From: "matz (Yukihiro Matsumoto) via ruby-core" Date: 2026-08-06T06:34:53+00:00 Subject: [ruby-core:126287] [Ruby Feature#22118] Introduce Basic Bit Operations into String Issue #22118 has been updated by matz (Yukihiro Matsumoto). Thank you for the update. The specification in the current Description matches my intent in #note-5, and both points raised in #note-6 are fine as proposed. bit_count and the bit-order keyword. Accepted as written. A population count over the whole string is independent of bit numbering within a byte, so nothing is lost by omitting the keyword here. When range arguments arrive, the keyword becomes meaningful and must be added at that time with the same default as the other bit_* methods. Very large bit offsets. Accepted as written: ArgumentError, raised before the read/write out-of-range rule applies. I agree that this bound belongs in the ticket rather than being left to each implementation's internal width. bit_put (#note-7). mame's point is a fair one, and a writer taking a value may well turn out to be necessary. But let us not put it in this proposal. It overlaps bit_set/bit_clear without replacing bit_flip, and it interacts with the range arguments hasumikin has in mind in #note-9, so it deserves to be designed together with them rather than bolted on now. Adding a method later is easy; removing one is not. Please open a separate ticket for it. Also, I withdraw the reservation about the lsb_first: keyword name recorded in #note-2. The explanation in the Description convinced me, and an encoding or internal flag would carry the bit order invisibly across method calls, which is worse. The specification is settled. Please proceed with the PR. Matz. ---------------------------------------- Feature #22118: Introduce Basic Bit Operations into String https://bugs.ruby-lang.org/issues/22118#change-118379 * Author: hasumikin (hitoshi hasumi) * Status: Open ---------------------------------------- This PR implements basic bit operations into String, incorporating revisions made following the discussion (esp. #note-5) PR URL: https://github.com/ruby/ruby/pull/17353 ## Single-bit read and mutation * `String#bit_get(offset, lsb_first: true) -> 1 | 0 | nil` * `String#bit_set?(offset, lsb_first: true) -> true | false | nil` * `String#bit_set(offset, lsb_first: true) -> self` * `String#bit_clear(offset, lsb_first: true) -> self` * `String#bit_flip(offset, lsb_first: true) -> self` **Note for `bit_set`/`bit_clear`/`bit_flip`:** As these methods are destructive-only (they have no non-destructive counterpart), no `!` is added. The `!` suffix is meant to distinguish the dangerous member of a destructive/non-destructive pair, following the `setbyte` precedent. ## Bitwise operators * `String#bitwise_not -> String` * `String#bitwise_not! -> self` * `String#bitwise_and(other) -> String` * `String#bitwise_and!(other) -> self` * `String#bitwise_or(other) -> String` * `String#bitwise_or!(other) -> self` * `String#bitwise_xor(other) -> String` * `String#bitwise_xor!(other) -> self` ## Popcount * `String#bit_count -> Integer` --- ## API Contracts ### 1. Bit-order keyword #### `lsb_first: true` (default) Within each byte, `offset = 0` is the LSB. Numbering proceeds upward through byte[0] and then continues at the LSB of byte[1]: ```text byte[0] byte[1] offset: 7 6 5 4 3 2 1 0 15 14 13 12 11 10 9 8 bit: b7 b6 b5 b4 b3 b2 b1 b0 b7 b6 b5 b4 b3 b2 b1 b0 ^ ^ LSB LSB offset = byte_index * 8 + bit_in_byte ``` #### `lsb_first: false` for MSB-first Byte order is preserved (`byte[0]` is still first), but within each byte numbering starts at the MSB: ```text byte[0] byte[1] offset: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 bit: b7 b6 b5 b4 b3 b2 b1 b0 b7 b6 b5 b4 b3 b2 b1 b0 ^ ^ MSB MSB offset = byte_index * 8 + (7 - bit_in_byte) ``` The keyword only controls bit numbering within each byte. Byte order itself is unchanged. LSB-first is the more common convention for general in-memory bitmap use, including formats and APIs such as Apache Arrow validity bitmaps, ext4 block bitmaps, Roaring bitmaps, Linux/BSD bitmap APIs, and hardware register descriptions. MSB-first is also needed for domains such as RFC-style network protocol diagrams, PNG low-bit-depth scanlines, BitTorrent bitfields, and some compressed bit streams. This is the reason that we define `lsb_first: true` as the default. The bit-order keyword must be accepted by all `bit_*` methods with the same default; a getter and a setter disagreeing on bit addressing would be disastrous. **One temporary exception:** Since the `bit_count` introduced in this version counts the population of the entire string without taking arguments, it does not accept a bit-order keyword. If arguments for specifying a range are introduced in the future, the bit-order keyword should be introduced at that time. ### 2. Out-of-range behavior Read methods: `String#bit_get` and `String#bit_set?` return `nil` when the bit offset is out of range. On the other hand, mutating methods such as `String#bit_set`, `String#bit_clear`, and `String#bit_flip` raise `IndexError` for an out-of-range bit offset. This follows the same general distinction as read-like access versus write-like access: reading a missing position can return `nil`, but mutating a missing position should not silently do nothing. If the bit offset is too large to be represented internally, all of these methods raise `ArgumentError` before applying the read/write out-of-range rule. This bound is specified explicitly because the representable offset range may differ between Ruby implementations. ### 3. Negative bit offsets Negative bit offsets are rejected with `IndexError`. Although Ruby often uses negative indices to count from the end, combining negative bit offsets with the `lsb_first: true/false` mechanism would make the numbering rule harder to explain and reason about. The API keeps bit offsets as non-negative flat positions from the beginning of the string. ### 4. Length mismatch for binary bitwise operations `String#bitwise_and`, `String#bitwise_or`, `String#bitwise_xor`, and their `!` variants require both operands to have the same byte size. If the byte sizes differ, they raise `ArgumentError`. This avoids implicit truncation or zero-padding. Since these methods operate on strings as fixed-size bitmaps, requiring equal sizes makes the operation explicit and predictable. Silent zero-extension would let corrupted results flow downstream in the typical use cases (checksums, masks); callers who want padding can `ljust` explicitly. ### 5. Encoding All bit operations treat the receiver as a byte sequence. For non-destructive methods that return a String, the return value's encoding is set to `Encoding::BINARY`. Destructive methods do not alter the receiver's encoding, consistent with `setbyte`, which likewise lets a byte-level mutation produce an encoding-invalid string without changing the encoding. -- https://bugs.ruby-lang.org/ ______________________________________________ ruby-core mailing list -- ruby-core@ml.ruby-lang.org To unsubscribe send an email to ruby-core-leave@ml.ruby-lang.org ruby-core info -- https://ml.ruby-lang.org/mailman3/lists/ruby-core.ml.ruby-lang.org/