Re: PKAlgs: 02: DSA Signature Keys needs re-wording

[Bodo Moeller <[e-mail]>]

On Fri, Mar 23, 2001 at 09:53:19AM +0100, [e-mail] wrote:

> Some further comments:
> 
> section 2.2.2 states:
> When the id-dsa-with-sha1 algorithm identifier appears as the algo-
>    rithm field in an AlgorithmIdentifier, the encoding SHALL omit the
>    parameters field.
> 
> This seems to contradict the intended contents of section 2.3.2 ?

No.  Section 2.3 is on algorithm identifiers for *keys* contained in
certificates; section 2.2 is on algorithm identifiers for *signatures*.
To verify a signature, you must already have the key, and that is
where you obtain the parameters from.

A couple of days ago I misunderstood the draft exactly as you do now
(except that I had looked at ECDSA where you now look at DSA :-).
So maybe it's not just me and the draft should have some additional
clarification in the text.  The confusion is caused by the use of
'AlgorithmIdentifiers' for two different purposes: in the
'signatureAlgorithm' field and in the 'subjectPublicKeyInfo' field.

Proposed change for the DSA case, section 2.2.2: Replace

   When the id-dsa-with-sha1 algorithm identifier appears as the algo-
   rithm field in an AlgorithmIdentifier, the encoding SHALL omit the
   parameters field.  [...]

by

   When the id-dsa-with-sha1 algorithm identifier appears as the algo-
   rithm field in a signatureAlgorithm field, the encoding SHALL omit the
   parameters field.  [...]

(Similarly for ECDSA.)

(But note that the 'id-dsa-with-sha1' algorithm identifier is never
actually used in 'subjectPublicKeyInfo' fields, so even when taken
out of context the original wording from 2.2.2 is not wrong
-- in 'subjectPublicKeyInfo' fields, you'd see the 'id-dsa' algorithm
identifier instead.)


-- 
Bodo Möller <[e-mail]>
PGP http://www.informatik.tu-darmstadt.de/TI/Mitarbeiter/moeller/0x36d2c658.html
* TU Darmstadt, Theoretische Informatik, Alexanderstr. 10, D-64283 Darmstadt
* Tel. [phone], Fax [phone]