Note
A different approach to enable devcontainer
is: https://github.com/behrica/devcontainer-templates
Babashka script which adds a devcontainer
configuration to a
clojure project in current folder.
Basic usage is via
bb -m behrica.adddev.main
But as the script acts on local folder, it makes more sense to have it installed in some form in your computer,
using bbin
for example.
bbin install io.github.behrica/add-devcontainer
Running the script creates a local folder .devcontainer
with a devcontainer.json
including Clojure, babashka and lsp.
You can add the following parameters to add support for python or R into the devcontainer.
This updates as well deps.edn
.
parameter | description |
---|---|
:with-R | adds R and clojisr |
:with-python | adds python and clj-libpython and a nrepl start script which runs nrepl in a configured venv |
:with-noj | add noj to deps.edn |
When the python support is activated, the script creates as well a nrepl.sh
script,
which combines the setup of the venv via poetry
and the nrepl start via clj
in one commmand.
This makes sure, that the clojure process starts with the activated venv, which makes libpython-clj
automatically
find the correct python environment at initialisation.
Using the appropiate poetry
and clj
commands this could as well be done more manual.
Assuming you have
installed and working on your machine,
you can get from "zero" to a clojure + python and R enabled devcontainer running nrepl (with working python bindings via [libpython-clj](https://github.com/clj-python/libpython-clj and R bindings with clojisr) ) on port 12345 via:
neil new scratch foo
cd foo/
adddev :with-python :with-R
chmod +x nrepl.sh
devpod up .
devpod ssh . --command ./nrepl.sh
As devpod will forward port 12345 to your host, you can do cider-connect
to localhost:12345
with Emacs (or any other nrepl based IDE)
All changes of python packages listed in pyproject.toml
will be picked-up and sychronized with the used venv
when you stop/start nrepl.sh
The files of your localfolder a bind mount into the runnover docker. So you can edit them "localy" with Emacs, and the running Docker will see all changes immidiately. (You could as well edit them via Emacs/Tramp over the ssh connection, but this is not needed)
Using those the devpod
CLI usage is not needed and the devcontainer
setup happens automatically by VSCode.
VSCode has a feature of "Reopen in folder in devcontainer" which does all automatically.
In this case the nrepl.sh
would be started from
inside VSCode / Codespaces
If the code is on Github, Github allows to use Codespaces for an in-Browser VSCode IDE.
Assuming that above listed tools are installed.
It adds as well noj
neil new scratch datascience-scratch
cd datascience-scratch
adddev :with-python :with-R :with-noj
# Recommendet when using Emacs.
# In VSCode `opening the folder` activates devcontainer
chmod +x nrepl.sh
devpod up .
devpod ssh . --command ./nrepl.sh