I am new to python. I guess there are libraries called numpy, anaconda etc. What the best init.el setup for this. (I want to work on ML or Data Science)
Following is my configuration for Python; I am using Emacs 29.1 --with-tree-sitter:
(use-package treesit :config (setq treesit-language-source-alist '((python . ("https://github.com/tree-sitter/tree-sitter-python.git")))) (add-to-list 'auto-mode-alist `("\\.py[iw]?\\'\\|python[0-9.]*" . python-ts-mode))) (use-package eglot :hook ((python-mode . eglot-ensure) ; pip install 'python-language-server[all]' (python-ts-mode . eglot-ensure)))
I have something like this
;; ----- language server (require 'eglot) (add-hook 'python-mode-hook #'eglot-ensure) (add-to-list 'eglot-server-programs `(python-mode . ,(eglot-alternatives '("pylsp" "jedi-language-server" ("pyright-langserver" "--stdio"))))) ;; ----- treesitter (use-package treesit-auto :ensure t) ;; treesit grammar should be installed here: "~/.emacs.d/tree-sitter/" ;; ----- select interpreter (setopt python-shell-interpreter ("~/miniconda3/bin/python")) (setopt python-shell-interpreter-args "-i") ;; ----- conda (use-package conda :ensure t :custom (conda-anaconda-home "~/miniconda3") (conda-env-home-directory "~/miniconda3") (conda-env-subdirectory "envs") :config (unless (getenv "CONDA_DEFAULT_ENV") (conda-env-activate "base")))) ;; ----- emacs-jupyter (use-package jupyter :commands (jupyter-run-server-repl jupyter-run-repl jupyter-server-list-kernels)) (org-babel-jupyter-override-src-block "python") ;; so python becomes jupyter-python in org-babel ;; ----- other settings (setopt python-indent-guess-indent-offset-verbose nil)
I had to uninstall the conda package because it was constantly disconnecting and reconnecting to the kernel and every time it’d do that Emacs would freeze for a couple of seconds. Probably happened about 10 times a minute before I finally deleted the package.
I use Emacs 29.1 or Emacs 30, because:
- they have built in LSP support by
Eglot
. - they have built in tree-sitter support by
python-ts-mode
And I use pyright as the LSP backend.
After install
pyright
, just start Emacs, open python file andM-x eglot
will fire up the lsp support.For the python virtual environment, I use
envr(direnv)
on macOS and Linux, usepyvenv
on Windows platform.You can see my personal configuration here:
- they have built in LSP support by
It sounds like you’re confused about Python and Emacs as well as their relationship. You’ll need to find a Python data science tutorial and read all the way through it.
This should help with the emacs side: https://github.com/wyuenho/emacs-pet
I like this article: https://robbmann.io/posts/006_emacs_2_python/
It covers Python configuration using mostly built-it modules and features
still using elpy, it werks
jedi-language-server + eglot only.
maybe i’ll extend it to support running unit tests.use https://github.com/emacs-jupyter/jupyter for jupyter integration.
use https://github.com/jkitchin/ox-ipynb to export as jupyter notebooks to share with collegues.
All the other packages only makes sense if you are working alone. If you have to share your work with others, and aimed for DS/ML work thats what I would recommend.