTML form (which in
this case is "file") rather than the path and name of the file itself ($file).

From the CGI documentation:

--------------------------------------------------------------------
When the form is processed, you can retrieve the entered filename by
calling param():

$filename = $query->param('uploaded_file');

... (some stuff snipped out) ...

To be safe, use the *upload()* function (new in version 2.47). When
called with the name of an upload field, *upload()* returns a
filehandle, or undef if the parameter is not a valid filehandle.

$fh = $query->upload('uploaded_file');
while (<$fh>) {
print;
}

On both 98 and 99,

my $fh = $q->upload( $file );

should be

my $fh = $q->upload( "file" );

In the first paragraph on 99:

"Be sure that you pass upload the name of the
file according to param, and not a different name..."

would be more clear as:

"Be sure the name that you pass upload is the same name
you would pass to param, and not a different name..."

As it is, it makes it sound like you are passing the name returned from
param, which is wrong (but it matches the code samples, which makes it even
more confusing.

Comfirmation that on Unix, Mac OS X and Mac OS that the argument to upload should
be the name of the param not the name of the file.

The regular expression to open a unique output file only works if the
user-supplied filename contains an extension separated by a period. If
this is not the case, the regex fails, a large amount of garbage is
printed to the error log, and the connection eventually times out.

The solution is to fix the regex on line 31 of the program so that it
validates the filename correctly, and adds an extension if one is
necessary. For example:

if ( $filename =~ /^(w[w-]*)(.[w.-]+)?/ ) {
$filename = $1 . ($2 || '.err');
}

Text description:

From the start of the line, ensure that there is a string that starts
with a word character and is continued by word characters or hyphens.

This may, but need not be, followed by a period, followed by one or more
characters that can be either word characters, periods, or hyphens (to
accommodate things like foo.tar.gz).

The filename is the part before the extension concatenated either with the
part after the extension (if present) or the string '.err' (if not
present).
<<

The user errata at the book web site indicate that others have had
problems with this, but the proposed solution is to delete the code and
replace it with a much simpler file check that does not have nearly as
much error handling. The above solution preserves this error handling and
should be universally applicable (it even works if the filename ends in a
period).

This would seem to preclude the use of frames with CGI.pm start_html because
FRAMESET is allergic to BODY tags. It must be possible since at the bottom of
p 104 the -target argument is discussed.

Is it possible for CGI.pm to redraw just one frame? My current cgi redraws
all the frames even though only one changes.

-name => "curtain"
should be:
-name => "look_behind"
because lines 21 - 23 have: NAME="look_behind"

In your example of EMBPerl, there appears to be an issue with the edit
function. Although it will work fine if run form the embpcgi.pl, editing a
story will not work at all while running in mod_perl, as it is currently
written. When the story is read, a shared lock is placed on the file, as
mod_perl does not terminate, like the cgi script, this lock is maintained and
will not allow you to write to the file. This results in no editability unless
the cgi is used...Obviously, a poor choice if mod_perl is an option. The
approach I would take is to add a cleanup sub to unlock the file for editing,
run every time the page is compleatly loaded...or at least from talking to
people this would seam to be the solution (mod_perl EMBPerl newsgroup).

Oh, just in case I am wrong...The reason that the lock came to mind was I
reset apache (and subsiquently mod_perl) and then saved after going to the
edit_article.epl?, and then it worked.

$q->textarea(........... -value => $comment ),

should be:

$q->textarea(........... -default => $comment ),

because textarea has no value property (per pg. 79).

The code appears exactly as printed in the next 2 lines:
open MAIL "|-" or exec( "/bin/mail", $email ) or
die "Cannot exec mail $!";

First, there should be a common between MAIL and "|-"

More importantly, although the text states that this code
uses the "fork and exec" trick, there is no explicit call
to fork (although the OPEN does implicitly call fork)

The exec will only take place if the open fails!

The query's don't actually return anything. I am unable to figure this out as I am
using the book to teach myself CGI programming. If you do a query without filling in
the fields of the query screen, it returns every entry in the database. If you put
anything in the fields, including exact matches, it returns nothing. I typed the
entire script from the book and when this problem occurred I downloaded the script
from your website (Under Chapter 10 9781565924192_examples.tar.gz) and it behaves the same
way. I compared the downloaded version to your book and it is the same.

The use of map(... " = ""
... """
on pg 259, and twice on pg 260 do not work for SQL code.

You should be using single quotes around where logic literals, as it was done on pg
258 in the doAdd subroutine, using

"push(@value_array, "'" . $value . "'");

The sort is coded to sort the number of matches from lowest to highest. Search
engines usually show highest number of matches first. To allow this, the sort
should be coded as follows (swapping $a->[0] and $b->[0]):

sort { $matches{$b->[0]} <=> $matches{$a->[0]} ||
$a->[1] <=> $b->[1] }