v3.1.0

What's Changed

Added

• Batch-at-a-time reads across the C ABI and all three bindings — a sheet no longer has
  to be materialized in one call, so peak memory is one batch instead of one sheet. New
  exports: xl_typed_reader_open/xl_typed_reader_next/xl_typed_reader_close for typed
  reads, and xl_parse_arrow_stream over the Arrow C stream interface.

  • Python: Workbook.iter_parse_typed, to_record_batch_reader, plus iter_pandas and
    iter_polars.
  • C++: xl::typed_reader<T> and xl::arrow_stream<T>.
  • Rust: Workbook::typed_chunks and arrow::parse_arrow_stream (behind the arrow
    feature).

  The .NET reader already streams row-by-row by construction and is unchanged. Each
  binding's README documents the semantics: one chunked read per workbook, what
  batch_size means, and what invalidates a live reader.

• find_package(excelreader) for the C++ package — cmake --install now installs the
  headers, the native binary and a config package, so a downstream project can consume an
  install prefix instead of vendoring the repo. The existing FetchContent route is
  unchanged, and the install rules stay off when cpp is pulled in as a subproject.

• Shared column style and width validation across the XLSX, XLSB and XLS sheet writers, so
  an invalid column definition is rejected the same way in all three.

Fixed

• A second chunked read on the same workbook is now rejected instead of producing undefined
  results, and xl_typed_reader_open zeroes *out_reader on every failure path — a caller
  that checks the out-param rather than the return code can no longer read a stale pointer.
• Guards against double-buffered batches and against use of an already-released Arrow
  stream; the Arrow stream's release, error and schema contracts were tightened to match
  what the C Data Interface requires.
• OpenTransient fault ordering, and several inaccuracies in the native C header.
• Agile encryption: the CBC path now verifies the transform consumed the whole segment
  instead of silently accepting a short read.
• Python: iter_pandas got its own pyarrow guard, and split_blocks is pinned so the
  pandas conversion no longer depends on a pyarrow default that can change.

Performance

• Agile package encryption reuses one CBC transform across segments instead of rebuilding
  it per segment.
• xl_parse_arrow now routes through the same resumable session as the chunked path, with
  its memory ceiling pinned by the test suite.

No public .NET API changes in this release — PublicAPI.Shipped.txt is untouched across
ExcelReader.Core and ExcelReader.Arrow, both target frameworks. The C ABI only gains
exports; every existing one keeps its signature and behavior.

Full Changelog: v3.0.2...v3.1.0

v3.0.2

What's Changed

Added

• ECMA-376 standard encryption support (read-only) — workbooks encrypted with the
  "standard" scheme (2.2/3.2/4.2: AES-ECB, SHA-1, 50,000 fixed iterations) now open the
  same way agile-encrypted ones already did, via Excel.Open(..., options) with a
  password. Validated against a real third-party fixture (Apache POI's published
  VelvetSweatshop-encrypted file), not just a self-forged one.
• Rust bindings: AllRows/read_all_blob for columnar bulk row reads.

Fixed

• Two container-parsing bugs surfaced by validating standard encryption against a genuine
  third-party file: an incorrect verifierHashSize bound, and the CFB reader demanding a
  full physical sector on a container's final, unpadded stream.
• xl_last_error/xl_read_all_blob were declared in the native C header but missing from
  the Windows .def export files — a latent link error for native/Rust/Python consumers.
• ExcelDataReader.Read() corrected pending-row handling.

Performance

• Standard-encryption package decryption reuses one cipher transform across all segments
  instead of rebuilding it per 4096-byte segment (~3.8x on a large package).
• ExcelDataReader.GetBytes/GetChars no longer round-trip through an allocated string.
• ExcelReader.Arrow's timestamp column reuses the library's existing allocation-free
  date/text parser instead of a duplicate hand-rolled one.
• Test suite: the encrypted-container mutation test now runs in parallel (deterministic
  failure reporting preserved) — full suite ~40s → ~23s.

No public API changes in this release — PublicAPI.Unshipped.txt is empty across
ExcelReader.Core and ExcelReader.Arrow, both target frameworks.

