Module shared_bit_vec 

SharedBitVec - cross-process bit-packed boolean array.

Fixed-size bit array stored in an MMF. Each underlying u64 word is operated on atomically (fetch_or for set, fetch_and for clear) so concurrent writers don’t lose updates.

Layout

+---------------------------+
| BitVecHeader (64B)        |
|   magic, capacity_bits    |
+---------------------------+
| words[ceil(cap/64)]       |  AtomicU64 each
+---------------------------+

Concurrency

• set(i) -> word.fetch_or(1 << bit, AcqRel). Multiple writers setting distinct bits in the same word compose correctly (RMW; no lost updates).
• clear(i) -> word.fetch_and(!(1 << bit), AcqRel).
• toggle(i) -> word.fetch_xor(1 << bit, AcqRel).
• get(i) -> word.load(Acquire) & (1 << bit) != 0.
• set_range(lo, hi) / clear_range(lo, hi) use RMW on boundary words and store on fully-covered interior words. Interior stores are safe because they overwrite all 64 bits; no concurrent writer can be modifying interior bits THIS call expects to keep (we’re setting/clearing them all).

Use cases

• Cross-process set membership (presence/absence flags).
• Bloom filter backing array.
• Allocation bitmaps (slot in use / free).
• Feature flag arrays.
• Multi-process work-stealing claim bits.

Structs

BitVecHeader

SharedBitVec

Enums

BitVecError

Constants

BITS_PER_WORD

BITVEC_MAGIC

Functions

bit_vec_file_size