3 ms·
I don't get why the discovery of a non-preimage attack is causing so much consternation. If the mere existence of collisions is not acceptable to your VCS, the
by bandrami 10y ago
I don't get why the discovery of a non-preimage attack is causing so much consternation.
If the mere existence of collisions is not acceptable to your VCS, then your VCS can't use a hash, period.
If you're worried about an intentional attack, it's no closer today than it was last week: the attacker doesn't control the output hash of the collision, or either input.
- Taek 10y ago> If the mere existence of collisions is not acceptable to your VCS, then your VCS can't use a hash, period What? It could use a secure hash. Collisions take more energy than the universe has to discover.
- bandrami 10y ago"Secure" is a time-constrained term. But any hash, whatsoever, has collisions, so if collisions are unacceptable, your VCS can't use them. (The point being that a VCS should handle collisions gracefully no matter what has is used.)
- lmm 10y agoNo VCS is 100% secure against the possibility of catastrophic failure. If an asteroid wiped out all life on earth there is no VCS that can handle that gracefully. So as long as a hash collision is less likely than that, using a hash and not handling collisions gracefully doesn't make the VCS substantially less safe.