Skip to content

Releases: G-Research/ParquetSharp

ParquetSharp 15.0.0-beta2

05 Mar 23:03
925d233
Compare
Choose a tag to compare
Pre-release

What's Changed

  • Document workarounds for writing TimeSpan values by @adamreeve in #422
  • Support the TimeOnly and DateOnly types added in .NET 6 by @adamreeve in #424
  • Add the key management tools API for encryption by @adamreeve in #426

Full Changelog: 15.0.0-beta1...15.0.0-beta2

ParquetSharp 15.0.0-beta1

07 Feb 01:42
3182e68
Compare
Choose a tag to compare
Pre-release

This version of ParquetSharp upgrades the underlying Arrow C++ Parquet library to version 15.0.0.

What's Changed

  • Upgrade C++ Arrow to 15.0.0 by @adamreeve in #417
  • Add support for reading and writing the .NET Half type by @adamreeve in #418
  • Expose page checksum verification reader and writer properties by @adamreeve in #419

Full Changelog: 14.0.2...15.0.0-beta1

ParquetSharp 14.0.2

02 Feb 00:10
50292d0
Compare
Choose a tag to compare

This version of ParquetSharp is identical to 14.0.2-beta1, which has been promoted to stable.

This version of ParquetSharp upgrades the underlying Arrow C++ Parquet library to version 14.0.2.

New features

  • Add Skip at the LogicalColumnReader level by @mfkl in #388
  • Add support for custom conversion logic with the row-oriented API by @jescalada in #393

Bug fixes

  • Fix segfault writing to no longer valid column writer by @adamreeve in #401

Miscellaneous

Full Changelog: 13.0.0...14.0.2

ParquetSharp 14.0.2-beta1

22 Jan 22:57
307a2bd
Compare
Choose a tag to compare
Pre-release

This version of ParquetSharp upgrades the underlying Arrow C++ Parquet library to version 14.0.2.

New features

  • Add Skip at the LogicalColumnReader level by @mfkl in #388
  • Add support for custom conversion logic with the row-oriented API by @jescalada in #393

Bug fixes

  • Fix segfault writing to no longer valid column writer by @adamreeve in #401

Miscellaneous

Full Changelog: 13.0.0...14.0.2-beta1

ParquetSharp 13.0.0

27 Nov 21:55
63ea83d
Compare
Choose a tag to compare

This version of ParquetSharp is identical to 13.0.0-beta1, which has been promoted to stable.

This version of ParquetSharp upgrades the underlying Arrow C++ Parquet library to version 13.0.0, and adds the ability to interoperate with the Arrow C# library, supporting reading Parquet data as Arrow record batches and writing Arrow data to Parquet. See the working with Arrow data documentation for further details.

Full list of changes

New Contributors

Full Changelog: 12.0.1...13.0.0

ParquetSharp 13.0.0-beta1

08 Nov 22:49
3be3f25
Compare
Choose a tag to compare
Pre-release

This version of ParquetSharp upgrades the underlying Arrow C++ Parquet library to version 13.0.0, and adds the ability to interoperate with the Arrow C# library, supporting reading Parquet data as Arrow record batches and writing Arrow data to Parquet. See the working with Arrow data documentation for further details.

Full list of changes

New Contributors

Full Changelog: 12.0.1...13.0.0-beta1

ParquetSharp 12.0.1

09 Jul 22:58
01f2eec
Compare
Choose a tag to compare

This version of ParquetSharp is identical to 12.0.1-beta1, which has been promoted to stable.

