Repository navigation
v0.2.0
Breaking release: match locations are reported by integer sequence id (SeqId) instead of by FASTA header string, removing one String allocation per located occurrence. Header <-> id lookup is O(1) in both directions via seq_header / seq_id, and sequence headers must now be unique (FmIndexError::DuplicateHeader). The on-disk index format is unchanged, but an index built by 0.1.0 from duplicate headers is now rejected on load and must be rebuilt. See CHANGELOG.md for the full entry.