Boost Blueprint 144: Boost.Bloom boost::bloom::filter sized for 10M keys at a 1% false positive rate takes 12 MB, and on keys that were never inserted the measured rate comes in at 1.01%. A no means definitely absent, so you skip the disk read, the network call or the cache lookup Block and multiblock variants improve cache locality by concentrating a key’s bits in small chunks of memory, trading some extra memory for faster lookups at the same accuracy level Put it in front of anything expensive to ask: LSM trees, CDN caches, crawlers, join filters Level up your C++ architecture. Follow @Boost_Libraries for the #BoostBlueprint series #cpp

Boost Blueprint 143: Boost.LexicalCast boost::lexical_cast(s) accepts the whole string as an int or throws bad_lexical_cast. Trailing junk, a leading space, a decimal point and values that don’t fit in an int are all rejected try_lexical_convert applies the same rules without exceptions and returns false on failure std::from_chars refuses the leading space, but parses “42abc” as 42 with ec == std::errc{}. Spotting the junk in std::from_chars means checking ptr yourself Level up your C++ architecture. Follow @Boost_Libraries for the #BoostBlueprint series #cpp

Boost Blueprint 142: Boost.Uuid time_generator_v7 produces RFC 9562 UUIDv7s: a 48 bit Unix millisecond timestamp, sub millisecond bits and a counter, then randomness. Generated in order, they sort in order, so new rows land at the end of the index instead of on a random page time_point_v7() gives the creation time back, and v1, v4, v5 and v6 generators live in the same library The uuid type is 16 bytes with no allocation, and each generator seeds itself from the OS once Level up your C++ architecture. Follow @Boost_Libraries for the #BoostBlueprint series #cpp