Update docs

INSTALL.rst

@@ -180,6 +180,8 @@ libpywrapper two environment variables need to be present:
 building without linking against libpython we must set ``NMODL_PYLIB``
 before running cmake!
 
+.. _testing-installed-module:
+
 Testing the Installed Module
 ----------------------------
 

packaging/README.rst

@@ -4,8 +4,8 @@ Building Python Wheels
 Note: This is only slightly adapted from NEURONs
 `scripts <https://github.com/neuronsimulator/nrn/tree/master/packaging/python>`__.
 
-NMODL wheels are built in a manylinux2010 image. Since the generic
-docker image is very basic (CentOS 6) a new image, which brings updated
+NMODL wheels are built in a manylinux2014 image. Since the generic
+docker image is very basic (CentOS 7), a new image, which brings updated
 cmake3 (3.12), flex and bison was prepared and made available at
 https://hub.docker.com/r/bluebrain/nmodl (tag: wheel).
 
@@ -18,64 +18,66 @@ Linux `here <https://docs.docker.com/engine/install/ubuntu/>`__ and on
 OS X `here <https://docs.docker.com/docker-for-mac/install/>`__. On
 Ubuntu system we typically do:
 
-::
+.. code::sh
 
    sudo apt install docker.io
    sudo groupadd docker
    sudo usermod -aG docker $USER
 
 Logout and log back in to have docker service properly configured.
 
-Pull and start the docker image
--------------------------------
+Launch the wheel building
+-------------------------
 
-We mount local neuron repository inside docker as a volume to preserve
-any code changed. We can use -v option to mount the local folder as:
+For building the wheel we use the ``cibuilwwheel`` utility, which can be installed using:
 
-::
+.. code::sh
 
-   docker run -v /home/user/nmodl:/root/nmodl -it bluebrain/nmodl:wheel bash
+   pip install cibuildwheel
 
-where ``/home/user/nmodl`` is the NMODL repository on the host machine.
-We mount this directory inside docker at location ``/root/nmodl`` inside
-the container.
+Then to build a wheel for the current platform, run:
 
-Note that for OS X there is no docker image but on a system where all
-dependencies exist, you have to perform next building step.
+.. code::sh
 
-Launch the wheel building
--------------------------
+   cibuildwheel
 
-Once we are inside docker container, we can start building wheels. There
-is a build script that loops over the pythons ``>=3.5`` in
-``/opt/python``, build and audit the generated wheels. Results are
-placed in this wheelhouse directory.
+If you have Docker installed, you can also build the Linux wheels using:
 
-::
+.. code::sh
 
-   cd /root/nmodl
-   bash packaging/python/build_wheels.bash linux
+   cibuildwheel --platform linux
 
-For OSX on a system with the all dependencies you have to clone the
-NMODL repository and do:
+Note that, if you happen to have Podman installed instead of Docker, you can
+set ``CIBW_CONTAINER_ENGINE=podman`` to use Podman instead of Docker for this
+task.
 
-::
+Furthermore, in order to build wheels on MacOS, you need to install an official
+CPython installer from `python.org <https://www.python.org>`__.
 
-   cd nrn
-   bash packaging/python/build_wheels.bash osx
+For a complete list of all available customization options for
+``cibuildwheel``, please consult the
+`documentation <https://cibuildwheel.readthedocs.io/en/stable/options/>`__.
 
-Updating neuron_wheel docker image
-----------------------------------
+Testing the wheel
+-----------------
+
+On MacOS, the testing of the wheel is launched automatically when running
+``cibuildwheel``. On Linux, you will need to test the wheel manually, please
+see :ref:`testing-installed-module` for the instructions.
+
+
+Updating the NMODL Docker image
+-------------------------------
 
-If you have changed Dockerfile, you can build the new image as:
+If you have changed the Dockerfile, you can build the new image as:
 
-::
+.. code::sh
 
    docker build -t bluebrain/nmodl:tag .
 
-and then push image to hub.docker.com as:
+and then push the image to hub.docker.com as:
 
-::
+.. code::sh
 
    docker login --username=<username>
-   docker push bluebrain/nmodl:tag 
+   docker push bluebrain/nmodl:tag