CHANGELOG.md
<a id='changelog-0.3.5'></a>
# 0.3.5 (2024-03-19)
## 🐛 Bug Fixes
- Add codespell config and workflow, and use it to typos found throughout the codebase. [PR #131](https://github.com/datalad/datalad-neuroimaging/pull/131) (by [@yarikoptic](https://github.com/yarikoptic))
## 📝 Documentation
- Fixup intersphinx_mapping for python. [PR #132](https://github.com/datalad/datalad-neuroimaging/pull/132) (by [@yarikoptic](https://github.com/yarikoptic))
## 🏠 Internal
- Set up releasing workflows. Fixes [#127](https://github.com/datalad/datalad-neuroimaging/issues/127) via [PR #130](https://github.com/datalad/datalad-neuroimaging/pull/130) (by [@jwodder](https://github.com/jwodder))
## 0.3.4 (Oct 24, 2023) -- Make BIDS great again
- BIDS metadata extractor(s): do not require `participants.tsv` file to be present
(it is RECOMMENDED not REQUIRED in BIDS)
- `cfg_bids` procedure: do not annex any `README*`, `LICENSE`, or
`.bids-validator-config.json` files
- documentation: fix builds on RTD
## 0.3.3 (Oct 28, 2022) -- Oktoberfest
- raise dependency on datalad-deprecated which was prematurely released
## 0.3.2 (Oct 25, 2022) -- ... but grandparents always are
- increase dependency on datalad to 0.16.7
- depend on datalad-metalad and datalad-deprecated for both styles of metadata
extractors
- update the metadata extractors to work with current numpy versions
- BIDS extractor compatible with pybids>=0.15.1 and BIDS v1.6.0
- fix nifti1 metadata extractor not dealing properly with byte strings
- various internal changes and fixes around testing and CI setup
## 0.3.1 (Jun 3, 2020) -- Uncles aren't always nice either
- be compatible with pydicom 2.0.0
## 0.3.0 (Feb 26, 2020) -- ... and parents become grumpier
- DataLad 0.12 series is now minimal supported
## 0.2.4 (Feb 05, 2020) -- Kids are growing
Minor bugfix release to account for changes in pybids and datalad
- pybids:
- demand pybids >=0.9.2
- account for new field "extension" provided by BIDS now
- use get_dataset_description if available
- pandas:
- use .iloc instead of deprecated .ix
- datalad
- use -d^ instead of deprecated (in 0.12) -S in an example script
## 0.2.3 (May 24, 2019) -- Old is not bad
Minor quick bugfix release to demand pybids < 0.9 since we are not yet fully
ready for its full glory
## 0.2.2 (May 20, 2019) -- It was like that way before!
Minor bugfix release
- revert back to old ways of installing package data so test data and procedures
get properly installed
## 0.2.1 (May 17, 2019) -- Why wasn't it that way before?
- include a procedure 'cfg_bids' to configure BIDS datasets
- fix several issues with troublesome dependency declarations
## 0.2.0 (Feb 09, 2019) -- Am I compatible with you honey?
- Make compatible with (and demand) pybids 0.7.{0,1}. 0.7.0 introduced
change of terms: modality -> suffix, and type -> datatype, which would now
require to either reaggregate all previous metadata or somehow fixup
in-place existing metadata files. And for 0.7.1 workaround was added to
not return `suffix` at least when participants.tsv was queried.
- Make compatible with pydicom 1.0 in treatment of MultiValue
- Refactorings:
- tests:
- use `export_archive` instead of plain `tarfile.open`
- make compatible with recent (0.11.x) DataLad which now extract annex
keys etc
## 0.1.5 (Sep 28, 2018) -- BIDS robustness
- Assorted improvements of the BIDS metadata extractor performance on datasets
in the wild.
## 0.1.4 (Aug 02, 2018) -- PyBIDS
- Fixed compatibility with pybids 0.6.4 and now demand it as the minimal
PyBIDS version
## 0.1 (Apr 28, 2018) -- The Release
### Major refactoring and deprecations
- This is the first separate release of DataLad's neuroimaging functionality as an
extension module.
- Metadata
- BIDS metadata now uniformly refers to subjects and participants using the
metadata key 'subject'
### Enhancements and new features
- Extractors now report progress (with DataLad 0.10+)
- BIDS participant metadata is now read via pybids
### Fixes
- Fix issue with unicode characters in BIDS metadata
- DICOM metadata now also contains the 'PatientName' field that was previously
excluded due to a too restrictive data type filter