[TLS] comment on draft-ietf-tls-session-hash-01

[Nikos Mavrogiannopoulos <[e-mail]>]

Hello,
 The text in draft-ietf-tls-session-hash-01 specifies:
   "Implementation note: As described in Section 4, the "session_hash"
is
   used in the extended master secret computation.  Hence, it must be
   possible to compute the session_hash before the master secret is
   computed.  In SSL 3.0, the master secret is first needed in the
   Client's CertificateVerify message.  Hence, it is widespread
   implementation practice to compute the master secret as soon as the
   "pre_master_secret" is available, typically immediately before or
   after sending the Client Key Exchange message."

My first comment is that this text is pretty confusing.

The second is to clarify my understanding. Does this text suggest that
for SSL 3.0 in order to calculate the "extended master secret", must
first calculate the "master secret" it is supposed to replace?

If yes, I find the argument unconvincing for the introduction of such
complexity in the master secret calculation.

I suggest the following alternatives:
1. Do SSL 3.0 extended master secret calculation as in TLS 1.0 when this
extension is present
2. Explicitly state that SSL 3.0 is not in the scope of the document;
anyway there is no real expectation of such an extension to be
implemented in an SSL 3.0-only implementation.

regards,
Nikos