The ultimate python development environment

Last update: 20 February 2023

In this post, we're going to discuss how to setup an optimal development environment for python. We're going to look at the some important tools that you may want to include in your development experience and some tips to get the best coding experience possible.

Let's get started

Install python

You can install python in many different ways. The most common way is to use the package manager that comes with your OS.

In Arch Linux:

$ pacman -Syu python

That will install the latest version available in your OS.

In case you want to have a different version of python, you can use pyenv to switch between different versions easily.

Bootstrap new python project

The directory layout for your project may differ depending on what you're building.

Here are some examples:

  • For a web application
├── app/ # your code goes here
├── Containerfile # Used to containerize your web app
├── Justfile # commands runner
├── pyproject.toml # configuration
├── # used to run your web app
└── tests/ # your tests go here
  • For a library
├── Justfile # commands runner
├── pyproject.toml # configuration
├── src/ # your code goes here
└── tests/ # your tests go here
  • For a cli
├── # used to invoke your cli
├── Justfile # commands runner
├── pyproject.toml # configuration
├── src/ # your code goes here
└── tests/ # your tests go here


Justfile is where you specify the commands to run. It's a modern replacement of Makefile.

You need to install just to be able to use it.

Here is an example:

# Justfile
    @just --list

    #!/usr/bin/env bash
    source .venv/bin/activate &&
    pytest tests/

By running

$ just test

Will run the pytest command after sourcing the virtual environment.


This where many tools configuration will be. We're gonna see how it looks like as we go through the next steps.

ℹ️ Note
You may see projects with file. This has been deprecated in favor of pyproject.toml file.
Tips 💡
You can create templates for your projects using cookiecutter so you don't copy manually the project structure each time you want to start a new one.

Package/Dependency manager

Python does not come with an official package manager, so many tools have been built to fill that gap.

Initially, people would write the list the dependencies in a text file and install them using pip.

I guess there is still many people using this approach but it has some issues.

The modern way to handle python dependencies is to use a package/dependency manager. Here are some popular ones:

For the rest of the post, we're gonna focus on Poetry.

To install Poetry

$ curl -sSL | python3 -

Once it's installed, we're gonna use this configuration

$ mkidr ~/.config/pypoetry
$ cat <<EOF > ~/.config/pypoetry/config.toml
create = true
in-project = true

This tells Poetry to install the dependencies in the same directory as our code under .venv folder. Otherwise Poetry install them under ~/.cache/pypoetry directory.

To bootstrap a new project using Poetry you can run:

$ poetry new -n --name=simplewebap simplewebap

That will create a new directory simplewebap and inside of it you'll find the skeleton for a new python project

├── pyproject.toml
├── simplewebap
│  └──
└── tests

Then we can add our first dependency

$ cd simplewebap
$ poetry add flask

You'll notice that Poetry has updated `pyproject.toml

$ cat pyproject.toml
python = "^3.10"
flask = "^2.2.2"

Check the Poetry documentation to get familiar with it

Using NeoVim as the main editor

neovim is one of the best editors out there if not the Best. With some configuration, you can really boost your productivity to the next level.

ℹ️ Note
I'm gonna use vim-plug to manage neovim plugins.

When you write your code, you need to have two main components integrated in your neovim:

1. Code linter and formatter

We're gonna use black and ruff for code formatting and linting respectively.

Let's install them first

$ pip install --user black ruff
Be sure to have $HOME/.local/bin in your $PATH to be able to access them

We then need to define the configuration for those tools in pyproject.toml

Here is an example that you can use

line-length = 100
exclude = '''
  | \.mypy_cache
  | \.pytest_cache
  | \.tox
  | \.venv
  | __pycache__
  | build
  | dist

line-length = 100
target-version = "py310"

exclude = [

docstring-quotes = "double"

To integrate ruff and black into neovim, we're gonna use the null-ls plugin.

Let's install the plugin first

  Plug 'jose-elias-alvarez/null-ls.nvim'

Once it's installed, we can use this configuration

lua << EOF
    local null_ls = require("null-ls")
    local augroup = vim.api.nvim_create_augroup("LspFormatting", {})

        sources = {
            -- formatting

            -- diagnostics

         on_attach = function(client, bufnr)
                if client.supports_method("textDocument/formatting") then
                    vim.api.nvim_clear_autocmds({ group = augroup, buffer = bufnr })
                    vim.api.nvim_create_autocmd("BufWritePre", {
                        group = augroup,
                        buffer = bufnr,
                        callback = function()
                            vim.lsp.buf.format({ bufnr = bufnr })

So every time we save a python file, black and ruff will run and format the code automatically.