• feat: ECMA-376 standard encryption support, native bulk-read export, and reader/crypto performance fixes by @GabrielMarquezMatte in #105

Full Changelog: v3.0.1...v3.0.2

v3.0.1

What's Changed

• chore: promote public API for v3.0.0 by @github-actions[bot] in #103
• Enhance public API and add trimming compatibility attributes by @GabrielMarquezMatte in #104

Full Changelog: v3.0.0...v3.0.1

v3.0.0

⚠️ Breaking changes

• IExcelRecordMap<T>.ConfigureExcelRecordMap is now generic in the row writer.
  Old: static abstract void ConfigureExcelRecordMap(ExcelRecordMapBuilder<T> builder);
  New: static abstract void ConfigureExcelRecordMap<TRow>(ExcelRecordMapBuilder<T, TRow> builder) where TRow : IRowWriter;

  This lets the mapped record-writer path compile each column's write action against the concrete row writer (XlsxRowWriter, XlsbRowWriter, ...) instead of dispatching through the IRowWriter interface on every cell — the same optimization the reflection-based writer already had via its per-TRow plan.

  Who's affected: only hand-written IExcelRecordMap<T> implementations. If your types use [ExcelSerializable] (the source generator), nothing to do — just recompile. A hand-written implementation needs updating:

  // before
  public static void ConfigureExcelRecordMap(ExcelRecordMapBuilder<MyRecord> builder) { ... }
  
  // after
  public static void ConfigureExcelRecordMap<TRow>(ExcelRecordMapBuilder<MyRecord, TRow> builder)
      where TRow : IRowWriter
  { ... }

New: encrypted OOXML writing

• Excel.EncryptPackage / EncryptPackageAsync: encrypt an OOXML package into an agile-encrypted CFB container (file and stream overloads). Pairs with the encryption reading support shipped in v2.3.0 — round-tripping a password-protected workbook no longer needs a third-party tool.
• Underlying pieces now public/available: agile EncryptionInfo descriptor serialization, the write direction of agile key derivation, and an OLE compound writer generalized to N streams with mini-FAT support.

New: ADO.NET access

• ExcelDataReader implements IDataReader, giving Excel/CSV data a standard IDataReader surface (GetValue, GetSchemaTable, GetOrdinal, typed GetXxx accessors, etc.) for code that expects one — DataTables, data-binding, existing ADO.NET pipelines.
• ExcelRowReaderExtensions.Sheets() / ExcelSheet: enumerate a workbook's sheets (index + name) without opening each one.

New: native bindings parity

• Python: SheetWriter, a streaming write handle — write_row with type inference, in-memory streaming, and columnar writing. Encrypted writing wired into the native bindings.
• Rust: Workbook::rows, a zero-allocation lending row cursor, and Workbook::read_all_decoded for an RAII row set.
• C++: the same Workbook::rows buffer-reusing cursor and read_all_decoded counterparts.
• ExcelReader.Arrow: RecordBatch write support (previously read-only).

Fixes

• XLSB integers are ~2x smaller. Numeric cells with an integral value in [-2^29, 2^29) now encode as the 4-byte BrtCellRk (fInt) form instead of always paying for the 8-byte BrtCellReal — matching what Excel itself emits for a sheet of integers.
• XlsbRowWriter.Write<T> overflow errors now name the type. A caller-defined IUtf8SpanFormattable that overflows to ±Infinity during numeric conversion previously surfaced a bare "non-finite value" error; it now names the source type, matching the message you already got for a non-numeric value.
• Corrupt mini-stream sectors fail loudly. ReadMiniStream now throws InvalidDataException for an out-of-bounds mini-sector slice instead of letting the read overflow silently.
• Fixed a NativeAOT startup crash affecting the Python wheel and C/C++/Rust bindings. The native library no longer pins IlcInstructionSet=native. That setting baked the build machine's CPU features into the binary as a hard requirement; because the binary is built on one machine and consumed on another (CI runner, or an end user's machine after installing the wheel), a CPU-feature mismatch made the process abort at startup with "The required instruction sets are not supported by the current CPU." — intermittently in CI, and potentially for any user whose CPU lacked whatever the build host happened to have (e.g. AVX-512). Vectorized code paths still light up at runtime on capable CPUs via Vector128/256.IsHardwareAccelerated checks, which are unaffected by this change.
• Data integrity checks strengthened; mapped-record-writer benchmarks added for the new ExcelRecordMapBuilder<T, TRow> path.
• Dependency bumps: dorny/paths-filter 3→4, analyzer packages.

