Greetings friend!

I’m writing you from a spa retreat, and I'm not even gonna try to do the apology thing anymore. 2025 was a year of beautiful madness that kept me reeling well into January 2026, and I’m here to get a grip again.

Ultimately, I wrote a talk, one keynote (keep reading!), and one bridesmate wedding speech at my best friend’s wedding. One would not think that it can take more than a month to write a five-minute speech, but only if you don’t know me or don’t realize how much it meant to me.

At the same time, I was juggling my baseline FOSS obligations (Python 3.14 was… fun… for attrs, and have you even been to Hawai‘i if you haven’t spent hours merging Dependabot PRs on your apartment’s balcony?!) and running a huge, important project at work.

A project that at this point has yielded several subprojects that I really hope to get to talk more about in the future and that are kinda surprisingly not satisfactorily solved problems in Python yet:

• A robust Postgres-based worker queue that supports transactional enqueuing and confirmations inspired by Go’s River.

• A Postgres-backed event-sourcing package that implements the concepts from this article, including snapshotting and inline subscribers. It works entirely on immutable classes and is type-safe end-to-end.

• A database-agnostic helper for transaction management after SQLAlchemy 2 showed me how much of my code worked by accident (hashtag autobegin). Interestingly, I’ve run into Kraken Tech people at PyCon UK who made something similar for Django: Django Subatomic.

• And finally, truly generic helpers for filtering, sorting, and paging results – both database and anything else. I’ve only found a bunch of 80% solutions that don’t work together or are inadequate for my needs. This should be a solved problem, but somehow it isn’t, and it’s a tricky problem of composability. Not necessarily in the sense of inheritance vs composition but more in the sense of: how to combine these things while allowing for customization but allowing for static typing? For example: different sources like various databases, web APIs, or just lists and different destinations like various web frameworks or CLI output. It seems to me one of the many cases where something that appears to be a single problem is actually several of them – like validation – and untangling them is the first important step.

I may open-source some of these eventually, but I’m currently buried under work, need to be able to evolve them without breaking half of the world, and my other FOSS projects really need my love. I might be reaching the practical limit of how many open-source projects a single human can maintain without frustrating everyone. :(

Finally, there's, of course, also the good old newsletter paradox where sending out newsletters means losing subscribers, which is my cue for dog content! I didn’t have Barnaby since the last installment (I agree – booo!), so please meet Augustus, whom I met in a Starbucks close to the real-world Arconia (yes, I’m a fan; yes, I took a gate photo)!

As usual, there’s gonna be another good boy at the end for the faithful who all the way to the end!

But now let's take it from the top what happened since… oh no… July 2025!

My first real keynote

Especially in the golden times of conferences pre-2019, I've headlined several of them, but I've never outright written a classical keynote. Honestly, mainly because I didn't feel I had one in me. But in the past few years, an important topic was brewing in my head: what I consider Python's underappreciated and misunderstood superpower. I've announced that idea in August 2022, and it took until 2025 that I finally got to present it at PyCon UK 2025 in the gorgeous CONTACT theater in Manchester.

As I’ve already written elsewhere: this is the only keynote I can give at the moment. I’ve poured everything that I have in me into it and it was a huge relief to see how well it resonated with the very diverse audience.

You see, my biggest fear was to give an unrelatable talk for people who aren’t into Hynek Content™. When I give a regular talk I don’t think about this at all since people in the room chose me and my topic – I “just” give my best. When I’m keynoting, everyone is locked with me in the room.

I know I’ve glazed PyCon UK before, but I shall again: it’s one of my favorite Python conferences. Run well, impeccable choice of venues, extremely friendly despite co-run by my sworn enemy (c.f., intro to my keynote), surprisingly diverse, and not three PyDatas in a trench coat. You should come!

You may have noticed that I haven’t posted the keynote to my homepage for a long time. I was stalling because I hoped to maybe give it again before I make it public-public. It doesn’t look like that’ll work out, so it’s online now. I’m afraid the fact that it’s one hour long with no fat to cut (originally it was 2 hours long, natch) is narrowing the opportunities to virtually zero.

So, here you go:

Upcoming appearance

But! I will be keynoting/headlining PyTexas which runs April 17–19 in Austin’s lovely public library! Since my slot is only 45 minutes, it will be a refreshed version of my IMHO very important Design Pressure talk which filled every room so far. Get your tickets and come and say hi!

Since I love western-style horseback riding and want to milk my ESTA, I will also be on a recon mission to figure out if there’s a way to do a cool cowboy vacation around there. So if you know something, please come and talk to me. 🤠

I’m also fully booked for PyCon US / Long Beach, but I will not be speaking. Hope to run a few fun open spaces, though!

Upcoming video

Speaking of upcoming things! My last video on local uv workflows is now almost a year old, which is extremely embarrassing even all things considered.

The upside is that the comment section has spawned an entire genre of people making fun of waiting for part three, which makes me smile even under the weight of my own guilt. It’s especially sad/funny because part three was the video I set out to do in the first place! But then I realized I have to build the foundation for it first, which resulted in part one and two.

As I’m sure everyone has experienced themselves, the longer you don’t do something, the more difficult it is to get back on the bandwagon. It didn’t help that my retries video is still doing very poorly, and the second uv installment leveled out quite brutally, too. So, one of the reasons for this retreat is truly to reset myself, break through all the resistance that built up over the months (this newsletter being part of it), and get back into creating. 🧘‍♂️ I will not write a new talk this year; I will focus on FOSS, YouTube, and my blog – in that order.

I’m somewhat morbidly curious how all this will play out now. While the traffic to my blog has all but collapsed, in these days of AI slop abundance, I think/hope people will appreciate humanity more than ever – as shown by the latest trend of YouTubers striving for imperfection and holding their lav mics in their hands.

Anyway, March 2026.

Final bow

And all this is only possible thanks to my GitHub Sponsors that I still can’t thank enough! I feel like I’m in a delivery debt, despite the fact that I wrote two talks and shipped more than 15 PyPI releases – not to mention my GitHub Actions. Your support makes that possible and is the reason I can travel to Manchester or Austin and give keynotes and share my experience with y’all. Please consider joining them, guilting me into delivering. It remains incredibly humbling to see people appreciating my work enough to chip in ❤️ and it’s not too late to be featured in my next video! 😇

Of course, a lot more happened since July, but let’s keep it to the important stuff. So here’s your payoff: please meet Orlando, a Silken Windsprite by my friend Caspar, who also runs the objectively best online score board, Keep The Score:

Thanks for sticking with me on this weird ride and have a great 2026!

Cheers,
—h