W3C home > Mailing lists > Public > public-webcrypto@w3.org > June 2012

Re: Strawman proposal for the low-level API

From: Wan-Teh Chang <wtc@google.com>
Date: Fri, 22 Jun 2012 16:14:10 -0700
Message-ID: <CALTJjxEA9z9AiyWiU3CC=uET9-9t1yRs_o3HE4ERo6Emfdh4oA@mail.gmail.com>
To: Mike Jones <Michael.Jones@microsoft.com>
Cc: Ryan Sleevi <sleevi@google.com>, David Dahl <ddahl@mozilla.com>, "public-webcrypto@w3.org" <public-webcrypto@w3.org>
On Wed, Jun 20, 2012 at 1:52 PM, Mike Jones <Michael.Jones@microsoft.com> wrote:
> As a data point, the IETF JOSE work intentionally has only algorithm
> identifiers and no separate algorithm parameters.  Instead, the particular
> parameter choices are baked into the identifier.
> So for instance, rather than having a generic RSA signature algorithm and
> hash functions as parameters, there are separate algorithm identifiers such
> as:
>                 RS256 - RSA signature using SHA-256 hash algorithm
>                 RS384 - RSA signature using SHA-384 hash algorithm
>                 RS512 - RSA signature using SHA-512 hash algorithm

I think this is a reasonable approach.  This approach is reflected in
the OIDs for signature algorithms in ASN.1.  For example, PKCS #1
specifies signature algorithm OIDs such as sha1WithRSAEncryption and
sha256WithRSAEncryption, and NIST specifies ecdsa-with-SHA256,
ecdsa-with-SHA384, etc.  PKCS #11 has "combo" signature mechanisms
such as CKM_SHA256_RSA_PKCS and CKM_ECDSA_SHA1.  (Note: PKCS #11 also
has the "sign hash"-style signature mechanisms such as CKM_RSA_PKCS
and CKM_DSA, similar to what I describe next.)

On the other hand, in Windows CryptoAPI and CNG, a signature is
generated by signing a hash that is computed separately. There is a
clear separation of the hash component in the sign and verify
operations.  So they are in line with Ryan Sleevi's proposal :-)

In my opinion, the hardest problem with choosing a signature algorithm
is which hash algorithm should be used.  It turns out the hash
algorithm should match the signature key size.  So "combo" algorithms
such as "RS256" still won't help solve this problem.

Received on Friday, 22 June 2012 23:14:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:17:10 UTC