Bill Studenmund <skippy@macro.stanford.edu> writes:
> Stanford uses the andrew string-to-key routine, which for 8 character
> passwords, encrypts the password using "#~" as the salt. Yes, "#~". It
> uses an illegal salt. But it does it quite blatently, and it seems that
> there is some sort of convention as to what a salt of "#~" means.
> 
> Any thoughts as to what it means and how to teach our libcrypt to deal
> with it?

Try using the salt "p1".  I believe that's legal; at least the modern
crypts I've used (include NetBSD's) don't choke on it the way that they do
on "#~".  And it works, too, with AFS Kerberos;   I can only assume that
the older implementations of crypt did some sort of table lookup that
wrapped around.

kcr