CHANGES
=======

v0.3.2
------

* fix: sampler length works as intended and check for empty datastream
* feature: implement method take for datastream
* doc: adds pypi badge
* doc: missing space in error message

v0.3.1
------

* Use textwrap to shorten debug description
* Cache dataset in-memory
* Better message by using same traceback
* Better exceptions for starmap with module logging
* Wrap star function for better exception
* Try-except for better exceptions
* Conform group\_split to the name change Dataset.functions -> get\_item
* Add Dataset.group\_split

v0.3.0
------

* Add some examples in docstrings
* Split map into map/starmap and compose functions without starcompose

v0.2.8
------

* Zip dataframe
* Fix space when publishing
* Fixed missed self reference in RepeatSampler
* Fix linting errors/warnings and remove unused imports
* Make from\_subscritable work with lazy subscriptables

v0.2.7
------

* Link to function in documentation
* Show results for example
* More anvanced examples in readme
* Create split file directory and write with indent
* Prefer nunique over len after unique
* Change exception type in split
* Rename typevar to more standard
* Remove subset requirement from split
* Add seed option to Dataset.split
* Change frozen requirement to subset
* Allow frozen splits
* Replace json tool

v0.2.6
------

* Make dataset combine/zip/concat use lazy dataframes
* Fix split docstring formatting

v0.2.5
------

* Many new degenerate tests
* Test validation of proportions. Update split docstring with how saved splits work
* Mention limitations of Dataset.from\_subscriptable in docstring
* Fix stratification bug
* Added Dataset.split. Fix incorrect version. Move tool functions to module
* Minor documentation update

v0.2.4
------

* Fix package in setup

v0.2.3
------

* Run tests
* First iteration of CI
* Fix Dockerfile
* Set dependencies

v0.2.2
------

* Let subset take function
* Add package to setup
* Update project summary

v0.2.1
------

* Fix README parse error on pypi
* Add license from ml-workflow
* Convert readme to rst

v0.2.0
------

* Minor docstring fix
* Read the docs config
* Move out samplers from datastream module
* Rename source now that it is always a dataframe
* Subset only takes mask
* Added generics. Attempt to fix version in docs. Documentation improvements
* Documentation of methods and generating documentation using sphinx
* Minimum pydantic implementation passes all tests
* Datastream and two samplers using pydantic. Test for multi\_sample
* Dataset using pydantic
* Function showcase
* Add tests for subscript and subset
* More usage examples in readme
* Fix representation of short dataset

v0.1.0
------

* Allow execution of publish.sh
* First commit. Split from ml-workflow
