tr/// does not use regular expressions so your title has thrown everyone off.

tr/// only handles compile-time values so you have to use eval to get tr/// to deal with variable data.

If you are going to use the tr/// more than once, then for efficiency, you should do something like:

sub generateTr {
    my( $from, $to )= @_;
    my $sub= eval <<END;
        sub {
            my( \$copy )= \@_;
            \$copy =~ tr/\Q$from\E/\Q$to\E/;
            return \$copy;
        }
END
    die "$@" if $@;
    return $sub;
}

my $xlate= generateTr( "abc", "def" );
my $c= $xlate->( "abc leppard" );
# $c is now "def leppdrd"
[download]

Try:

$a = "abc";
$b = "def";
$c = "abc leppard";

$c =~ s/$a/$b/;
[download]

I was trying to come up with an interesting example but failed. I actually _do_ want to do translation. (I'm writing a little crytpogram helper tool.) I want to replace all the letters from an alphabet, say $a, with all the letters from the new alphabet, say $b.

# the alphabet
$a = "abcdefghijklmnopqurtuvwxyz";
# i've discovered that a represents T in the cryptogram, for example
$b = "T     E                   ";

$c = "the text i want to decipher";

$c =~ tr/$a/$b/;
[download]

So basically, is there any way to get this to work? Thanks for your help.
-Paul

Ok, here it is:

Right from the perlop manpage:

Because the transliteration table is built at
compile time, neither the SEARCHLIST nor the
REPLACEMENTLIST are subjected to double quote
interpolation.  That means that if you want to use
variables, you must use an eval():
 
eval "tr/$oldlist/$newlist/";
die $@ if $@;

eval "tr/$oldlist/$newlist/, 1" or die $@;
[download]

Do a perldoc perlop to read all about tr

#!/usr/bin/perl
$a = "abc";
$b = "def";
$c = "abc leppard";

$c =~ s/$a/$b/;
print $c;
[download]

bbc leppbrd
[download]

try :

$c =~ tr/$a/$b/;
[download]

$c =~ s!$a!$b!;
[download]

Cheers,
KM

$c =~ s/$a/$b/;

Update oops, read it wrong... and other people have answered it.
                - Ant