2. LSP integration.

There are many LSP servers solution for python to use. I choose pyright, but you're free to use anything else that works for you.

Let's install pyright first

$ pip install --user pyright

You need first to install those plugins that would enhance your coding experience

  Plug 'neovim/nvim-lspconfig'
  Plug 'hrsh7th/cmp-nvim-lsp'
  Plug 'hrsh7th/cmp-buffer'
  Plug 'hrsh7th/cmp-path'
  Plug 'hrsh7th/cmp-cmdline'
  Plug 'hrsh7th/nvim-cmp'
  Plug 'hrsh7th/cmp-vsnip'
  Plug 'hrsh7th/vim-vsnip'

And then, you can have this configuration for your lsp

lua <<EOF

    virtual_text = false,
    signs = true,
    underline = false,
    update_in_insert = false,
    float = {border = "rounded"},
    severity_sort = false,

  vim.o.updatetime = 250
  vim.cmd [[autocmd CursorHold,CursorHoldI * lua vim.diagnostic.open_float(nil, {focus=false})]]

  local has_words_before = function()
    local line, col = unpack(vim.api.nvim_win_get_cursor(0))
    return col ~= 0 and vim.api.nvim_buf_get_lines(0, line - 1, line, true)[1]:sub(col, col):match("%s") == nil

  local feedkey = function(key, mode)
    vim.api.nvim_feedkeys(vim.api.nvim_replace_termcodes(key, true, true, true), mode, true)

  local on_attach = function(client, bufnr)
    local bufopts = { noremap=true, silent=true, buffer=bufnr }
    vim.keymap.set('n', '<leader>d', vim.lsp.buf.definition, bufopts)

  -- Pyright
  local configs = require('lspconfig/configs')
  local util = require('lspconfig/util')

  local path = util.path
  local function get_python_path(workspace)
    -- Use activated virtualenv.
    if vim.env.VIRTUAL_ENV then
      return path.join(vim.env.VIRTUAL_ENV, 'bin', 'python')

    -- Find and use virtualenv in workspace directory.
    for _, pattern in ipairs({'*', '.*'}) do
      local match = vim.fn.glob(path.join(workspace, pattern, 'pyvenv.cfg'))
      if match ~= '' then
        return path.join(path.dirname(match), 'bin', 'python')

    -- Fallback to system Python.
    return exepath('python3') or exepath('python') or 'python'

  local cmp = require'cmp'

    snippet = {
      expand = function(args)
    window = {
      completion = cmp.config.window.bordered(),
      documentation = cmp.config.window.bordered(),
    mapping = cmp.mapping.preset.insert({
      ['<C-k>'] = cmp.mapping.scroll_docs(-4),
      ['<C-j>'] = cmp.mapping.scroll_docs(4),
      ['<CR>'] = cmp.mapping.confirm({ select = true }),
      ["<Tab>"] = cmp.mapping(function(fallback)
      if cmp.visible() then
        elseif vim.fn["vsnip#available"](1) == 1 then
          feedkey("<Plug>(vsnip-expand-or-jump)", "")
        elseif has_words_before() then
      end, { "i", "s" }),
      ["<S-Tab>"] = cmp.mapping(function()
        if cmp.visible() then
        elseif vim.fn["vsnip#jumpable"](-1) == 1 then
          feedkey("<Plug>(vsnip-jump-prev)", "")
      end, { "i", "s" }),
    sources = cmp.config.sources({
      { name = 'nvim_lsp' },
      { name = "vsnip" },
      { name = 'nvim_lsp_signature_help' },
        name = 'spell',
        option = {
            keep_all_entries = false,
            enable_in_context = function()
                return true
    }, {
      { name = 'buffer' },

    before_init = function(_, config)
        config.settings.python.pythonPath = get_python_path(config.root_dir)
    on_attach = on_attach,
    capabilities = capabilities,


With this configuration, you'll be able to have:

  • code autocompletion using Tab
  • going to definition using <leader>+d
  • Navigate through the documentation if it's available with ctrl+j and ctrl+k

Check out neovim-lsp for more options.

Here are some screenshots:

Tips 💡
You may want to have telescope plugin as well for an easy file browsing.
You can check my personal configuration here for a complete setup of my neovim.

Debug python code

For many years, I used to use print to debug my code, what an awful and inefficient way to debug the code !

There is a debugger for python called pdb. This will make you debug your code as a pro. I personally use a pbpp which is an advanced version of pdb.

