NEWS
tantivyr 0.1.3
- Updated the bundled 'Tantivy' engine to 0.27.0 and refreshed the vendored
'Rust' crates. The on-disk index format is unchanged (v7), so existing
indexes keep working. This also replaces the
lru crate with a version
that fixes RUSTSEC-2026-0253 (the unsound code was never reachable from
'tantivyr').
- Fixed the macOS linker warnings reported by CRAN's M1mac additional checks
("object file was built for newer 'macOS' version than being linked"): on
macOS the Cargo build now pins
MACOSX_DEPLOYMENT_TARGET to rustc's own
default for the platform (11.0 on Apple silicon) unless it is already set,
so the C sources compiled by Cargo carry a minimum version that is never
newer than the one R links against.
tantivyr 0.1.1 (2026-09-22)
- New
fold_accents argument in tnt_text() and tnt_index_df(): removes
diacritics from indexed words and queries (ASCII folding), so a query typed
without accents, such as "orcamento", finds "orçamento". Stored text and
snippets keep their accents.
- New
pt_news dataset: 36 short, fictional news stories in Brazilian
Portuguese for trying out stemming, stop words, filters and highlights.
- New vignette "Searching Portuguese text" covering stemming, stop words,
accents and the query syntax.
- Updated the bundled 'Tantivy' engine to 0.26.2 (bug-fix release; no API
changes) and refreshed the vendored 'Rust' crates.
SystemRequirements now states the minimum supported 'Rust' version
(rustc >= 1.88), so an outdated toolchain is reported clearly at configure
time instead of failing during compilation.
- Corrected the example return value in the
tantivy_version() documentation.
tantivyr 0.1.0 (2026-07-15)
- Initial release.
tnt_index_df() indexes a data frame in one call (auto-schema).
- Explicit schema API:
tnt_schema() with tnt_text(), tnt_i64(),
tnt_u64(), tnt_f64(), tnt_bool(), tnt_date() and tnt_json().
tnt_index() creates or opens on-disk and in-memory indexes.
tnt_add(), tnt_commit(), tnt_delete() and tnt_update() provide
incremental updates.
tnt_search() returns a tibble with BM25 scores, structured filters
(comparison expressions or Tantivy query strings), snippet highlighting and
fast-field ordering; tnt_count() returns total match counts.
- Portuguese and English stemming and stop words, plus other Snowball languages.