Python in Production

I’m missing a key part from the public Python discourse and I would like to help to change that.

The other day I was listening to a podcast about running Python services in production. While I disagreed with some of the choices they made, it acutely reminded me about what I’ve been missing in the past years from the public Python discourse.

It’s about moving from a huge Zope app (I’ve never used Zope) to a hybrid of Zope surrounded by Flask microservices (amen to Goodbye Microservices ), about mindblowing Postgres features you may not know about (but are overkill for me), and they explain how and why they picked their cloud provider (I run nothing in the cloud).

And yet despite the fact that the details aren’t relevant to me, the mindsets , thought processes , and stories around it captivated me and I happily listened to it on my vacation.

Python conferences were a lot more like this. I remember startups and established companies alike to talk about running Python in production, lessons learned, and so on. Nowadays I feel like I’m one of the few remaining people talking publicly about doing more with Python than scripting AWS and/or Kubernetes (Instagram and to a certain degree Spotify being notable exceptions). The remaining places where we get to talk about these things are insider IRC channels, Twitter, and open spaces at PyCon. Even just being able to constructively disagree with someone is a gift in itself.

The scientific stack and education have carried Python through the Python 2 to 3 transitions that was – as my friend Glyph put it – a huge disaster and I’ll always be grateful for that. Nonetheless, nowadays I often feel strangely estranged at many Python conferences.

And while Node, Go, and Rust certainly ate much of that market, I know for a fact that the number of companies running Python services is not in the single digits.

But from my time on the PyCon US program committee I also know that they are not rejecting such talks out of spite. It’s simply that the vast majority of proposals nowadays reflect the composition of the Python community: science and education. The amount of accepted talks that are not from those two areas are usually way out of proportion with the submissions.

An Offer

So in a completely egoistical move, I would like to encourage people who do interesting stuff with Python to run websites or some kind of web and network services to tell us about it at PyCons, meetups, and in blogs. To help with that, I’m open to mentoring people who feel like they might have something to share but find the process daunting. At this point I have a 7 for 7 acceptance rate at PyCon US and I know a thing or two about speaking at conferences – so I think I can increase your chances to get through.

People keep saying that the hard production/web problems are solved and Python is boring. But that’s patently not true:

• Many companies are going through the Python 2 to 3 migration right now . Often, “just use six and port all of it” is not an option and other techniques are employed. Some are well-known but not by everyone. A lot of time is wasted right now by everyone figuring it out in their own.
• Most people are not running their apps in Kubernetes in the cloud – often for good reasons. Why? And how?
• Docker is old news and yet it’s not a solved problem as my friend Itamar’s site proves: https://pythonspeed.com/docker/ . Did you know that using Alpine-based base images for Python apps is a terrible idea?
• The async ecosystem today is where the sync ecosystem was ten years ago: we have many frameworks and experiments and it’s both exciting and overwhelming if you try to follow it closely.
• Standards like OpenAPI and GraphQL have emerged. There’s a bunch of libs and frameworks of varying maturity – someone is probably using them? How do they fit into larger systems?
• There are much better patterns at using micro frameworks like Flask or Falcon than returning dictionaries and strings. Armin talked about this four years ago and yet I don’t see people talk about this at all.
• Is there finally an OAuth 2.0 library for Python that won’t make its users want to walk on LEGO bricks? (I legitimately don’t know – I just remember it being a huge pain point years ago.)

I could go on and on. Not all of these topics will get you a spot at PyCon US but if the ecosystem showed more presence at smaller conferences, meetups, podcasts, and blog posts, it would help everyone. And I’d feel less lonely which of course should be everyone’s ultimate goal.

Your chances of me helping are increased if you’re part of an URM and/or if I find your topic interesting. Please accept my apology if I can’t help you specifically, but I’ll try to find time for as many people as my time permits.

Please reach out: https://hynek.me/about/ .