Closed prjemian closed 3 years ago
NOTE: the CI was restarted multiple times to ensure this report was not a temporary problem on the GitHub actions server.
Building a Dockerfile to test this locally in a container:
FROM continuumio/miniconda3
# docker build -t test_py36 ./
# docker run -it --rm test_py36
USER root
CMD /bin/bash
WORKDIR /opt
# tip from https://pythonspeed.com/articles/activate-conda-dockerfile/
# enables commands: conda init bash & conda activate
SHELL ["/bin/bash", "--login", "-c"]
RUN conda init bash
RUN conda update conda
RUN git clone https://github.com/bluesky/hklpy
WORKDIR /opt/hklpy
RUN sed -i s:'python >=3.6':'python =3.6':g environment.yml
RUN conda env create -n py36 -f environment.yml
RUN echo "conda activate py36" >> ~/.bashrc
RUN conda list
RUN conda env list
# RUN pytest -vvv .
Here's the full output from pytest -vvv .
:
Digging through the [xarray]() repository, note that python 3.6 was dropped in release 0.17.0.
The last xarray release before this was 0.16.2. Since hklpy does not specify a requirement for xarray, one of our requirements is missing this restriction for Py3.6.
I'm happy to drop support for Python 3.6 (ahead of the numpy drop of it at the end of this year). @mrakitin : Agree?
Works for me!
We need to make a coherent change in https://github.com/conda-forge/hklpy-feedstock too. Right now Python 3.6 is still supported there.
Building the documentation requires these additional packages: make flake8 black pandoc
Discovered by building this Dockerfile: (docker build -t test_py ./
)
FROM continuumio/miniconda3
# docker build -t test_py ./
# docker run -it --rm test_py
USER root
CMD /bin/bash
WORKDIR /opt
# tip from https://pythonspeed.com/articles/activate-conda-dockerfile/
# enables commands: conda init bash & conda activate
SHELL ["/bin/bash", "--login", "-c"]
RUN conda init bash
RUN conda update conda
RUN git clone https://github.com/bluesky/hklpy
WORKDIR /opt/hklpy
RUN sed -i s:'python >=3.8':'python =3.8':g environment.yml
RUN conda env create -n py38 -f environment.yml
RUN echo "conda activate py38" >> ~/.bashrc
RUN conda list
RUN conda env list
RUN pip install -e .
RUN conda install \
jupyter nbconvert sphinx sphinxcontrib-napoleon \
make \
flake8 black pandoc \
-c defaults -c conda-forge
RUN pip install sphinx_rtd_theme
RUN make
With recent merges, the CI is now failing (where it did not fail in the PR's branch).
One fail, in the Py3.8 build, is due to missing
pandoc
package. This fail blocks building of the documentation.Another failure, in the Py3.6 build, is due to a problem with the
xarray
package.