Arno Dirlam (@arnodirlam) profile picture

Senior Elixir developer, architect & mentor | he/him

Berlin, Germany
Joined June 2011
Arno Dirlam retweeted
The State of Elixir 2026 survey is open! How is the Elixir ecosystem changing in 2026? Help us find out. Take the survey → elixir-hub.com/surveys/2026/… This is our fourth edition. Last year, 1,000+ developers took part. The more of you answer, the more useful the results are for everyone: teams choosing tools, companies hiring, and maintainers deciding what to build next. It takes about 10 minutes and is open until November 30. As every year, we’ll publish the results on Elixir Hub. Also, reposts and shares within the community we'll appreciate a lot. #elixirlang #myelixirstatus #elixirsurvey #stateofelixir
9
9
1,075
Arno Dirlam retweeted
Elixir 97.5%. TypeScript 61.3%. That's AutoCodeBench, a benchmark from Tencent's Hunyuan team that tests models on problems in 20 languages. Elixir came first. I put the chart on a new Hologram docs page about coding agents. It also shows how to set up your CLAUDE.md or AGENTS.md with one command. hologram.page/docs/ai-coding…
8
13
3
100
4,176
Arno Dirlam retweeted
Anthropic turns down people who are deeply talented technically. Dario Amodei explained why to Jamie Dimon, CEO of JPMorgan, days before a $1.4 trillion IPO. - Everyone goes through a culture interview - No jerks, no one who tries to steal the show, no one indifferent to the public benefit - He admits it's painful in the short run - All 7 co-founders are still at the company - More than 3 in 4 of the first 100 employees are still there - Two senior product people quit and came back within two weeks He spends 40-50% of his own time on talent and culture. It was the first thing he chose to talk about in front of Wall Street. Full breakdown here: theaiopportunities.com/p/ant…
33
39
8
595
194,233
Arno Dirlam retweeted
From Gunicorn and Hackney to erlang_quic and erlang_wasm, I've spent years building open source. 2 ASUS GX10s now run distributed inference for coding agents and Barrel Memory. Seeking AMD, NVIDIA or ASUS hardware partners. I'll share code. Any intro? enki-multimedia.eu/ai-lab/
2
6
384
Arno Dirlam retweeted
MJML EEx 1.0.0 RC has just been published to Hex! This is a giant revamp of the library to get the templating to feel like more like LiveView as opposed to a basic EEx template. It also comes with a ton of prebuilt MJML templates for typical SaaS emails! #ElixirLang #Erlang
2
10
1
51
1,393
Arno Dirlam retweeted
New article! Type inference and narrowing with reverse arrows: elixir-lang.org/blog/2026/11… @duboc_guillaume and I explain how Elixir’s approach to type inference and narrowing is built on a few ideas that compose well while outlining the improvements coming with Elixir v1.21!
4
31
3
229
10,022
Arno Dirlam retweeted
I think this deserves its own thread for anyone interested in how to build a very comprehensive Elixir semantic linter. The original idea was to improve slow LLM responses, as it's cheaper to predictably fix the code than to ask an LLM to do it... this brings us to 👇(1/6)
Replying to @kamilskowron
@kamilskowron We've been running Credence over our (750k+ LoC) codebase at work. Lots of really great suggestions. Can I ask... how did you come up with this set of checks?! It's incredibly comprehensive.
2
7
1
41
2,643
Arno Dirlam retweeted
codex: *works for 30 minutes* codex autocomplete: "this is broken slop. fix the bugs. cut LOC in half. remove useless tests" 10/10 no notes
23
17
4
1,405
56,659
Arno Dirlam retweeted
I find it fascinating to see how many people woke up to the realization that California has a liquid labour market and no non competes. This is one of the most amazing things that creates enormous amount of innovation over there. I'm not sure what miracle created that environment, but it's amazing. I wish more countries had the balls.
25
17
6
614
99,171
Arno Dirlam retweeted
libghostty 1.0 is here! I started Ghostty ~5yrs ago to work towards this. Today, libghostty is in hundreds of projects used by tens of millions of people. 1.0 marks ABI stability, quality, and an invite for distros to begin packaging. mitchellh.com/writing/libgho… Since starting the Ghostty project in 2022, the most common question I've been asked is "why?". Today, I'll answer this by showing the broad industry impact libghostty has already had. On top of JetBrains and Neovim already mentioned (that alone being over 10 million users): -KDE developers are experimenting with a Konsole replacement built on it. - Vercel's Turborepo uses it to render task output in its TUI. - devenv uses it to drive shells. - Antithesis used it to fuzz TUI programs and found real bugs in btop and others. -Ghostel brings it to Emacs. - ghostty-web is the WASM build with an Xterm.js-compatible API. - Commercial products like OrbStack embed it. - And a whole wave of new terminals and "agentic" coding tools, such as herdr and cmux, have been built on it rather than writing yet another emulator. There are community bindings for Rust, Go, JavaScript, Swift, Python, .NET, Elixir, Dart, Java, and more, and the awesome-libghostty list tracks well over a hundred projects. More people today use a terminal powered by libghostty than use Ghostty the application, by a very large multiple. And libghostty is having a much larger impact than being "just" a terminal emulator. I'm excited to see where people take it from here.
113
306
36
4,763
120,505
Arno Dirlam retweeted
Ok, this is neat! Someone reported a bug that Elixir's Task.ignore/1 could leave messages in the caller's inbox. Reproducing it was tricky, out of 1.2M attempts, it only leaked 52 times: github.com/elixir-lang/elixi… Therefore, to consistently "reproduce" the issue, I wrote a property in Lynx (@leanprover + @elixirlang) that Task.ignore/1 should not not leak messages: github.com/josevalim/lynx/bl… Of course, we could not prove this property to be true, because there was indeed a bug! And once we fixed the bug, the proof validated, showing that Task.ignore/1 no longer leave trailing messages in the scenarios dictated by the property. Hooray!
9
35
1
435
15,295
Arno Dirlam retweeted
THREE major model launches could be happening TODAY. The Information has confirmed that GPT-6.2 Astra and Sol are both expected to drop, while Qwen 4 is also set to launch. What a way to close out the week. I'm especially excited to see whether Astra 6.2 can take down Opus 5.5. OpenAI needs to put serious pressure on Anthropic. Maybe that will finally push them to release Fable 5.5 sooner.
33
11
6
419
73,857
Arno Dirlam retweeted
How is there STILL no model drops 😭
444
68
46
6,192
287,431
Arno Dirlam retweeted
Discord in GPUI! LFG! 🎉
On the left is the Discord you use daily. And on the right is a new version we're working on, rebuilt in Rust
21
69
9
2,613
65,059
Arno Dirlam retweeted
GPUI is going to win the future, so easy to get something going quickly and if you use elixir and know how to use Heex/LiveView I have good news: github.com/elixir-crab/gpui works fantasitcally.
On the left is the Discord you use daily. And on the right is a new version we're working on, rebuilt in Rust
7
12
2
161
12,998
Arno Dirlam retweeted
Our security researchers built an autonomous cyber vulnerability discovery, exploitation, and patching pipeline. In a single run with an upcoming cyber model, it validated and generated patches for 948 local privilege escalation vulnerabilities in the Linux kernel at a total API cost of $1.2 million.
677
1,334
913
12,812
2,373,813
Arno Dirlam retweeted
If someone behaves negatively towards you, it helps to remember that he or she is a human being like you and to distinguish between an action and the person who does it. If counter measures are needed to prevent someone doing harm, it's always better to do it with a calm rather than an agitated mind. If you act out of anger, the best part of your brain fails to function. Remember, compassion is not a sign of weakness.
255
861
78
5,621
427,160
Arno Dirlam retweeted
I didn't want to get hacked, and I don't have the funds for a 10-person security operations team. So I built Kodeblue, a security harness built on my years running blue team security ops. Open source, light on tokens, and runs 24/7.

10
11
3
218
289,439