Skip to content

Tags

Tags give the ability to mark specific points in history as being important

  • v1.2.0

    1eb774ef · Publish v1.2.0 · 
    v1.2.0

Features:

* Add support for arrow v56
* Add support for FixedSizeBinary

Fixes:

* row_iterator: Always call check_schema
* Fully-qualified use of CheckableDataType::check_datatype in the macro

Internal:

* row_iterator: Fix warnings when executing cargo doc
* Switch from edition 2015 to 2021
* Fix Clippy warnings
* Add Cargo.lock to work around conflicting arrow versions in tests

  • v1.1.0

    d5d1de9d · Publish v1.1.0 · 
    v1.1.0

Add support for arrow v53

  • v1.0.0

    e9df8fb8 · Publish v1.0.0 · 
    v1.0.0

New features:

* Add support for naive Decimal128 decoding
* Add support for decoding timestamps from Decimal, so [the full range of ORC timestamps](https://github.com/datafusion-contrib/datafusion-orc/pull/96) is supported
* Add support for decoding dictionary-encoded with primitive values

Changes:

* Update from arrow v50 to v52

Fixes:

* ar_row_derive: Use fully-qualified names to avoid shadowing
* Fix error message on missing columns

  • v0.6.0

    9a624175 · Publish v0.6.0 · 
    v0.6.0

This release forks `orcxx` into a new crate, `ar_row` aimed at providing row-oriented
deserialization of Apache Arrow arrays, by reusing `orcxx`'s deserialization architecture,
traits, and `derive` macro.

This removes all ORC-specific code, except for tests (which now use `datafusion-orc`
to read data files into Arrow structures); and all C++ code.

  • v0.5.0

    d9fe6b6e · Publish v0.5.0 · 
    v0.5.0

* Compile with `-fPIC` to fix some linking issues
* Fix tests
* Add `ORC_USE_SYSTEM_LIBRARIES` env var to use system libraries instead of downloading
* Display nicer error messages when CMake or Make returns an error

  • v0.4.2

    19797269 · Timestamps need to be decoded as two i64 · 
    v0.4.2

Fixes:

* Fix decoding of timestamps, they are not a single i64 as I thought

  • v0.4.0

    17449395 · Publish v0.4.0 · 
    v0.4.0

Breaking:

* Use `thiserror` instead of nested `Result`
* Rename 'utils' module to 'errors'

Additions:

* Add support for deserializing Timestamp as i64 struct field

Internal:

* Replace `unsafe_unwrap` with stdlib's `unwrap_unchecked`

  • v0.3.0

    f6a1c142 · Publish v0.3.0 · 
    v0.3.0

Breaking:

* Make `Reader::row_reader()` take a ref instead of owned RowReaderOptions

Additions:

* Add ParallelRowIterator
* Implement ExactSizeIterator and DoubleEndedIterator for RowIterator
* Implement `row_number()` and `seek_to_row()` for RowReader
* Impl Sync and Clone for RowReaderOptions
* Implement Send for vectors

Fixes:

* Fix empty list deserialization

Internal:

* `orcxx_derive/tests/test1.rs`: Test more batch sizes (+deduplicate)

  • v0.2.3

    9b1818f3 · Publish v0.2.3 · 
    v0.2.3

Documentation:

* Avoid relative links between crate documentation

  • v0.2.2

    2ddf382c · Publish v0.2.2 · 
    v0.2.2

Documentation:

* Copy examples to the README
* Add links between documentation pages

Internal:

* Add pre-commit config
* Remove orcxx/README.md symlink

  • v0.2.1

    9ca7fcbc · Publish v0.2.1 · 
    v0.2.1

Documentation:

* `orcxx_derive`: Document RowIterator with an example

Internal:

* Use system libraries when building on docs.rs

  • v0.2.0

    b5d0ef0e · Publish v0.2.0 · 
    v0.2.0

Breaking:

* RowIterator: Always check the selected kind
* Simplify RowIterator::new() to automatically select columns

Additions:

* `OrcStruct::columns()`
* Support for escaping field names

Internal:

* Fix dependencies between crates + dedup metadata

  • v0.1.0

    441b1e24 · Fix package build and publication · 
    v0.1.0

Initial release.

Provides full read-only access to .orc files through three APIs:

* trees of vectors
* vectors of rows (structures generated with a custom derive)
* iterator on rows (ditto)