Manual Backport PR #48427 on branch 1.5.x (BLD: Refactor Dockerfile t…

…o not install dev enviornment on base) (#48450)

Backport PR #48427: BLD: Refactor Dockerfile to not install dev enviornment on base

.github/workflows/code-checks.yml

@@ -153,6 +153,9 @@ jobs:
       - name: Build image
         run: docker build --pull --no-cache --tag pandas-dev-env .
 
+      - name: Show environment
+        run: docker run -w /home/pandas pandas-dev-env mamba run -n pandas-dev python -c "import pandas as pd; print(pd.show_versions())"
+
   requirements-dev-text-installable:
     name: Test install requirements-dev.txt
     runs-on: ubuntu-latest

Dockerfile

@@ -1,4 +1,4 @@
-FROM quay.io/condaforge/mambaforge:4.13.0-1
+FROM quay.io/condaforge/mambaforge
 
 # if you forked pandas, you can pass in your own GitHub username to use your fork
 # i.e. gh_username=myname
@@ -10,16 +10,12 @@ ENV DEBIAN_FRONTEND=noninteractive
 
 # Configure apt and install packages
 RUN apt-get update \
-    && apt-get -y install --no-install-recommends apt-utils dialog 2>&1 \
+    && apt-get -y install --no-install-recommends apt-utils git tzdata dialog 2>&1 \
     #
-    # Install tzdata and configure timezone (fix for tests which try to read from "/etc/localtime")
-    && apt-get -y install tzdata \
+    # Configure timezone (fix for tests which try to read from "/etc/localtime")
     && ln -fs /usr/share/zoneinfo/Etc/UTC /etc/localtime \
     && dpkg-reconfigure -f noninteractive tzdata \
     #
-    # Verify git, process tools, lsb-release (common in install instructions for CLIs) installed
-    && apt-get -y install git iproute2 procps iproute2 lsb-release \
-    #
     # cleanup
     && apt-get autoremove -y \
     && apt-get clean -y \
@@ -35,18 +31,12 @@ RUN mkdir "$pandas_home" \
     && git remote add upstream "https://github.com/pandas-dev/pandas.git" \
     && git pull upstream main
 
-# Because it is surprisingly difficult to activate a conda environment inside a DockerFile
-# (from personal experience and per https://github.com/ContinuumIO/docker-images/issues/89),
-# we just update the base/root one from the 'environment.yml' file instead of creating a new one.
-#
 # Set up environment
-RUN mamba env update -n base -f "$pandas_home/environment.yml"
+RUN mamba env create -f "$pandas_home/environment.yml"
 
 # Build C extensions and pandas
-SHELL ["/bin/bash", "-c"]
-RUN . /opt/conda/etc/profile.d/conda.sh \
-    && conda activate base \
-    && cd "$pandas_home" \
+SHELL ["mamba", "run", "--no-capture-output", "-n", "pandas-dev", "/bin/bash", "-c"]
+RUN cd "$pandas_home" \
     && export \
     && python setup.py build_ext -j 4 \
     && python -m pip install --no-build-isolation -e .

doc/source/development/contributing_environment.rst

@@ -38,6 +38,16 @@ Run Container::
     # Run a container and bind your local repo to the container
     docker run -it -w /home/pandas --rm -v path-to-local-pandas-repo:/home/pandas pandas-yourname-env
 
+Then a ``pandas-dev`` virtual environment will be available with all the development dependencies.
+
+.. code-block:: shell
+
+    root@... :/home/pandas# conda env list
+    # conda environments:
+    #
+    base                  *  /opt/conda
+    pandas-dev               /opt/conda/envs/pandas-dev
+
 .. note::
     If you bind your local repo for the first time, you have to build the C extensions afterwards.
     Run the following command inside the container::

doc/source/whatsnew/v1.5.0.rst

@@ -332,6 +332,7 @@ Other enhancements
 - Added ``copy`` keyword to :meth:`Series.set_axis` and :meth:`DataFrame.set_axis` to allow user to set axis on a new object without necessarily copying the underlying data (:issue:`47932`)
 - :meth:`DataFrame.set_index` now supports a ``copy`` keyword. If ``False``, the underlying data is not copied when a new :class:`DataFrame` is returned (:issue:`48043`)
 - The method :meth:`.ExtensionArray.factorize` accepts ``use_na_sentinel=False`` for determining how null values are to be treated (:issue:`46601`)
+- The ``Dockerfile`` now installs a dedicated ``pandas-dev`` virtual environment for pandas development instead of using the ``base`` environment (:issue:`48427`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_150.notable_bug_fixes: