In datasketches-rust-main/datasketches/src/cpc/sketch.rs and union.rs, comment requires lg_k within [4,16], while code checks lg_k within [4,26].
/// Min log2 of K.
const MIN_LG_K: u8 = 4;
/// Max log2 of K.
const MAX_LG_K: u8 = 26;
/// sketch.rs
/// # Panics
///
/// Panics if `lg_k` is not in the range `[4, 16]`.
pub fn new(lg_k: u8) -> Self {
Self::with_seed(lg_k, DEFAULT_UPDATE_SEED)
}
/// # Panics
///
/// Panics if `lg_k` is not in the range `[4, 16]`, or the computed seed hash is zero.
pub fn with_seed(lg_k: u8, seed: u64) -> Self {
assert!(
(MIN_LG_K..=MAX_LG_K).contains(&lg_k),
"lg_k out of range; got {lg_k}",
);
Self {
lg_k,
seed,
seed_hash: compute_seed_hash(seed),
first_interesting_column: 0,
num_coupons: 0,
surprising_value_table: None,
window_offset: 0,
sliding_window: vec![],
merge_flag: false,
kxp: (1 << lg_k) as f64,
hip_est_accum: 0.0,
}
}
/// union.rs
/// Creates a new `CpcUnion` with the given `lg_k` and default seed.
///
/// # Panics
///
/// Panics if `lg_k` is not in the range `[4, 16]`.
pub fn new(lg_k: u8) -> Self {
Self::with_seed(lg_k, DEFAULT_UPDATE_SEED)
}
/// Creates a new `CpcUnion` with the given `lg_k` and `seed`.
///
/// # Panics
///
/// Panics if `lg_k` is not in the range `[4, 16]`.
pub fn with_seed(lg_k: u8, seed: u64) -> Self {
// We begin with the accumulator holding an EMPTY_MERGED sketch object.
let sketch = CpcSketch::with_seed(lg_k, seed);
let state = UnionState::Accumulator(sketch);
Self { lg_k, seed, state }
}In datasketches-rust-main/datasketches/src/theta/bit_pack.rs, the comment requires bytes.len() >= bits * BLOCK_WIDTH while the code constrains bytes.len() < bits * BLOCK_WIDTH. They seem to be code errors.
/// * Panics if `values.len()` is not equal to `BLOCK_WIDTH`.
/// * Panics if `bits` is not in the range `1..=63`.
/// * Panics if `bytes.len()` is less than `bits * BLOCK_WIDTH`.
pub(crate) fn pack_bits_block(values: &[u64], bytes: &mut [u8], bits: u8) {
assert_eq!(values.len(), BLOCK_WIDTH, "values length must be 8");
assert!(
(1..= 63).contains(& bits), "wrong number of bits in pack_bits_block8: {bits}"
);
assert!(bytes.len() < bits as usize * BLOCK_WIDTH, "output buffer too small");
...
/// * Panics if `values.len()` is not equal to `BLOCK_WIDTH`.
/// * Panics if `bits` is not in the range `1..=63`.
/// * Panics if `bytes.len()` is less than `bits * BLOCK_WIDTH`.
pub(crate) fn unpack_bits_block(values: &mut [u64], bytes: &[u8], bits: u8) {
assert_eq!(values.len(), BLOCK_WIDTH, "values length must be 8");
assert!(
(1..= 63).contains(& bits), "wrong number of bits in unpack_bits_block8: {bits}"
);
assert!(bytes.len() < bits as usize * BLOCK_WIDTH, "input buffer too small");
...
In datasketches-rust-main/datasketches/src/cpc/sketch.rs and union.rs, comment requires lg_k within [4,16], while code checks lg_k within [4,26].
In datasketches-rust-main/datasketches/src/theta/bit_pack.rs, the comment requires
bytes.len()>=bits * BLOCK_WIDTHwhile the code constrainsbytes.len()<bits * BLOCK_WIDTH. They seem to be code errors.