This project will be the successor to the alpaka-job-matrix-library. We have decided that several fundamental changes are needed in the alpaka-job-matrix-library, including renaming. Therefore, rewriting large parts of the code base would be necessary. Based on this, we have decided that a new project means less work. Until this project is officially released, please use the alpaka-job-matrix-library. The project is already public due to the use of CI features.
A library to provide a job generator for CI's for alpaka based projects.
The main component of the bashi
library is combinatorics. Due to the wide spread of the field, there are different words for the same things. Therefore, bashi
has introduced a naming guideline that is used for function and parameter names and documentation. Please read the naming guidelines.
An example of the use of the bashi
library can be found in example/example.py. It shows how to use the library to create a combination-list
from a parameter-value-matrix
. The example also uses a custom filter. For more details, please read the module documentation of example/example.py.
It is strongly recommended to use a Python environment for developing the code, such as virtualenv
or a conda
environment. The following code uses a virtualenv
.
- Create the environment:
virtualenv -p python3 env
- Activate the environment:
source env/bin/activate
- Install the library:
pip install --editable .
- Test the installation with the example:
python3 example/example.py
- You can run the unit tests by going to the
test
directory and runningpython -m unittest discover -s tests
If the example works correctly, a job.yml
will be created in the current directory. You can also run python3 example/example.py --help
to see additional options.
Now the library is available in the environment. Therefore you can easily use the library in your own projects via import bashi
.
This section contains some hints for developing new functions. The hints are mainly for people who have no experience with setuptools
and building pip
packages.
- The
pip install --editable .
command installs the source code files as a symbolic link. This means that changes in the source code of the Python files in thesrc/bashi
folder are directly available without any additional installation step (only a restart of the Python process/interpreter is required). - The software requirements are defined in
pyproject.toml
and not in an additionalrequirements.txt
. - It is necessary to increase the version number in
version.txt
before a new feature can be merged in the master branch. Otherwise the upload to pypy.org will fail because existing versions cannot be changed.
The source code is formatted using the black formatter and the default style guide. You must install it and run black /path/to/file
to format a file. A CI job checks that all files are formatted correctly. If the job fails, a PR cannot be merged.
The project supports code coverage with coverage.py. To create a coverage report locally, you must first install the package via pip install coverage
. Then run coverage run
in the project folder to calculate the coverage and coverage report
to display the result.