Compatibility

• Reading is unaffected — no breaking changes to any parser, reader, or the generated IExcelRowMap<T> side.
• .NET 8 and .NET 10 both supported, as before.

v2.3.0

What's Changed

• Streaming writer handle, CLI tool, schema inference, and write/Arrow bindings for C++/Rust/Python by @GabrielMarquezMatte in #93
• Fixing docs by @GabrielMarquezMatte in #94
• chore: promote public API for v2.2.0 by @github-actions[bot] in #95
• Bump the benchmarks group with 1 update by @dependabot[bot] in #97
• Bump the analyzers group with 6 updates by @dependabot[bot] in #96
• Release v2.3.0: encrypted workbooks, parallel CSV, ExcelReader.Arrow package by @GabrielMarquezMatte in #98

Full Changelog: v2.1.3...v2.3.0

v2.2.0

What's Changed

• Streaming writer handle, CLI tool, schema inference, and write/Arrow bindings for C++/Rust/Python by @GabrielMarquezMatte in #93
• Fixing docs by @GabrielMarquezMatte in #94

Full Changelog: v2.1.3...v2.2.0

v2.1.3

What's Changed

• chore: promote public API for v2.1.2 by @github-actions[bot] in #83
• Native bindings performance and fuzz testing improvements by @GabrielMarquezMatte in #84
• Bump Sep from 0.15.2 to 0.17.0 by @dependabot[bot] in #92
• Bump Microsoft.CodeAnalysis.Analyzers and Microsoft.CodeAnalysis.CSharp by @dependabot[bot] in #91
• Bump the test group with 2 updates by @dependabot[bot] in #90
• build(deps): bump softprops/action-gh-release from 2 to 3 by @dependabot[bot] in #88
• build(deps): bump actions/setup-python from 6 to 7 by @dependabot[bot] in #87
• build(deps): bump actions/download-artifact from 6 to 8 by @dependabot[bot] in #86
• Implementing C++ and Rust wrappers packages for native bindings with parsing and reading by @GabrielMarquezMatte in #85

Full Changelog: v2.1.2...v2.1.3

v2.1.2

What's Changed

• Changing python project name by @GabrielMarquezMatte in #80
• Corrigindo tag de linux by @GabrielMarquezMatte in #81
• New native APIs for improved performance by @GabrielMarquezMatte in #82

Full Changelog: v2.1.1...v2.1.2

v2.1.1

What's Changed

• Changing python project name by @GabrielMarquezMatte in #80

Full Changelog: v2.1.0...v2.1.1

v2.1.0

What's Changed

• chore: promote public API for v2.0.1 by @github-actions[bot] in #67
• Promote public API and add fuzz testing for ExcelReader by @GabrielMarquezMatte in #68
• Fix DateTime parsing for ISO dates below year 100 and add tests by @GabrielMarquezMatte in #69
• Bump SharpFuzz from 2.1.1 to 2.3.0 by @dependabot[bot] in #77
• Bump Sep from 0.15.1 to 0.15.2 by @dependabot[bot] in #76
• Bump Microsoft.Testing.Extensions.CodeCoverage from 18.9.0 to 18.10.0 by @dependabot[bot] in #75
• Bump the benchmarks group with 1 update by @dependabot[bot] in #74
• Bump the analyzers group with 6 updates by @dependabot[bot] in #73
• chore(deps): bump actions/checkout from 4 to 7 by @dependabot[bot] in #72
• chore(deps): bump actions/upload-artifact from 4 to 7 by @dependabot[bot] in #71
• chore(deps): bump actions/setup-dotnet from 4 to 6 by @dependabot[bot] in #70
• Update GitHub Actions and analyzer dependencies by @GabrielMarquezMatte in #78
• feat: native C ABI + Python bindings (excelreader package), with bulk read, date parsing, and PyPI publishing by @GabrielMarquezMatte in #79

Full Changelog: v2.0.1...v2.1.0