[issue27272] random.Random should not read 2500 bytes from urandom

[Tim Peters]

Tim Peters added the comment:

Ah!  Yes, .getrandbits(N) outputs remain vulnerable to equation-solving in Python 3, for any value of N.  I haven't seen any code where that matters (may be "a security hole"), but would bet some _could_ be found.

There's no claim of absolute security here.  To the contrary.  What I'm opposed to is making _all_ naive code vulnerable to easy script-kiddie brute force attacks against lame seeding.

The kinds of things people _were_ jumping up & down about were the many instances of stuff like this on the web:

https://stackoverflow.com/questions/3854692/generate-password-in-python

Again, I'd be impressed if you could write code under Python 3 to deduce the MT state from any number of outputs from his naive approach in reasonable time.  Of course he should be using urandom() instead (as an unaccepted answer urges) - but much code plain doesn't, and in Python 3 it's resistant to any attack the PHP paper exposed.

Make seeding lame again, and the easiest attacks can succeed again (the equation-solving stuff remains a footnote to me).

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue27272>
_______________________________________