Open mbargull opened 6 years ago
Additionally, the requirements which are used for the tests or to build the docs needn't be install_requires
. Test requirements can use the tests_require
parameter to setup(...)
.
To be honest, you seem to know more about it than I do. We have been aware that there are issues with the setup on some systems. If you just clone the repository, put the scripts folder in your PATH and the basepath in your PYTHONPATH, and install the Python dependencies with pip and somehow install the required mappers and variant callers, you should be OK. Sorry for not being as helpful as I could have been.
Hi @ulfschaefer, sorry for not providing more context upfront. I'm coming from over at https://github.com/bioconda/bioconda-recipes, where someone (@pvanheus) is looking into making PHEnix available to Conda / Bioconda users (and thus make your software more easily accessible, i.e., let a single conda install phenix
install everything -- including those non-Python mapper/var. caller dependencies -- for the user).
My comments are meant as suggestions/improvements to make PHEnix behave more like "conventional/standard" Python package which would ease user installation as well as packaging in Bioconda & co.
Happy to help if you need any assistance or more information!
Sure. I know @pvanheus and I have spoken to him separately. Thank you to you both for looking into doing this.
Hi there, I found a couple issues in your
setup.py
file:from setuptools import setup
instead offrom distutils.core import setup
since you use theinstall_requires
keyword, which is asetuptools
addition.package
parameter should be namedpackages
.pip.req.parse_requirements
. A simpleHowever I'd advice against using it at all, see: https://packaging.python.org/discussions/install-requires-vs-requirements/.
install_requires
should list compatible versions, e.g.,package>=1.0
orpackage>=1.0,<2.0
, which are usually not exact==
requirements as inrequirements.txt