This and other RFCs are available on the web at http://dev.perl.org/rfc/ =head1 TITLE crypt() default salt =head1 VERSION Maintainer: Mark Dominus <[EMAIL PROTECTED]> Date: 11 Sep 2000 Last Modified: 21 Sep 2000 Mailing List: [EMAIL PROTECTED] Number: 208 Version: 3 Status: Developing =head1 ABSTRACT A frequently-asked question is how to generate an appropaite random salt for password encryption. I propose that Perl generate the salt automatically if the salt argument is omitted in the call to crypt(). =head1 DESCRIPTION At present, crypt() requires two arguments: crypt PLAINTEXT,SALT It then passes these arguments directly to the C library crypt() function. When encrypting a new password, the programmer is required to generate a salt at random: @letters = ('A' .. 'Z', 'a' .. 'z', '0' .. '9', '/', '.'); $salt = $letters[rand@letters] . $letters[rand@letters]; $passwd = crypt($passwd, $salt); This is inconvenient and nonportable. It's also nonobvious: people frequently ask in the newsgroups how to do it. I propose that if the SALT argument is omitted, Perl should generate an appropriate salt internally and use that. $passwd = crypt($passwd); # Same as above On systems where the password format is different, Perl can do the appropriate thing. =head1 IMPLEMENTATION For the standard DES-based crypt, the implementation is straightforward trivial. Perl already has many functions that take an optional argument, and the C internals of the random-salt generator are well-known. Details will vary for systems using alternative password hashing schemes. On some systems, no salt need be generated. These can be taken care of with a suitably ifdef'ed section of code if necessary. If the random number generator has not yet been seeded, Perl should seed it. Michael Schwern has developed a partial demonstration implementation in pure Perl. It is available from http://www.pobox.com/~schwern/src/RFC-Prototype-0.01.tar.gz It has been suggested that C<crypt()> should have a private random number generator, to avoid interfering with the sequence of numbers produced by rand(). This would significantly complicate the implementation, and I believe it is probably unnecessary. See the REFERENCES for details. =head1 MIGRATION C<crypt()> with only one argument is presently a compile-time error, so there are probably few translation issues. The meaning of this program will change: $" = ', '; $code = "crypt(@ARGV)"; eval $code; die $@ if $@; But I don't think this is anything to worry about---it should fall into the "other 5%" category. =head1 REFERENCES perlfunc manpage for discussion of crypt() crypt(3) http://dev.perl.org/archive?35:mss:4500:200009:lmemkmdbnocclmnnijmc
