1.5.2

Minor improvements:

• Add support and testing for Python 3.8. @rly (#247)
• Remove code duplication and make Type/Value Error exceptions more informative. @yarikoptic (#243)
• Streamline CI and add testing of min requirements. @rly (#258)

Bug fixes:

• Update hdmf-common-schema submodule to 1.1.2. @rly (#249, #252)
• Add support for np.array(DataIO) in py38. @rly (#248)
• Fix bug with latest version of coverage. @rly (#251)
• Stop running CI on latest and latest-tmp tags. @rly (#254)
• Remove lingering mentions of PyNWB. @rly (#257, #261)
• Fix and clean up documentation. @rly (#260)

1.5.1

Minor improvements:

• Allow passing HDF5 integer filter ID for dynamically loaded filters @d-sot (#215)

Bug fixes:

• Fix reference to hdmf-common-schema 1.1.0 (#231)

1.5.0

Minor improvements:

• Improve CI for HDMF to test whether changes in HDMF break PyNWB. #207 (@rly)
• Improve and clean up unit tests. #211, #214, #217 (@rly)
• Refactor code to remove six dependency and separate ObjectMapper into its own file. #213, #221 (@rly)
• Output exception message in ObjectMapper.construct. #220 (@t-b)
• Improve docstrings for VectorData, VectorIndex, and DynamicTableRegion. #226, #227 (@bendichter)
• Remove unused "datetime64" from supported dtype strings. #230 (@bendichter)
• Cache builders by h5py object id not name. #235 (@oruebel)
• Update copyright date and add legal to source distribution. #232 (@rly)
• Allow access to export_spec function from hdmf.spec package. #233 (@bendichter)
• Make calls to docval functions more efficient, resulting in a ~20% overall speedup. #238 (@rly)

Bug fixes:

• Fix wrong reference in ObjectMapper.get_carg_spec. #208 (@rly)
• Fix container source not being set for some references. #219 (@rly)

Python 2.7 is no longer supported.

1.4.0

Minor features:

• Write from all DataChunkIterators at once. #182 (@ajtritt)
• Allow enforcing unique ids on DynamicTable. #186 (@oruebel)
• Allow search of IDs in ElementIdentifiers. #187 (@oruebel)
• Update requirements. #176 (@rly)
• Enhance selection of rows in DynamicTable. #191 (@oruebel)
• Removed redundant shape check functions. #205 (@oruebel)

Bug fixes:

• Pass through DataIO parameters after formatting strings. #173 (@ajtritt)
• Fix bug in getting h5.Dataset shape. #175 (@ajtritt)
• Remove extra print statements from export_spec. #174 (@rly)
• Resolve references properly. #179 (@ajtritt)
• Check for correct types when indexing into DynamicTables after read. #184 (@ajtritt)
• Avoid reading first chunk of DataChunkIterator on init in certain conditions. #189 (@rly)
• Fix _version.py reporting version. #190 (@rly)
• Raise error if expected attribute does not exist on a Container. #196 (@rly)
• Handle extra or duplicate values passed to docval. #194 (@rly)
• Add shape property for Data. #202 (@oruebel)
• Fix support for scalar np.bool_. #203 (@rly)

1.3.3

Minor features:

• Improve documentation of data_utils. #149 (@oruebel)
• Improve CI with PyNWB. #148 (@rly)
• Improve documentation for cloning HDMF repo. #148 (@rly)
• Add ability to apply DataIO to Data after construction. #159 (@ajtritt)
• Add append and extend for Datasets in Data objects. #161 (@bendichter)
• Update copyright year in documentation. #164 (@t-b)
• Allow SZIP compression options. #167, #168 (@oruebel)
• Add export_spec utility function for use by extensions. #170 (@rly)

Bug fixes:

• Fix copy of DataIO and DataChunk. #155 (@oruebel)
• Fix default ObjectMapping bug for DynamicTableRegion columns. #171 (@oruebel)

1.3.2

hdmf 1.3.2

1.3.1

Bug fix:

• Add missing scipy requirement to setup.py #151 (@rly)

1.3.0

Major features:

• Add common data structures - Container, Data, DynamicTable, CSRMatrix, most of which are moved from PyNWB. #134, #150 (@ajtritt)

Minor features:

• Add functionality to slice a DataIO object using object[slice] which calls object.data[slice]. #141 (@oruebel)
• Add more reasonable defaults for BuildManager and NamespaceCatalog. #133 (@ajtritt)
• Replace Travis CI and Appveyor CI with Azure Pipelines. #137 (@rly)

Bug fixes:

• If a spec does not specify a dtype or uses the dtype 'numeric', infer the dtype from the data. This fixes an issue where a TimeSeries that used a DataChunkIterator as data would always write the data as floats instead of the original data type. #143, #146 (@rly)
• Fix infinite recursion when copying H5DataIO object. #141 (@oruebel)
• Fix typos and clean up repository. #138 (@rly)

1.2.0

New features:

• Enhance DataChunkIterator to support h5 datasets better, skip empty blocks, and work on arbitrary iteration axes #132 (@oruebel, @rly)
• Drop Python 2.7 support #126 (@rly)

Minor enhancements:

• Always use parent names in autogenerated child spec names #130 (@rly)
• Improve unit testing #131 (@rly)
• Improve coverage testing #128 (@rly)
• Implement better support for floating point data for Python 3.5 on Windows #127 (@rly)

1.1.2

Minor enhancements:

• Add LabelledDict class from PyNWB (@rly)
• Add ability to resolve link to links (@ajtritt)
• Support Scalar HDF5 dataspaces in H5SpecReader (@t-b)
• Determine dtype for empty numpy datasets (@oruebel)

Bug fixes:

• Fix shape enforcement in docval when argument does not have len (@rly)