ietf-openpgp
[Top] [All Lists]

Re: Changing HAVAL definition.

1998-06-16 11:58:56
On Tue, 16 Jun 1998, Jon Callas wrote:

At 05:10 PM 6/15/98 -0700, Hal Finney wrote:
   In looking at the source to PGP 5.5, I see that it had defined hash
   algorithm 4 as an experimental double-width version of SHA.  It never
   created messages with that hash, but if it sees hash algorithm 4, it will
   think that is double-SHA rather than HAVAL as the spec has it.  Can we
   reserve hash algorithm 4 and redefine HAVAL to be 7?  That way 5.5 will
   know that it is seeing an unrecognized hash if it ever sees hash 7,
   and produce an error message.  Otherwise it is likely to report a bad
   signature if it sees hash 4 because someone used HAVAL.
   
I have no objection to this. Anyone else?

To all: by the bye, we still don't have OIDs for HAVAL and Tiger. I am
content to leave this situation as it is. But I would be delighted to add
OIDs if someone were to scrounge them. I am even willing to drop them
completely if there's consensus in that direction. I'm just bringing up the
issue. No response means you agree it's okay to leave it as it is.

Please restate this since you don't want to "leave it as it is".

What I think is being proposed:

Hash#   Old     New
4       HAVAL   (reserved)
7       (undef) HAVAL

However, the next revision of PGP after 5.5 should treat hash #4 as
reserved (and use 100+ for experimenting).  Since I expect one of the hot
items for the next version of this spec will be a wide hash (to go with a
wide DSA variant), I will expect such a hash to be dropped in as #4. 

Someone should also create a patch for the source of 5.5 that disables
hash #4.  So when is there going to be a PGP source code journal?

As far as the OIDs (or lack thereof), I made a suggestion to alternately
use an octet string, or IA5 string with the string from the Hash:
clearsign header in place of the OID.  This way the issue goes away.  I
plan on proposing this for the next revision.  There won't be an OID for
the wide hash whatever it is.

--- reply to tzeruch - at - ceddec - dot - com ---


<Prev in Thread] Current Thread [Next in Thread>