Jeff Bailey <[EMAIL PROTECTED]> writes:
> I guess I expect it to give more information about "unknown publickey
> algorithm". Does that mean it doesn't recognise the publickey I sent it,
> or does it mean that it doesn't recognise the encryption with which it
> was coded?
The former.
> Maybe some sort of detail: "You sent the following fingerprint. XX
> XX XX XX. You have registered the following fingerprints: XX XX XX
> XX, XX XX XX XX. No match" Right now I don't realy understand the
> error message.
It might be possible to include the algorithm name in the message.
Although not trivial to do right, as the names are generally forgotten
as soon as they have been interned to ATOM_FOO (if foo is a known
magic word) or zero (if the symbol was not present in the atoms.in).
In the latter case, the information is lost.
Fingerprints are more difficult; if the client says "this is a
gnomovision public key", it is difficult for the server to figure out
what kind of fingerprinting makes sense for the gnomovision system.
(For a concrete example, substitute openpgp for gnomovision).
> Yes. It created a directory called ~/.lsh/authorized_keys_sha1 and
> placed the 2 public keys there that I put in in some binary format. (I'm
> testing with both lsh and ssh2 and being equally unloved)
It's not enough to copy the files there; they must also have the
right names (in fact, it is *only* the names of the files in that
directory that matter, the server doesn't even read their contents).
The documented way to add keys to that directory is to use the
lsh-authorize script.
> Thanks I'll try this, I just need to get Scheme and the new M4 loaded on
> to subversions.
Newer m4 is not crucial; it's needed only for regenerating some files
in the testsuite.
For scheme, IIRC the needed debian packages are guile and slib.
/Niels
