[NNTP] Proposed changes to STARTTLS-07

[Russ Allbery]

Ken Murchison <ken at oceana.com> writes:

> Based on comments from the list and Russ' suggestion, here is my current
> diff for STARTTLS.  Note that a lot of this is just moving text around
> (back to -06 placement).  It feels like we need more text describing
> *why* the connection should be closed after a failed TLS (session in
> indeterminate state, interop problems, etc), but I couldn't come up with
> anything that I like.  Suggested text is welcome.

How about:

    While recovery may be theoretically possible, in practice TLS
    negotiation failure often leaves the session in an indeterminate
    state.

The other changes look good to me.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>