Re: DISCUSS comments on publickeyfile-09

[Joseph Galbraith <galb-list%vandyke.com@localhost>]

>    The introduction to this document is very lightweight.  Please expand
>    it to provide some context.  At a minimum, this needs to say that the
>    document is about the SSH protocol and the role of public keys.  It
>    is also desirable to cover the trust model, explaining why these
>    files are very important.
>
>    The introduction should be expanded to discuss fingerprints and how
>    they are used in SSH.

Is this better:

  The SSH protocol supports the use of public/private key pairs
  in order to perform authentication (public-key authentication.)
  However, in order to use public-key authentication in the SSH
  protocol, public keys must first be exchanged between client
  and server.

  This document formally describes an existing public-key file
  format which can be used with any of the common existing file
  transfer mechanisms in order to exchange public keys.

  The SSH protocol also uses public/private key pairs to
  authenticate the server.  In this scenario, it is important
  to verify that the public key provided by the server is
  indeed the server's public-key.

  This document describes a mechanism for creating a short text
  string that uniquilly represents a public-key (fingerprinting)
  for use in manually comparing public keys.

>    In section 3.6, please change "me@myhost" to "me%example.com@localhost".

Done.

>    The last paragraph of the security considerations needs to be
>    expanded to provide a bit of context.  MD5 has some known weakness,
>    but they are not a problem in this situation because ...

Is this better?

  The public-key fingerprint method presented here relies on
  the MD5 hash, which is known to have certain weaknesses
  regarding it's collision-resistance; however, the particular
  use made of MD5 here depends solely on it's 2nd-preimage
  resistance, not on it's collision-resistance.

  MD5 is used here for historical reasons.

Thanks,

Joseph