RE: Digest mess
Paul Leach (paulle@microsoft.com)
Wed, 7 Jan 1998 09:30:03 -0800
I agree with Scott that digest without message integrity is not very secure
-- just the assurance that some part of the message was once issued by the
client.
> ----------
> From: Scott Lawrence[SMTP:lawrence@agranat.com]
> Sent: Wednesday, January 07, 1998 7:18 AM
> To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> Subject: Re: Digest mess
>
>
> >>>>> "DK" == Dave Kristol <dmk@bell-labs.com> writes:
>
> DK> Repeat after me "Digest is meant to replace Basic."
>
> I would just extend that to
>
> "Digest is meant to replace Basic with something useful."
>
> DK> Basic provides no message integrity. There may indeed be a need
> DK> for a lightweight way to detect POST/PUT message integrity. But
> DK> let's not necessarily force the latter on the former.
>
> Even GET without at least some response integrity is of questionable
> value, and without a digest you don't even get that (and you can if
> you include a client nonce). I don't accept the argument that we
> are doing something good by replacing one inadequate mechanism with
> another inadequate one.
>
> DK> Are you saying that the original Digest wouldn't play through proxies?
> DK> If so, how would the more recent version do any better? (Unless
> you're
> DK> referring to the way proxies may mung headers. Original Digest wasn't
> DK> dependent on other headers.)
>
> It was dependant on the headers for the entity integrity, and that's
> essential (in my admittedly not-so-humble opinion :).
>
> DK> I would hate to see the best be the enemy of the good here.
>
> As I see it, I'm far from arguing for the 'best' here... just
> 'adequate'.
>
> DK> I have had server support for Digest for nearly three years (and
> DK> SimpleMD5 before that), and I've longed for widespread support of
> DK> Digest in clients.
>
> ...and one argument that client vendors have made for why they have
> not deployed it is that it is too weak - why would they be any
> more motivated to provide Digest without integrity protection?
>
> They have argued that Basic with SSL is better, and as things appear
> to be going I would have a tough time making a technical case
> otherwise.
>
> --
> Scott Lawrence EmWeb Embedded Server
> <lawrence@agranat.com>
> Agranat Systems, Inc. Engineering
> http://www.agranat.com/
>