Vasanth retweeted
systemdで隔離環境作れるsystemd-nspown、2.0.0でdocker engine互換のAPIを実装したらしい
GitLabのCI Runnerで使う事例とかが紹介されてる
blog.nspawn.org/posts/nspawn-3…
GitLabのCI Runnerで使う事例とかが紹介されてる
blog.nspawn.org/posts/nspawn-3…
2
53
208
10,103
Vasanth retweeted
Now available for ACM Members: "Programming Language Fundamentals: A Metalanguage Approach in Elm," by Martin Erwig. For undergraduate students in CS, as well as researchers and practitioners working with programming languages. This book is an intro...
share.percipio.com/cd/0FLnChWtn
share.percipio.com/cd/0FLnChWtn
2
10
39
1,567
Vasanth retweeted
An OpenSSH bug that took 46 years to ripen!
1980
UNIX puts * in the password field of system accounts. The idea is simple: it’s not a real hash, so it can’t be brute-forced and you can’t log in with a password.
2012
The musl libc developers decide that returning NULL from crypt() on failure is dangerous - programs often forget to check it and crash with a segfault. So they return an error string instead.
Guess which one?
The same asterisk: *.
Three weeks later they add a length limit: passwords longer than 256 bytes burn CPU. Anything longer is rejected. And since it’s an error, crypt() returns *.
2016
OpenSSH fixes a timing attack (CVE-2016-6210). If a user doesn’t exist, the server still hashes the password with a fake/borrowed salt so an attacker can’t tell valid usernames by response time.
2022
Alpine Linux changes the locked root password in /etc/shadow from ! to * so SSH keys work properly on cloud images.
… and the loop closes!
You connect over SSH as root (whose shadow entry is *).
You type any password longer than 256 characters (e.g. 300 As).
crypt() hits the length limit → returns * → OpenSSH compares it with the value in the database → that value is also * → you’re in as root.
That chain took 46 years (!!!) to complete.
Every individual change was reasonable. Together they produced a perfect * == * authentication bypass (CVE-2026-123158, nicknamed *sshole). Fixed in OpenSSH 10.6p2 and Alpine packages.
1980
UNIX puts * in the password field of system accounts. The idea is simple: it’s not a real hash, so it can’t be brute-forced and you can’t log in with a password.
2012
The musl libc developers decide that returning NULL from crypt() on failure is dangerous - programs often forget to check it and crash with a segfault. So they return an error string instead.
Guess which one?
The same asterisk: *.
Three weeks later they add a length limit: passwords longer than 256 bytes burn CPU. Anything longer is rejected. And since it’s an error, crypt() returns *.
2016
OpenSSH fixes a timing attack (CVE-2016-6210). If a user doesn’t exist, the server still hashes the password with a fake/borrowed salt so an attacker can’t tell valid usernames by response time.
2022
Alpine Linux changes the locked root password in /etc/shadow from ! to * so SSH keys work properly on cloud images.
… and the loop closes!
You connect over SSH as root (whose shadow entry is *).
You type any password longer than 256 characters (e.g. 300 As).
crypt() hits the length limit → returns * → OpenSSH compares it with the value in the database → that value is also * → you’re in as root.
That chain took 46 years (!!!) to complete.
Every individual change was reasonable. Together they produced a perfect * == * authentication bypass (CVE-2026-123158, nicknamed *sshole). Fixed in OpenSSH 10.6p2 and Alpine packages.
🚨 Critical OpenSSH flaw allows passwordless account access using any password over 256 characters
sshhole.com
A newly disclosed vulnerability dubbed *sshole has been discovered in OpenSSH installations using musl libc, potentially allowing attackers to bypass access to accounts that should reject password logins entirely.
The flaw affects certain configurations of Alpine Linux, OpenWrt, Buildroot, and Yocto where PAM is disabled and password authentication remains enabled. Under specific conditions, attackers can authenticate to locked accounts using arbitrary passwords between 257 and 1,024 characters.
The researcher demonstrated that affected Alpine Linux cloud images could allow attackers to gain root privileges through a passwordless administrative account.
OpenSSH 10.6p2 addresses the issue. Administrators are urged to update immediately or disable SSH password authentication until patches can be applied.
