FWIW, my opinion is this is likely another case of computer engineers
trying to optimize something that doesn't need to be optimized (a sin I
have been guilty of many times).
And I'm saying that as someone who went to the trouble of writing a spec
and implementing a prototype SASL mechanism for this purpose (plaintext
password encrypted only during the authentication phase). I now think that
work was largely a waste of time (although I had fun doing it and learned a
lot).
The cost of symmetric ciphers is small to negligable on modern hardware,
particularly a wimpy cipher like RC4 which is the most common in SSL/TLS.
Rather than making TLS implementations more complicated (and less secure)
to support mid-stream down-negotiation, or introducing another SASL
mechanism to do this, why not just optimize the RC4 code? That will
benefit _all_ protocols using TLS and reduce the complexity of the Internet
suite of protocols.
Encrypting data that doesn't need to be encrypted is good for overall
security of the system.
- Chris
- IMAP and Netnews Charles Lindsey
- Re: IMAP and Netnews Mark Crispin
- Re: IMAP and Netnews Timo Sirainen
- Re: IMAP and Netnews Ken Murchison
- Re: IMAP and Netnews Chris Newman
- Re: IMAP and Netnews Ken Murchison
- Re: IMAP and Netnews Mark Crispin
- Re: IMAP and Netnews Timo Sirainen
- Re: IMAP and Netnews Mark Crispin
- Re: IMAP and Netnews Charles Lindsey
- Re: IMAP and Netnews Mark Crispin
- Re: IMAP and Netnews Ken Murchison
- Re: IMAP and Netnews Mark Crispin
- Re: IMAP and Netnews Charles Lindsey
- Re: IMAP and Netnews Mark Crispin