Skip to content

v0.5.1

Choose a tag to compare

@stephenberry stephenberry released this 27 May 21:26
· 72 commits to main since this release

Paged Fixed Array data block support (read + write), fixing chunked datasets with more than 1024 chunks (#14).

Previously the writer emitted a flat data block while advertising the paged page size, corrupting files above 1024 chunks; the reader rejected paged Fixed Array data blocks outright. Both directions are now spec-compliant and verified against the reference C HDF5 library. No public API changes.

See the changelog for details.