2 ms·
You don't need luck, you just need enough timing data on the collisions and ordering and then use a solver iteratively to produce testcases to get down to the s
by rurban 10y ago
You don't need luck, you just need enough timing data on the collisions and ordering and then use a solver iteratively to produce testcases to get down to the seed. With normal sized hash tables and measurable collision timings you work with 10 bits. It's a matter of a few hours CPU time.
Exploiting public JSON interfaces e.g sounds easiest.
My claim stands that SipHash for hash tables is pure snake oil. Use a proper hash table instead, esp. when they are faster also. Linked lists are 1980ies technology. Even djb can make mistakes. But to his rescue his old solution to this problem is still one of the best. I would rather blame the other guy who came up with these questionable recommendations, and esp. the complang guys for trusting snake oil.