Releases: HDFGroup/h5pyd
Release v 0.8.0
Added support for OpenID and Active Directory authentication
Changed hsload --s3link option to --link, added support for --link with VLEN data
Pass credentials when accessing external links
Added create_dataset_like and create_scale methods (h5py compat)
Support HTTP Compression
Misc. bug fixes
Release v 0.7.0
Updated Travis test to not depend on Kita Lab
Fix for dimscales with anonymous datasets
Support loading HDF5 files stored in S3 without downloading the file
Note: to load an HDF5 file that is stored on S3 (e.g. s3://mybucket/myfile.h5) use hsload with the --s3link option:
$hsload --s3link s3://mybucket/myfile.h5 /home/myfolder/
This will copy the file metadata to HSDS storage, but chunk data will be accessed from the original file.
Release v 0.6.0
- Return extra chunk properties for CHUNK_REF layouts
- Support for accessing anonymous datasets
- Remove py27 compatibility code
- Removed support for Python 3.5, added support for Python 3.8
- Fix for get root Folder object
Release v 0.5.0
- Dropped support for Py2
- Changed default File mode to read-only
- Fixed hsls utilization report
- Support hsget/hsload for domains including Tables
- Changed min numpy version to v1.17.3
- Fix for hsinfo when endpoint is given
- Return dict for dset.layout when chunk type is not H5D_CHUNKED - see https://github.com/HDFGroup/hsds/blob/master/docs/design/single_object/SingleObject.md
Release v 0.4.4
Added multi-file query functionality in Folder class.
Added query option to hsls.
Release v 0.4.0
Added PyTables like Tables class.
Removed read_where method from Dataset.
Release v 0.3.4
Changes since v0.3.3:
Added retry parameter to File constructor
Various bug fixes