Pynwb

PyNWB 1.1.0 does not work with HDMF>=1.3. This release will work with HDMF>=1.3.2.

Minor improvements:
- Support newly added channel-specific conversion factor for ElectricalSeries 1072 (bendichter)
- Move generic types out of PyNWB into hdmf-common. 1061 (ajtritt)
- Update documentation to reflect the above changes. 1078 (rly)
- Add new case to the iterative write tutorial. 1029 (oruebel)
- Improve CI. 1079 (rly)
- Pin the current latest version of HDMF to requirements for setup.py. 1083 (rly)

New features:
- Add object ID to all neurodata types 991 (ajtritt, rly)
- Add NWBFile shallow copy method 994 (ajtritt, rly)
- Drop official Python 2.7 support 1028 (rly)
- Add scratch space 1027 1038 (ajtritt, rly)
- Support multiple experimenters 988 1035 (ajtritt, rly)
- Support multiple related publications 1047 (rly)
- Update schema to 2.1.0 (see release notes in https://nwb-schema.readthedocs.io/en/latest/format_release_notes.html) (rly, bendichter, ajtritt, oruebel, t-b)

Minor enhancements:
- Add iterative write check for TimeSeries timestamps 1012 (bendichter, oruebel)
- Add functions to convert between pixel mask and image mask for ophys data 766 (mamelara)
- Add cortical surface extension example 1040 (bendichter)
- Match API with schema defaults 1033 (rly)
- Core schema is now a git submodule 1045 (ajtritt)
- Implement better support for floating point data for Python 3.5 on Windows 1043 (rly)
- Enhance iterative write tutorial 1029 (oruebel)
- Allow empty data in DynamicTable with non-empty VectorIndex 887 (ajtritt)
- Allow OpticalSeries constructor argument 'field_of_view' to be H5Dataset 1063 (bendichter)
- Clarify documentation for deprecated ImageSeries constructor arg 'bits_per_pixel' 1065 (rly)
- Adjust code to explicitly map properties after changes made in HDMF 1.2 1048 1069 (rly)
- Improvements to CI, documentation, and GitHub repo structure 1055 (rly)

New/modified functionality:
- Add MPI functionality to NWBHDF5IO (bendichter)
- Add option to exclude columns from DynamicTable.to_dataframe() (NileGraddis)
- Remove ability to add DecompositionSeries to LFP (bendichter)
- Remove num_samples from TimeSeries (NileGraddis)
- Automatically detect ragged arrays in from_dataframe (bendichter)
- Cache the spec by default on write (rly)
- Improve printing of NWB objects (rly)
- Change ProcessingModule.add_data_interface() to .add(), ProcessingModule.get_data_interface() to .get(), NWBFile.modules to NWBFile.processing (bendichter)
- Remove unused SpecFile type (oruebel)
- Add ability to validate files against the cached spec (t-b)
- Make CurrentClampSeries/VoltageClampSeries parameters optional (t-b)
- Update documentation (t-b, rly)
- Update copyright/license
- Improve tests and CI
- Update requirements
- See also HDMF changes https://github.com/hdmf-dev/hdmf/releases/tag/1.0.4

Bug fixes:
- Fix dynamictableregion iteration failure after roundtrip (NileGraddis)
- Fix from_dataframe for children of DynamicTable (bendichter)
- Fix for modular (cross-file) storage of timeseries timestamps (NileGraddis)
- Fix bug on loading lists of strings from hdmf 1.0.4 (rly)
- Fix IO for intervals (bendichter)
- Fix round trip for Subject.date_of_birth (bendichter)

Schema changes:
- DecompositionSeries "source_timeseries" link is no longer required (bendichter)
- Reorder keys (rly)
- Remove NWBFile "specifications" group (oruebel)
- CorrectedImageStack and ImagingRetinotopy inherits from NWBDataInterface instead of NWBContainer (rly)
- Fix typo in unit of resistance_comp_prediction/correction (t-b)
- Add option for third dimension for Units "waveforms" dataset to represent different electrodes (bendichter)
- Update NWBFile.nwb_version to 2.0.2