On 01/16/2012 07:46 PM, Samuel Thibault wrote: > Sebastian Ramacher, le Mon 16 Jan 2012 19:34:58 +0100, a écrit : >> Is this a known bug with Hurd's /dev/urandom implementation or does it >> behave differently by design? > > It's a known behavior. It's not by design, the random server just > does not have the loop to fill the buffer. AIUI, that behavior is > posix-compliant, however. Thank you. So I'll probably need to fallback to Crypto.Random.OSRNG.fallback on hurd-i386 which uses Python's os.urandom that just waits on /dev/urandom to return something. Do you have a pointer to the relevant part of POSIX, by any chance? I couldn't find anything. Are /dev/random and /dev/urandom even part of POSIX? Kind regards, -- Sebastian Ramacher
Attachment:
signature.asc
Description: OpenPGP digital signature