1+ What's new in h5py 3.3
2+ ======================
3+
14New features
25------------
36
7+ * Compatibility with the upcoming HDF5 1.12.1 and possibly 1.14 (:pr: `1875 `).
48* H5T_BITFIELD types will now be cast to their ``numpy.uint `` equivalent by default
59 (:issue: `1258 `). This means that no knowledge of mixed type compound dataset
610 schemas is required to read these types, and can simply be read as follows:
@@ -14,9 +18,11 @@ New features
1418 .. code ::
1519
1620 arr = dset.astype(numpy.bool_)[:]
17- * Validate key types when accessing groups.
18- * A new :meth: `Group.build_virtual_dataset ` method acting as a context manager
19- to assemble virtual datasets.
21+
22+ * Key types are validated when accessing groups, to give more helpful errors
23+ when a group is indexed like a dataset (:pr: `1856 `).
24+ * A new :meth: `.Group.build_virtual_dataset ` method acting as a context manager
25+ to assemble virtual datasets (:pr: `1905 `).
2026* If the source and target of a virtual dataset mapping have different numbers
2127 of points, an error should now be thrown when you make the mapping in the
2228 :class: `VirtualLayout `, rather than later when writing this into the file.
@@ -35,7 +41,7 @@ Bug fixes
3541---------
3642
3743* Trying to open a file in append mode (``'a' ``) should now give clearer
38- error messages when the file exists but can't be opened.
44+ error messages when the file exists but can't be opened ( :pr: ` 1902 `) .
3945* Protect :func: `h5py.h5f.get_obj_ids ` against garbage collection invalidating
4046 HDF5 IDs while it is retrieving them (:issue: `1852 `).
4147* Make file closing more robust, including when closing files while the
0 commit comments