Python - A Reflection in 2022#
I built my career on Python. In 2014, I had to make a decision. It was between either choosing to learn Ruby or Python. I was helping to create a tool and process for a team of content writers at Flipkart – I was a content writer there, and I’d used Excel for it at first, until Excel couldn’t handle the queries anymore. I called my friend who recommended Python. I learnt Python, almost over a course of a weekend, and I was making a PyQT4 application in a few weeks.
Note
Before you read this post, it might be prudent to read For Those Who Came in Late
I knew nothing about software development. I was using Github to store the code, this was a personal project. And I was having fun. It was enjoyable, really. The I installed Python on all my colleagues’ laptops, and my application ran. I eventually found PyInstaller and gave them all executables instead. It was really fun.
Somehow, I wonder if the entrypoint to Python has changed in the years since. I still like the language – it’s still my primary source of income. However, something about it seems to have changed at a fundamental level.
I didn’t learn about virtual environments until a long time later. I was using Python 2.7, since I learnt from Zed Shaw, who was a staunch opposer of Python 3 for the longest time. I didn’t need any of that.
Today, if someone came to me and asked me about getting started with Python, what would I tell them? Would it be the same things I experienced, or would I overwhelm them with unnecessary tooling?
What’s the best way to install Python on your machine? I no longer run sudo
apt-get install python python-pip
. I don’t remember the last time I had to use
get-pip.py. I have used
miniconda, Anaconda, asdf,
pyenv, and poetry.
Right now, I like asdf
, but I have a feeling I won’t use it too long. That
deserves a post in and of itself really.
What’s the best way to manage dependencies? Is it creating a
virtual environment
and capturing the dependencies in a requirements.txt
file, or perhaps you’d
like to use poetry
to do that and store only your direct dependencies in a
pyproject.toml file?
Maybe you’ve heard of pip-tools and
want to use a requirements.in
file which you can then use to generate a
requirements.txt
. Or do you want to write the requirements file yourself?
Have you heard how nodejs,
Rust and
other languages generate a .lock
file or with the hashes of the dependencies
so you’re safer when using other registries? Maybe someone has told you you
should be looking at the output from dependabot
when you get an alert about malicious packages or packages with
vulnerabilities.
And deployment. How do you deploy? Do you recreate your development environment in a cloud server? Perhaps you have a nice little shell script that does this. Or maybe you’ve heard of Ansible, to automate this. Or maybe Packer. Or maybe your company has its own deployment tool. Do you use Docker or Kubernetes? Do you still use Docker or should you use Podman? And how do you build a Docker image anyway? Should you redo the steps you’re doing on your laptop on your container or should you just use a version-tagged python container and then install packages at a global level because the container runs in an isolated environment, using something you’ve heard called cgroups?
Isn’t this exhausting? I find it to be. It wasn’t this way before. And I’ve still not spoken about Typing and uploading your .tar.gz or wheel file to pypi.org. And I’ve not spoken about Python versions yet.
What’s even going on in the Python world? Somehow, the language is becoming less enjoyable to me. I remember having so much fun with it. What happened to the Zen of Python.
There should be one– and preferably only one –obvious way to do it.
What happened to that? As it stands, these are only the ways I remember off the top of my head. I haven’t spoken about virtualenvwrapper or pipenv or the deadsnakes PPA for Ubuntu. The role of all these tools is to make things easier, so why does it feel so tiresome? I cannot imagine how it must feel as a newcomer to this world. Do bootcamps cover all this? They’d lose me on day 1. I have been in the business of mentoring younger developers and students who reach out to me for help, and I never really tell them all this stuff. It seems tiresome to me, and I cannot imagine what they must feel.
In the last few weeks I’ve been learning Golang and I admit, I’ve had the longest resistance to Golang. I wanted to, want to, learn Rust instead, but Go is more popular, and it’s easier to convince people to pick it up as opposed to getting them excited about Rust. I still want to learn Rust and build things with it, but I want to use Golang for whatever I’ve been doing with Python.
gopls has been a sheer joy to use with Neovim, and I’ve never had that sort of experience with Python, even though I use pyright.
But I still love Python. It’s been my main language for about 8 years now. I didn’t know any other language enough to build things with before it. And I had fun with it. I cannot imagine even building half the things I’ve built with Python in any other language. But I feel like I want to return to the basics with it.
What does that entail? I’m not sure really. I’ve tried downloading and building
versions of Python and using python3 -m venv env
to create environments,
but it feels too clunky a process. Or perhaps, whenever I need a new
environment, I can build an entirely new version of Python, store them all in
~/python/py<version>/<project-name>
, and use direnv
to automatically enable/disable it when I enter the folder.
Tip
Did you know about shims? I had been searching for something like this for ages and didn’t know what the right term was! Direnv is amazing.
The state of Python both excites and exhausts me. It’s good that there are so many tools now, but it cannot be easy. If I delve on this too much, I might end up making my own tool. It’ll end up being like the battle of the Frameworks. And for what it’s worth, I still think Flask is better than Django. I’m more attuned to doing things explicitly than having an entire framework do things for you. Call me old fashioned.