Symmetric encryption

[Per Tunedal Casual]

At 07:58 2004-10-26, you wrote:
 >On Mon, 25 Oct 2004, Per Tunedal Casual wrote:
 >
 >> I thought once more about this :-) It's hard enough to put up a
 >> passphrase that matches an 128 bit-key. The hash, if any, doesn't
 >> matter, because it can hardly decrease the strength of the passphrase,.
 >>
 >> Thus any cipher with a key longer than 128-bits is overkill when you
 >> encrypt symmetrically. Right?
 >===================
 >
 >i have several passphrases that are 20+ characters. 22 characters of
 >uppercase, lowercase, numbers and spaces is *stronger* than 128 bit.
 >
 >that's not counting the full range of characters that can be used...
 >assuming that 95 characters are generally suitable (read: safe) for use in
 >passphrases, it only takes a theoretical 19.5 characters to equal 128 bits
 >and 39 characters to hit 256 bits. such passphrases may not be suitable
 >for all applications, but they're not entirely ridiculous. IMHO it's
 >posible to come up with passwords that are that long, easy to remember and
 >strong.
 >
 >AFAIK, the session key used for symmetric encryption is derived from the
 >user supplied passphrase, then salted and iterated (this may have changed
 >recently). if symmetric and asymmetric encryption is used then the session
 >key is derived in the usual random way.
 >
 >
 >- --
 >         ...atom
 >
In that case it's important to know how the session key for symmetric 
encryption is derived from the passphrase. If hashing with SHA-1 i used 
that will set the limit to 160 bits.

In that case AES192 would be appropriate - only 32 bits wasted :-). AES256 
or TWOFISH would be ridiculous, wouldn't it?

Per Tunedal