Few days ago, in AIML cohort that I teach, one of the student faced a problem where some python package refused to install because it required python 3.10 or higher.

So I asked the student to install the latest python using

uv python install 3.14

But we kept getting the same error.

Turns out, uv uses system python by default. (Which was 3.8)

Here are the steps that ensure uv uses specific version for your project.

I wanted to try this new (still under preview, but when has that stopped me 😄) type checker named ty ¹ from astral

Installing(?) was as easy as uvx run ty ¯_(ツ)_/¯

I wanted to try it on Django project I’m working on.

Recently there was an article on LinkedIn about how debugger statement was left in the production code (front end javascript) and thus user could not move forward.

People commented about the code review.

But why invest manual efforts when such things can be automated ? See #4 in the list below.

pre-commit hook is an under appreciated (dare I say, unknown) feature of git

As the name suggests, set of commands/scripts can be executed before git commit

What is direnv

direnv is a tool that allows you to change your environment based on the configuration in that folder.

e.g. You can set different environment variables for different folders.

The reason I revisited direnv is because for python project, we need to switch to different virtual environment each time we change a project.

Wouldn’t it be nice if correct virtual environment was activated when you change to that directory

Default command for creating a new project is :

django-admin startproject my_new_project

Django will then create a folder called my_new_project (if it does not exist)

But if folder my_new_project exists (even if empty) we get error like :

CommandError: '/path/to/my_new_project' already exists

The way around is to specify a folder name after the project name like

django-admin startproject my_new_project existing_folder

See the documentation for details.

It is (almost) drop-in replacement for pip. We just need to invoke it as uv pip instead of pip

How to install ?

• pipx install uv : This makes it available everywhere.
• Other alternatives are curl (recommended) and brew (on macOS)

It is superfast

I tried installing Django

$ uv pip install Django
Resolved 3 packages in 138ms
Downloaded 3 packages in 2.53s
Installed 3 packages in 228ms
 + asgiref==3.8.1
 + django==5.0.6
 + sqlparse==0.5.0

This took 3s (As opposed to 33s taken by pip)

Today I wrote (OK, copy/pasted sample code and modified) python after more than a year. (Is it a year or an year - confusing. I think either works)

I had to start from installing python.

Why did I use python ?

At work, we are trying to connect to Azure Service Bus over AMQP. Ruby is not officially supported by MS anymore.

Python is.

But we wanted to try non-azure client. So went with qpid-proton (which also has Ruby version, but I was unable to get it even installed - on macOS - after struggling a lot)

If I were to explain bundler to pythonista (like myself) I would say

Bundler in Ruby land is like poetry in python land, except it does not create sandbox environment

To explain it a little more. It tracks dependencies in a Gemfile (and Gemfile.lock) which then goes in your repo. Other engineers sharing your code would then run bundle install, and they get exact same Gems (including versions) on their machines.