Jonathan Brachthäuser

I am more than proud of my PhD students, colleagues and collaborators! Together we managed to have 5 papers at this year's OOPSLA'25 (R1 and R2) and 1 at the colocated ICFP'25. 🥳🎉 The papers cover a wide range of PL and compiler topics 🧵

Claude code, with no special training for theorem proving, converted zlib to Lean and proved the roundtrip correct. Minimal human guidance. Decompress after compress always returns the original data. github.com/kim-em/lean-zip

goto makes a comeback. The goto rule might be simple but Hoare pointed out its application is complicated for human brains, but LLM-code generators don't care.

1/2 Excited about Flix or effects? Help us out by solving some programming challenges in our interactive study. Takes 10-15min. Link👇

It has been an honor and pleasure doing my Postdoc with Martin!

Martin Odersky, EPFL professor and head of the Programming Methods Laboratory, developed the programming language Scala, which is used by thousands of companies worldwide. In 2025, he was awarded the Programming Languages Achievement Award. go.epfl.ch/390f0f

The program for the Oregon Programming Languages Summer School 2026 looks exciting: cs.uoregon.edu/research/summe… Come or send your students!

Anybody at POPL (@poplconf) this week? If so, please say hi to our student Alexander Phi Goetz, who is presenting his work on type system design for delimited control operators at the SRC. It is his first conference and I am sure he would be happy if you say hi :)

@aramh Maybe this article by my friend David could interest you (if you didn't know it already) binderdavid.github.io/publications/B…

I really don't understand focusing (proof theory), but I feel like I should.

We're looking to hire an experienced editor with a taste for fancy type-systems to work on education for OxCaml! Please share with anyone you think might be a good fit for this role. (And...retweets wouldn't hurt.)

@aramh We wrote a grant proposal about this very point ~4 years ago. It was rejected. Reviewers said: who cares about this? By now we can motivate it way more practically, but I am still sad why mathematical elegance and curiosity isn't enough motivation.

In a programming language you need to control when things are executed. Call by push value was a first attempt towards that but a more general approach comes when all your types are polarized and dual (e.g. Par or Polarity, or any language that is a term calculus for a symmetric logic such as classical linear logic or classical logic). In a distributed system in order to be robust and reliable (provide backpressure, etc) you need to control when things are executed. This basically means control on push vs. pull behavior and a way to turn one into the other (another form of duality). In a programming language you need to control when your program produces effects and how the outside environment interacts with your program. There are various ways to do this with differing tradeoffs, but an increasingly popular way is through algebraic effects, where effects and co-effects are algebraically dual (I believe, not quite sure about this, please correct me if I am wrong). In functional programming it's easy to add more interpretations to a data types, but it is difficult (costly) to extend it. In an object oriented language it is trivial to add more cases to a data types, but it is difficult (clunky, expensive) to add more interpretations (visitor patterns and other abstrusions). This is, of course, the famous Expression Problem. Turns out that if you make FP and OOP dual you can also solve the expression problem (again see Polarity). I see a pattern here, let me add to it. Session type duality: producers are dual to consumers (also relevant to the distributed system case). Player vs opponent. Program vs. context. Syntax-semantics duality (at least for full abstraction). Various Stone dualities in the same area. Forward vs. reverse automatic differentiation. Etc, etc, etc. It's time to recognize that if your system doesn't have duality, it's probably missing some fundamental feature. It should be a good design principle that whatever system you're designing has to have an explicit dual structure.

POPL'26 registration has opened! We will have an amazing program -- join us! popl26.sigplan.org/attending/regi…

The ACM SIGPLAN Workshop on Partial Evaluation and Program Manipulation (PEPM'26), co-located with POPL'26 in Rennes, France is calling for papers! Submission deadline: Friday 24 Oct 2025 popl26.sigplan.org/home/pepm-2026

I am very happy that you enjoy our work :) Let me know if you have any questions

@AppleSupport Here is the command I use to delete the plist files in order to allow reconnecting (after reboot): sudo rm /Library/Preferences/com.apple.windowserver.displays.plist sudo rm ~/Library/Preferences/ByHost/com.apple.windowserver*.plist This workaround is not viable in practice.

@AppleSupport I can reproduce this bug reliably. Similar issues showed up multiple times now on different projectors which makes it impossible to reconnect. This is a major blocker for me since I need to mirror my screen in my lectures.

@AppleSupport M1 MacBook Pro (macOS 15.7.1): Using "Mirror app window" via HDMI then "Stop mirroring window" permanently breaks external display detection. Projector shows black screen, display disappears from settings. Fix: delete display plist files via Terminal & reboot.

@grashalm_ Thanks Christian :)

@__protected Impressive line-up. Congratulations to everyone.

That's it from us for this ICFP / SPLASH 🫳🎤 If you missed our talks, you can read about them here se.cs.uni-tuebingen.de/singapore2025/ and I'm sure the recordings will pop up on the SIGPLAN YouTube channel at some point.

@icfp_conference / @splashcon is over. Thanks to @ilyasergey, Charles, @ShriramKMurthi, @sukyoungryu, Dominique, @NeringaYoung, and all other organizers for their amazing work, making all of this happen! It has been a lot of funand I hope to see you all very soon again!

Really great talk, Jiri!

You can find our paper here: se.cs.uni-tuebingen.de/publications/b… The online demo is here: se-tuebingen.github.io/oopsla-2025-ar… And if you can't attend, you have the chance to watch it live: youtube.com/live/j_nXuGhMF…

The sixth and last talk of my group and collaborators will be by @jiribenes_ on the Simple Essence of Overload Resolution. It'll be at 11:45 local time in Orchid East, so make sure to drop by.