Eric Rescorla wrote:

> That said, TLS and SSLv3 are nearly identical. The differences
> essentially come down to:
> 
> (1) A tightening up of the Key Derivation Function (tying it more
> closely to HMAC)
> (2) A replacement of the ad hoc (and somewhat broken) MAC used in
> SSLv3 with HMAC.

Yes, HMAC is of course much better -- it's one of the few
crypto components we use that has provable strength.

> (3) A truncated MAC for the handshake verify function.
> (4) Some additional alerts.
> (5) Some clarifications.
> (6) A requirement to implement DH/DSS. This is going to be changed
> in the next draft however.

The last was initially motivated by the IETF requirement to
avoid technologies encumbered by patent.  Patent expired.  I 
myself am an advocate of using DH wherever third-party verification
is not required, however.

> I've certainly heard plenty of arguments that the changes made to TLS
> were unnecessary (in fact I've made such arguments myself) but I've
> never heard any even remotely convincing arguments that they render
> the protocol less secure. In fact, there are some plausible arguments
> that they render the protocol more secure. The only real debate
> was whether the rather modest improvements in security were worth
> the price of incompatibility.
> 
> Michael, do you have some argument to make the TLS is inferior
> to SSL?

No, I don't.  I think I was trying to point out that the effort/quality
of product ratio was higher w/TLS.   IETF task forces seem enormously
bureaucratic to me these days, and I was using this as an opportunity
to complain.  

> [0] There were, however, some relatively well known names there: Hugo
> Krawczyk, Ran Canetti and Dan Simon come to mind. Schneier even
> chaired an early rump group meeting.

Krawczyk, Canetti and Bellare are responsible for HMAC -- a thing
of beauty.  Don't get me started on Bruce -- the showman of crypto.
He could take Carl Sagan's place. ;-)

Mea culpa, mea culpa, mea culpa.
______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
User Support Mailing List                    [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to