Let's install it

$ pip install --user pdbpp

To have a better experience with pdbpp, you can add this configuration

$ cat <<EOF > ~/
import pdb

class Config(pdb.DefaultConfig):
    sticky_by_default = True
    use_pygments = True
    current_line_color = 50
    editor = "nvim"

To use pdbpp, you put breakpoints in places that you want to debug, then you run your code. The program will stop when it encounters breakpoints. Then you'll be presented an interactive interface where you can check the state of your variables, calls etc.


Any code that is not tested is a broken code. It's okay not have 100% code coverage but a descent chunk of your code should be tested.

The most commmon test framework for python is pytest

You can enhance your testing by integration those pytest plugins:

pytest-sugar: Progress bar and improved test results.

pytest-icdiff: Better output for asserts.

pytest-clarity: Improved output with colors.

pytest-coverage: Test coverage

You can add pytest configuration in pyproject.toml file

python_classes = "Test* *Tests"
addopts = "-vv -x -s --cov=app --cov-report term-missing"

Using Github actions to run your tests

If you're using GitHub to host your code, then you should use GitHub actions to run your CI pipelines.

Here is a simple CI workflow to start with

$ mkdir .github/workflows
$ cat <<EOF > .github/workflows/ci.yaml
name: CI
on: [push]
    runs-on: ubuntu-22.04
      - uses: actions/checkout@master
          fetch-depth: 1
      - uses: taiki-e/install-action@just
      - uses: abatilo/actions-poetry@v2
      - run: just setup
      - run: just test

in Justfile we need to define those targets: setup and test.

# Justfile
    @just --list

    #!/usr/bin/env bash
    poetry install --no-root

    #!/usr/bin/env bash
    source .venv/bin/activate &&
    pytest tests/

Build and publish your app

wheel is a packaging format for python code. When you install a package with pip, chances are you installed a wheel package.

Poetry has a builtin functionality to build wheels packages

$ poetry build

That will produce a wheel package in dist directory

If you are not using Poetry, then to be able to build a wheel package, you first need to install build and wheel packages

$ pip install --user build wheel

Then update the pyproject.toml file to include the necessary information

And finally build the wheel

$ python -m build --wheel

More info here

Containerize your app

To be able to distribute your web app easily, one way is to package it in an OCI image

We're going to use podman, which is an alternative to Docker, to build the OCI image for our web app.

We need first to create a Containerfile

$ cat <<EOF > Containerfile
FROM python:3.11-alpine AS builder


WORKDIR /build

RUN apk update && \
    apk add git gcc curl && \
    curl -sSL | python3 -

COPY poetry.lock pyproject.toml  ./

RUN poetry export \
    --without-hashes \
    -f requirements.txt \
    --output requirements.txt \
    --only main

RUN pip install --prefix /local --no-cache-dir pip && \
    pip install --prefix /local -I --no-cache-dir -r requirements.txt

FROM python:3.11-alpine
RUN apk update && apk add just tzdata
RUN cp /usr/share/zoneinfo/UTC /etc/localtime
RUN adduser --home /app --disabled-password app
COPY --from=builder /local/ /usr/local
COPY --chown=app:app . /app
USER app

Then build the OCI image

$ podman -f Containerfle -t simplewebapp:latest .

That will produce an OCI image that you can list as follows:

$ podman images --filter=reference=localhost/simplewebapp

Improve your git experience

Use lazygit

lazygit is a fantastic tool to manage the git flows with ease.

taken from lazygit repo

Check the github repo to learn how to use it.

Tips 💡
There is a plugin to integrate lazygit into neovim called lazygit.nvim

Use pre-commit

pre-commit will let you catch some errors before pushing your code.

You need first to install it:

$ pip install --user pre-commit
ℹ️ Note
If you're using poetry, you can add it as a dev dependency
$ poetry add --group=dev pre-commit

Then you need to create a configuration file .pre-commit-config.yaml where you specify the hooks to run:

Here is an example of configuration file you can use:

$ cat <<EOF > .pre-commit-config.yaml
  - repo:
    rev: v4.3.0
      - id: debug-statements
      - id: end-of-file-fixer
      - id: trailing-whitespace

  - repo:
    rev: 22.10.0
      - id: black

  - repo:
    rev: v0.0.246
      - id: ruff

Then run this command to update the hooks

$ pre-commit autoupdate

Finally install the hooks scripts with

$ pre-commit install

now pre-commit will run after every git commit

What's next ?

Maybe you want to explore how to deploy your app in k8s and you want to have a CD pipeline. You can check my previous post about CD pipeline using fluxcd 👉 here

Read more ...

compile deeplearning libraries for jetson nano

Create your own image for jetson nano board

simple facial recognition system