On Mon, Feb 15, 2021 at 08:25:27PM +0900, Michael Paquier wrote:
> Again a new rebase, giving v5:
> - Fixed the APIs to return -1 if the caller gives NULL in input, to be
> consistent with cryptohash.
> - Added a length argument to pg_hmac_final(), wiht sanity checks.
So, this patch has been around for a couple of weeks now, and I would
like to get this part done in 14 to close the loop with the parts of
the code that had better rely on what the crypto libs have. The main
advantage of this change is for SCRAM so as it does not use its own
implementation of HMAC whenever possible.
Any objections?
--
Michael
