On Wed, Sep 09, 2009 at 03:08:24PM -0700, Bart Smaalders wrote:
> [...]
> signature type=x.509 hashtype=sha-256 sigtype=rsa-pkcs public_key=<hash> 
> [public_key=<hash>] ... value=<hash> ...
> [...]
> For testing and process verification purposes, a signature of type
> "identity" is also supported; this presents the unsigned hash as
> the value directly:
> 
> signature type=identity hashtype=sha256 value=<hash>

The 'type' attribute seems confused.

In the first case (type=x.509) what's really being indicated is a
certificate type (could be x.509, could be a raw public key ala SSH),
while in the second case the type is really like sigtype.

I recommend:

signature hashalg=sha-256 sigalg=rsa-pkcs certtype=x.509 public_key=<hash> ...
signature hashalg=sha-256 sigalg=rsa-pkcs certtype=none public_key=<hash> ...
signature hashalg=sha-256 sigalg=none value=<hash> ...

The last one corresponds to type=identity.  The second one corresponds
to raw public keys, without using certificates.

Also, why can there be more than one public_key=<hash> per signature
action?

Finally: what should be named by public_key= in the case of x.509
signatures: a hash of the public key, or the certificate?  And if the
certificate, by issuer-and-serial, or hash of the certificate?

> Policy constraints regarding required signatures on pkgs will specify
> a public key that must appear in the certificate path of a valid signature
> on a manifest to enable installation; such policy constraints may be
> applied to all repos or just one.

I'm not sure what this means.  I think you might have meant that the key
used to make a signature action should be from a certificate that is
valid, including a valid path to a trust anchor.

Otherwise it looks good to me.

Nico
-- 
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to