[Sks-devel] Re: Key strangeness

[David Shaw]

On Sun, Feb 08, 2004 at 02:10:01PM -0500, Yaron Minsky wrote:
> Ah, that explains why I couldn't find it.  Does this problem come up with
> PKS, or is it SKS specific?  Is one possibility here that the lead packet
> is malformed, and as a result some difference in implementation of rfc2440
> results in differently evaluated keyids?  i.e., maybe SKS things that this
> key really does have keyid 3A546EC2, and GPG thinks it has a different
> one, basically because the key packet is malformed.

My guess is the same, but there are some other factors that are very
strange:

1) The problem is not SKS specific, and pgp.mit.edu (a pks
   installation) shows the same thing.

2) Downloading the "bad" key, and then re-submitting it to the
   keyserver results in the correct record - i.e. the keyserver is
   able to parse the key properly the second time around.

I'm vaguely wondering if the key was slightly corrupt - say, a badly
formatted MPI - and when the keyserver received it, it
re-canonicalized the MPI to make it correct, but still indexed it
under the bad key ID.  Could SKS do such a thing?

David