Release history

libasdf 0.1.0a2 (2025-12-05)

General

  • Changed version scheme to follow PEP-440 particularly for pre-release tags.

  • Fixed missing CMake files in release tarball.

Feature

  • Support for reading compressed block data with zlib or bzip2.

    Includes experimental “lazy decompression” (Linux only at the moment) that can transparently decompress blocks sequentially on an as-needed basis (e.g. it’s possible to read just the first few pages of a block without full decompression). (#37)

  • Support for reading lz4 compressed block data as produced by the Python asdf. (#88)

Bugfix

  • Fix bug in correctly identifying the number of binary blocks when the block index is missing or corrupt. (#93)

  • Fixed segfault when passing an unknown log level name to ASDF_LOG_LEVEL

libasdf 0.1.0-alpha1 (2025-11-12)

Misc

  • Fixed building with CMake

libasdf 0.1.0-alpha0 (2025-11-12)

General

  • Preview alpha release of libasdf! Read support (only) for ASDF files; write support will be in a later version.

    Supports much of the ASDF 1.6.0 standard, albeit with the following key features still missing:

    • Reading compressed block data

    • ndarrays with inline data

    • ndarray masks

    • ndarrays with compound datatypes

    • External block data for ndarray is not supported yet

      • Some preliminary support exists for parsing compound datatypes, but no routines exist yet for reading records/columns out of ndarrays.

    • YAML anchors in the tree

    • Not all of the core schemas are implemented yet (complex, externalarray, etc.)

    • …and likely other less common minor features.

    All of these will be addressed in future releases, likely with priority based on demand.

Misc

  • This release also contains a preview of the libasdf-gwcs extension, which partially supports reading GWCS objects into C-native datastructures (not actually evaluating the GWCS transforms, however).

    This will later be moved to a separate libasdf extension plugin, but for this release it is included in the main library for ease of evaluation.