A status update on Debian's reproducible builds

It's also an inherent part of a linear hash algorithm - buckets are created and destroyed continuously, and must be able to hold multiple keys. So a linear scan of the hash table - while keys will be approximately in the same order - will return keys with an identical hash function in random order. Sorting them in their buckets would be an unacceptable overhead, even if it's only a small overhead.

(I don't know for certain, but linear hashing definitely isn't unique to Pick - it's used by BerkeleyDB aka Sleepycat, and I believe it's the algorithm behind hashes in Perl and Python.)

Cheers,
Wol