On Tue, May 12, 2015 at 06:10:39PM +0000, Salz, Rich wrote: > You can't easily have test vectors for DSA signatures since they > include a random. Any test vector would have to include the random, > and any API would have to be able to accept the random as part of the > "sign" API. Verification should be okay. It'd be nice to have derandomized *DSA forms for OpenSSL. CFRG is on the case, thankfully, so eventually there should be a derandomized ECC signature scheme in OpenSSL. (Assuming the consensus ends up being in favor of having a deterministic, state-less signature scheme.) Nico --