What's Changed

  • Fix using internal F# types with the row oriented API by @adamreeve in #332
  • Use ISO 8601 date format for Date.ToString by @adamreeve in #330
  • Allow setting process-wide default writer properties by @adamreeve in #337
  • Don't retry writing in Dispose after encountering an error in ParquetRowWriter by @adamreeve in #344
  • Defer checking decimal precision and type length until creating converter by @adamreeve in #348
  • Add WriteRowSpan method to ParquetRowWriter by @adamreeve in #349
  • Upgrade to Arrow 12.0.1 by @adamreeve in #356

Misc

Full Changelog: 10.0.1...12.0.1

ParquetSharp 12.0.1 Beta1

20 Jun 03:50
d766af3
Compare
Choose a tag to compare
Pre-release

What's Changed

  • Fix using internal F# types with the row oriented API by @adamreeve in #332
  • Use ISO 8601 date format for Date.ToString by @adamreeve in #330
  • Allow setting process-wide default writer properties by @adamreeve in #337
  • Don't retry writing in Dispose after encountering an error in ParquetRowWriter by @adamreeve in #344
  • Defer checking decimal precision and type length until creating converter by @adamreeve in #348
  • Add WriteRowSpan method to ParquetRowWriter by @adamreeve in #349
  • Upgrade to Arrow 12.0.1 by @adamreeve in #356

Misc

Full Changelog: 10.0.1...12.0.1-beta1

ParquetSharp 10.0.1

12 Feb 21:07
054d5cd
Compare
Choose a tag to compare

This version of ParquetSharp is identical to 10.0.1-beta1, which has been promoted to stable.

This version upgrades Arrow to 10.0.1, which writes Parquet files with the version 2.4 format by default, and dictionary encoded data will use the RleDictionary encoding rather than PlainDictionary. Version 1 files can continue to be written by specifying the file format version in the WriterProperties.

What's Changed

  • Integrate Slack Github Action by @pavlovic-ivan in #308
  • GitHub actions updates to fix deprecation warnings by @adamreeve in #310
  • Support round-tripping nested values by @adamreeve in #301
  • Avoid use of ManagedOutputStream and RandomAccessFile after GC by @adamreeve in #313
  • Support reading decimal data with int32 or int64 physical type by @adamreeve in #315
  • Convert paths to extend length format on Windows and add documentation on long path support by @adamreeve in #311
  • Allow creating a ParquetFileReader/Writer directly from a .NET stream by @adamreeve in #320
  • Support of optional nested-nested types reading by @oldukhno in #319
  • Upgrade to Arrow 10.0.1 by @adamreeve in #321
  • Change version to 10.0.1 to match Arrow by @adamreeve in #323

Incompatible API Changes

  • LogicalColumnStream.SchemaNodesPath has been replaced by LogicalColumnStream.GetSchemaNodesPath

New Contributors

Full Changelog: 8.0.0...10.0.1

ParquetSharp 10.0.1 Beta1

14 Dec 06:40
386d91b
Compare
Choose a tag to compare
Pre-release

This version upgrades Arrow to 10.0.1, which writes Parquet files with the version 2.4 format by default, and dictionary encoded data will use the RleDictionary encoding rather than PlainDictionary. Version 1 files can continue to be written by specifying the file format version in the WriterProperties.

What's Changed

  • Integrate Slack Github Action by @pavlovic-ivan in #308
  • GitHub actions updates to fix deprecation warnings by @adamreeve in #310
  • Support round-tripping nested values by @adamreeve in #301
  • Avoid use of ManagedOutputStream and RandomAccessFile after GC by @adamreeve in #313
  • Support reading decimal data with int32 or int64 physical type by @adamreeve in #315
  • Convert paths to extend length format on Windows and add documentation on long path support by @adamreeve in #311
  • Allow creating a ParquetFileReader/Writer directly from a .NET stream by @adamreeve in #320
  • Support of optional nested-nested types reading by @oldukhno in #319
  • Upgrade to Arrow 10.0.1 by @adamreeve in #321
  • Change version to 10.0.1 to match Arrow by @adamreeve in #323

Incompatible API Changes

  • LogicalColumnStream.SchemaNodesPath has been replaced by LogicalColumnStream.GetSchemaNodesPath

New Contributors

Full Changelog: 8.0.0...10.0.1-beta1