You first need to install if you want to contribute to the code.
You may want to take a look into qcip_tools, the base library behind nachos.
- The code is written in Python 3, and follows the (in)famous PEP-8. You can check it by running
make lint
, which launch theflake
utility. - Codes and comments are written in english.
- The code is documented using docstrings and Sphinx. The docstrings must contains the basic description of the function, as well as a description of the paramters (with the
:type
instruction, please). - The code is tested. You can launch the test series by using
make test
. Every functionality should be provided with at least one unit test. Every script should be provided with at least one unit test. You may need test files to do so, but try to make them small (say, don't use d-aug-cc-pVDZ while STO-3G could do the job). - The package is documented. You can generate this documentation by using
make doc
. Non-basic stuffs should be explained in this documentation. Don't forget to cite some articles or website if needed.
Adapted from the (in)famous Git flow.
- Development is made in
dev
branch. - Functionalities are added through pull requests (PR) to the
dev
branch. Do not work indev
directly, but create a new branch (git checkout -b my_branch upstream/dev
). - Theses pull requests should be unitary, and include unit test(s) and documentation if needed. The test suite must succeed for the merge request to be accepted.
- The pull requests will be reviewed before acceptance.
- At some (random) points, a new version will appear, with a tag of the form
vXX
.
Note
Since nachos
now rely on pip-tools, the workflow is currently the following :
- Normal installation use
pip-sync && pip install -e .
(make init
) - To update the dependencies from upstream,
pip-sync
(make sync
). - To update the
requirements.txt
(and thus the actual version of the dependencies), a specific merge request is done, with the result ofpipenv lock
(followed bymake sync
on the dev's machine).