Turns out le Chonk is into rocket science.
Mistral Large 4 scoring well in engineering and astronomy on Ulam’s private benchmark, STEMBench: https://t.co/OW3reppzWN
@samruby At the pace this is moving, it won't be long before you can compile any Rails app into a single binary and keep maintaining it in the language you actually love 🙌
Getting closer with #roundhouse:
Update on compiling a real Rails API app to a native binary (Ruby → C via roundhouse + #Spinel): it builds with 0 errors on today's upstream, links, and serves.
Route-by-route vs real Rails: 91 of 199 called routes match exactly (45.7%); the external API is at 27/39 (69%).
All fixes went upstream as PRs and issues to rubys/roundhouse and matz/spinel.
The next batch should add about 20 more routes.
Compiling a real production @rzilient_club Rails API into a single native binary with #roundhouse + #Spinel (AOT Ruby → C):
it now builds, boots and serves, with responses checked field by field against Rails. Along the way: 69 PRs merged into roundhouse and 31 Spinel bugs filed (all 31 fixed).
Next: a full route-by-route comparison against Rails. And for Mastodon, following @samruby 's advice, the new Futamura target, keep Rails and the gems, then profile and specialize the hot spots.
What a time to contribute to open source 🔥
Agentic setup: point an agent at my Rails API, run it through roundhouse → spinel, get a native binary, report every issue it hits upstream and visualize it all on a custom dashboard locally, crazy
@REDACTED_SECRET@dhh Wow, I wouldn't make a statement like that. Just my opinion, but history is full of people from every background building and sharing the same societies. If anything, I think there's a trend to focus more on what makes us different than on what makes us equal.
We’ve lost an icon. Margaret Hamilton, a computer science pioneer best known for her work at MIT leading development of the flight software for the Apollo missions to the moon, has died at 90. Read about her legacy via MIT News: https://t.co/ODkbdIsoHw
@samruby If roundhouse gets close to a hand port while keeping the gems, the rewrite-vs-compile debate changes completely. Getting close may be good enough.
Pues nada, en estas elecciones generales voy a votar por correo, ni que sea para tocar los huevos. Y como se les ocurra impugnar mi voto, ¡pa'lante denuncia !
Right! 👏 In Ronny the order is allowlist then spam check, but every allowlisted mail still goes through it. Being on the list is necessary, not sufficient, for exactly the reason you give.
Your comment made me look again, the allowlist matches the From: header, and the header checks only looked at SPF and DKIM, which don't verify that address.
I'll add a dmarc=fail check, so a spoofed allowlisted sender is caught deterministically rather than left to the model.
I’ve been building Ronny, a personal assistant that watches my mailbox and talks to me on Telegram. It’s written in Zig 0.16 and runs on my home server: about 15k lines in 28 files, with 133 tests.
https://t.co/RHmVn3FwRo
What it does
Watches the inbox with IMAP IDLE and notifies me on Telegram about mail from an allow-list of senders. A local spam check runs first, so even an allow listed sender’s marketing or phishing gets dropped.
Takes text or voice commands to search, read, summarize and draft replies, and to add calendar entries. Voice notes are transcribed with whisper.cpp on the GPU, and summaries can come back as voice notes made with piper and ffmpeg.
New: meeting reminders. It checks Google Calendar and messages me N minutes before any entry with a link to join. That link can be a Meet link, a conference add-on, or a known meeting host found in the location or description. If the entry has an agenda, it’s summarized by the local model and sent as text or a voice note, depending on my settings. The lead time and the on/off switch are chat settings (“remind me 15 minutes before meetings”).
AMD quietly built its own Linux distro for AI developers.
And it’s not just Ubuntu with AMD branding.
It’s called the Ryzen AI Developer Platform OS, and AMD built it specifically for its Ryzen AI Halo hardware.
AMD built it on Debian instead of using a stock Linux distro. It comes with the drivers and AI tools already set up for Ryzen AI hardware.
The latest release now runs Linux 7.2.7 with ROCm 10.0 and includes tools like PyTorch, vLLM, llama.cpp and ComfyUI.
So AMD isn’t just shipping AI chips anymore.
It’s building the Linux stack developers use to actually run AI on them.
Source: https://t.co/sBCvJ34h5M