Struct SortColumns 

pub(super) struct SortColumns<'a> {
    lower_pool: &'a [u8],
    name_off: &'a [u32],
    name_len: &'a [u16],
    size_lo: &'a [u32],
    size_ovf: &'a FxHashMap<EntryId, u64>,
    mtime: &'a [i64],
}

Borrowed view of the sort-key columns, so permutation maintenance can hold &mut permutation arrays while comparing through the one definition of each key’s order (a drifting duplicate of cmp_by would silently corrupt the merge).

Fields

lower_pool: &'a [u8]

name_off: &'a [u32]

name_len: &'a [u16]

size_lo: &'a [u32]

size_ovf: &'a FxHashMap<EntryId, u64>

mtime: &'a [i64]

Implementations

impl<'a> SortColumns<'a>

pub(super) const fn new( lower_pool: &'a [u8], name_off: &'a [u32], name_len: &'a [u16], size_lo: &'a [u32], size_ovf: &'a FxHashMap<EntryId, u64>, mtime: &'a [i64], ) -> Self

fn lower_name(&self, id: EntryId) -> &[u8]

fn size_of(&self, id: EntryId) -> u64

pub(super) fn cmp_by(&self, key: SortKey, a: EntryId, b: EntryId) -> Ordering

Strict total order (id tie-break): no two distinct ids compare equal, which is what makes merged permutations byte-deterministic.