in reply to Fastest Encrypt/Decrypt

I think the RC4 algorithm is generally considered a good compromise between speed and security. There are several CPAN modules implementing it, e.g. Crypt::RC4::XS.

Update: More generally, if speed is of paramount importance (and you feel like playing with this), find a pseudo random number generator that's extremely fast but still produces reasonable quality randomness, i.e. 'random' enough to match your idea of what isn't "extremely simple to decrypt".  Then, to encrypt, bitwise XOR the data with the pseudo random number sequence. To decrypt, just do the same XORing once again with the same random sequence, i.e. using the same PRNG seed, which could be derived from a passphrase, together with some 'salt' (the latter is some additional random data generated anew for every encrypt/decrypt cycle, and stored with the message).

Replies are listed 'Best First'.
Re^2: Fastest Encrypt/Decrypt
by zwon (Abbot) on Jan 16, 2009 at 14:25 UTC

    Note, that when using RC4 you should never use one key twice, so maybe this is not a best choice.

    Update: just a little demonstration of the problem:

    use strict;
use warnings;
use Crypt::RC4::XS;

my $line1 = pack 'a32', '';
my $line2 = pack 'a32', 'some secret data';
my $enc1 = RC4('password', $line1);
my $enc2 = RC4('password', $line2);
my $xor = $enc1 ^ $enc2;
print "Result: $xor\n";
__END__
Result: some secret data

    [download]
[reply]
[d/l]
      you should never use one key twice, so maybe this is not a best choice.

      Yes, this is a (rather well-known) property of the algorithm, which immediately follows from the underlying XORing used.

      However, this doesn't necessarily mean it would be a bad choice, as the 'problem' can easily be worked around using some salting mechanism, which would provide one-time keys, while still allowing the same passphrase to be used multiple times.

      For example, just compute the key from the passphrase plus some nonce using a one-way hash function such as MD5. The nonce is stored/transmitted together with the data.  (The RC4 algorithm has a few other more subtle problems, but for the moderate security needs as requested here, they're pretty much irrelevant.)

[reply]

        Yes problem is solvable, I just drew OP attention to it. Solution is simple, but it requires some accuracy and understanding of the problem.

[reply]

In Section Seekers of Perl Wisdom