100
561
4,336
226,264
Vasanth retweeted
Here are a few ways to make your app feel faster:
- Use ease-out instead of ease-in for instant feedback
- Keep most UI animations under 300ms
- Use custom easing curves for snappier interactions
- Make your loading spinners spin faster
- Use ease-out instead of ease-in for instant feedback
- Keep most UI animations under 300ms
- Use custom easing curves for snappier interactions
- Make your loading spinners spin faster
55
119
3,248
99,725
Vasanth retweeted
⚠️Did Microsoft just MOG llama.cpp?
It dropped a local inference bombshell.
Qwen3.8-27B running at 266 tps on a 1x RTX 5090.
... and at 260K cx, still 187 tps.
Microsoft's Foundry Local is challenging llama.cpp in a big way. 🇺🇸
Using the same 27B model at 4-bit, Microsoft's ts
Context → Foundry / llama.cpp
⚡ 4K → 265.8 / 150.6 tps
⚡ 16K → 260.2 / 164.5 tps
⚡ 32K → 234.3 / 151.7 tps
⚡ 65K → 200.4 / 146.9 tps
⚡ 131K → 212.4 / 109.7 tps
⚡ 260K → 187.2 / 96.1 tps
That's almost 2× faster at 260K context.
And on the NVIDIA DGX Spark?
At 260K context
🔵 Foundry Local → 30.21 tps
🟠 llama.cpp → 12.82 tps
A 2.36× improvement.
What's under the hood?
🧠 ONNX Runtime
🚀 DFlash2 speculative decoding
⚙️ Optimized CUDA kernels
💾 INT4 weights + quantized KV cache
📚 Prefix caching
🔄 Continuous batching
And here's what I really like.
Microsoft now lets you connect Qwen3.8-27B running on your OWN GPU directly to GitHub Copilot.
🏠 Local inference
💾 ~18GB quantized model
🎮 NVIDIA GPU with 24GB+ VRAM
🔓 Open-weight model
💰 No per-token inference charges
Foundry Local 0.13.0 is available now.
⚠️Remember Microsoft tested against llama.cpp, not NInfer, vLLM, Strata, or every other optimized engine.
The llama.cpp baseline's speculative-decoding configuration also needs clarification.
Link in ALT
It dropped a local inference bombshell.
Qwen3.8-27B running at 266 tps on a 1x RTX 5090.
... and at 260K cx, still 187 tps.
Microsoft's Foundry Local is challenging llama.cpp in a big way. 🇺🇸
Using the same 27B model at 4-bit, Microsoft's ts
Context → Foundry / llama.cpp
⚡ 4K → 265.8 / 150.6 tps
⚡ 16K → 260.2 / 164.5 tps
⚡ 32K → 234.3 / 151.7 tps
⚡ 65K → 200.4 / 146.9 tps
⚡ 131K → 212.4 / 109.7 tps
⚡ 260K → 187.2 / 96.1 tps
That's almost 2× faster at 260K context.
And on the NVIDIA DGX Spark?
At 260K context
🔵 Foundry Local → 30.21 tps
🟠 llama.cpp → 12.82 tps
A 2.36× improvement.
What's under the hood?
🧠 ONNX Runtime
🚀 DFlash2 speculative decoding
⚙️ Optimized CUDA kernels
💾 INT4 weights + quantized KV cache
📚 Prefix caching
🔄 Continuous batching
And here's what I really like.
Microsoft now lets you connect Qwen3.8-27B running on your OWN GPU directly to GitHub Copilot.
🏠 Local inference
💾 ~18GB quantized model
🎮 NVIDIA GPU with 24GB+ VRAM
🔓 Open-weight model
💰 No per-token inference charges
Foundry Local 0.13.0 is available now.
⚠️Remember Microsoft tested against llama.cpp, not NInfer, vLLM, Strata, or every other optimized engine.
The llama.cpp baseline's speculative-decoding configuration also needs clarification.
Link in ALT
29
36
363
29,841
Vasanth retweeted
your code isn’t “clean”. it’s just short because you outsourced the hard parts to 900 packages written by strangers.
CUI:
• plain C, ~150 components
• 240 KB, zero dependencies
• Windows, macOS and Linux
write the hard part yourself for once. or make your AI agent suffer t.co/F8nQ3kQeyI
CUI:
• plain C, ~150 components
• 240 KB, zero dependencies
• Windows, macOS and Linux
write the hard part yourself for once. or make your AI agent suffer t.co/F8nQ3kQeyI
6
8
117
4,250
Vasanth retweeted
The bug is like a joke.
