NEWS

rgbio 0.4.0

Improved GenBank file read performance by ~25% compared to v0.3.0:
- Moved feature location parsing and validation to Rust for better performance and more consistent error handling.
Added dependency-free format: format = "base".
- This means that rgbio can now be used without any additional dependencies if needed, though the tidy (requires tibble) and bioconductor (requires Biostrings, S4Vectors, IRanges, GenomicRanges) formats are still recommended for most users.

Improved GenBank file read performance by ~25%:
- Improved feature location handling internals with vectorized validation and parsing paths.
- Reduced overhead in read_gbk() feature and sequence extraction by reusing vectorized intermediates.
- Updated write-path validation to validate location vectors in one pass for better performance and consistent errors.
- Expanded internal test coverage for vectorized location validation, including scalar/vector error parity checks.

Initial full release
Complete rewrite of high-level R functions
- Added new primary GenBank I/O API: read_gbk() and write_gbk().
- Added format = "tidy" and format = "bioconductor" output modes for reading.
- Added selective read controls (sequences, features, metadata) and record selectors (records).
- Added write support for named character vectors and DNAStringSet sequence inputs.
- Added multi-record write orchestration and append-mode writing.
- Kept read_genbank() and write_genbank() as deprecated compatibility wrappers.
Added more extensive testing with test files from gb-io.py, gb-io, and biopython