In all cases, please submit code to the main repository via a pull request. Upon approval, please merge via "Squash and Merge" on GitHub to maintain a clean commit history.
Please submit pull requests as early as possible (you can still push to the branch once submitted) to allow discussion.
cellfinder uses Black to ensure a consistent code style. Please run
black ./ -l 79 --target-version py37 before making any commits. To prevent any errors, it is easier to add a formatting check as a pre-commit hook. E.g. on linux by adding this to your
black ./ -l 79 --target-version py37 --check || exit 1
cellfinder uses pytest for testing. Please try to ensure that all functions are tested in
tests/tests/test_unit and all workflows/command-line tools are tested in
Some tests may take a long time (e.g. those requiring tensorflow if you don't have a GPU). These tests should be marked with
import firstname.lastname@example.org test_something_slow():slow_result = run_slow_processes()assert slow_result == expected_slow_thing
During development, these "slow" tests can be skipped by running
pytest -m "not slow".
All commits & pull requests will be build by Travis. To ensure there are no issues, please check that it builds:
pip install . and run all of the tests:
pytest before committing changes.
Travis will automatically release any tagged commit on the master branch. Hence to release a new version of cellfinder, use either GitHub, or the git command-line tool to tag the relevant commit and push to master.
The code in the cellfinder repository should be primarily for cell detection. Other changes should be added to the relevant repository:
Brain imaging data IO: brainio
Visualisation & general utilities: neuro
General imaging processing (and other) tools: imlib.
All file paths should be defined in
cellfinder.tools.prep.Paths. Any intermediate file paths, (i.e. those which are not of interest to the typical end user) should be prefixed with
tmp__. These should then be cleaned up as soon as possible after generation.
Cellfinder has recently (2019-08-02) dropped support for Python 3.5. Following this, a number of new python features will be adopted throughout.
.format() or using the old
In all cases, please aim to replace old calls to
.format() with pathlib object methods and f-strings.