…n to reflect the API changes.
Crypto pubkey types
…one on the DRBG library (which seems more portable).
Move to CryptoRandomGen from RandomGen
…domGen class). The tests were also updated and complete successfully on my machine. This is a move to a class of generators intended for instantiation only by cryptographically secure generators that allow for failure and reseed. For RSA, failures are in the form of exceptions (of type GenError). This patch has one semantic change besides the obvious RandomGen->CryptoRandomGen replacement! The rsaes_oaep_encrypt function was accepting an integer seed in the range of +/-2^29 (or whatever Int size is on the host) and translating it to a bytestring in the range of [0..2^29]. Thus, we were getting only 4 or 8 bytes of entropy (maximum) for a 32 byte long bytestring (assuming a default of SHA256). It now accepts generators and will generate random bytestrings of length equal to the size of the digest for the provided hash algorithm. I'm not sure what, if any, security implications this has - see RFC 3447 Sec 7.1.1 and see line 247 in RSA.hs (line 248 pre-patch, I think).
Dear Adam, Attached is a patch to let RSA build with the newest random (22.214.171.124). I had to create the patch using diff because I couldn't find the source repository in the cabal file. Could you also add that? Thanks, Bas
…sts test, to avoid a bug hit by Michael Snoyman (email@example.com).
… new test framework stuff.
Thanks to Kevin Ulin <firstname.lastname@example.org> for the catch!
…but adds a warning in others), and bump the version number.