On Alpine Linux, type 257 to 1024 characters of anything and you get root. 💀
On Alpine Linux, type 257 to 1024 characters of anything and you get root. 💀
‼️ BREAKING: OpenSSH has rushed out a fix for a critical flaw that lets attackers log into some Linux servers without the right password.
The team learned of the bug a day earlier. OpenSSH's Damien Miller says long embargoes now mostly help attackers, because AI agents mes within hours.
It hits Linux systems on the musl C library with PAM off, such as Alpine Linux. Miller told admins to disable password logins or SSH itself "IMMEDIATELY".
Version 10.6p2 works around it. Systems on glibc or PAM are not affected, and Miller says the root cause sits outside OpenSSH.
450
2,474
19,760
2,217,910
Vasanth retweeted
The code: github.com/pthom/imgui_ri…
In C++: a few lines of CMake (FetchContent).
In Python: it comes with pip install imgui-bundle, as imgui_bundle.rich_md.
In C++: a few lines of CMake (FetchContent).
In Python: it comes with pip install imgui-bundle, as imgui_bundle.rich_md.
0
1
3
172
Vasanth retweeted
I wanted to understand the structural and philosophical differences between GPUs, TPUs and newer alternatives (like Groq LPUs) without getting drowned in specs and jargon.
Mispronunciation of the day: "soldered" 🤦♀️...
0:00 Intro
0:56 The origin story of TPUs
3:42 The first TPU: iplication
5:21 SRAM vs DRAM
7:21 The memory bandwidth problem
9:14 Systolic arrays
12:48 TPU vs GPU
17:51 Groq’s LPU
25:38 GPU vs LPU
28:36 NVIDIA on the AI chips boom
Mispronunciation of the day: "soldered" 🤦♀️...
0:00 Intro
0:56 The origin story of TPUs
3:42 The first TPU: iplication
5:21 SRAM vs DRAM
7:21 The memory bandwidth problem
9:14 Systolic arrays
12:48 TPU vs GPU
17:51 Groq’s LPU
25:38 GPU vs LPU
28:36 NVIDIA on the AI chips boom
29
175
1,543
100,776
Vasanth retweeted
For those still wondering whether AI will impact the real economy... here's something that a very small group of us pulled off in just a couple of months itnews.com.au/news/westpac-r…
30
16
339
76,747
Vasanth retweeted
Today, we're shipping real-time graphics.
Using Sol 6.1 Ultrafast we can generate custom overlays, diagrams, and even interactive 3D scenes right on a user's screen, no chat.
It's a new visual interface for AI.
Here's me learning about LLMs + hardware.
Demo:
Using Sol 6.1 Ultrafast we can generate custom overlays, diagrams, and even interactive 3D scenes right on a user's screen, no chat.
It's a new visual interface for AI.
Here's me learning about LLMs + hardware.
Demo:
702
843
11,717
1,265,718
Vasanth retweeted
You need just a Mac mini M4 24GB for running this stack.
You can buy one at ~1.5k, put on your desk and having all the local intelligence you need for your daily tasks.
You can buy one at ~1.5k, put on your desk and having all the local intelligence you need for your daily tasks.
I give you a Local AI stack for office:
• MiniCPM-OCR – document parsing
• Liquid LFM2.5 – lightweight agents
• Qwen3 8B – office assistant
• MiniCPM-V – vision
• Qwen3 Embeddings – RAG
• Whisper – transcription
All local.
33
25
644
100,479
Vasanth retweeted
WHAT?! SOMEONE JUST OPEN-SOURCED TERRAFORM FOR AI KNOWLEDGE GRAPHS 🤯
It's called Open Ontologies and it makes AI reasoning verifiable.
Here's the crazy part:
→ Change 1 line, uncover 901 hidden consequences in 0.04s
→ Verify 259 facts, reject a forged one
→ Preview the lying changes
→ Generate proofs anyone can independently check
→ Validate RAG claims against their sources
Works with Claude Code, Cursor, Windsurf, VS Code and more via MCP.
One Rust binary. Runs locally. Fully open source.
100% Free and open-source.
Repo link in 🧵↓
It's called Open Ontologies and it makes AI reasoning verifiable.
Here's the crazy part:
→ Change 1 line, uncover 901 hidden consequences in 0.04s
→ Verify 259 facts, reject a forged one
→ Preview the lying changes
→ Generate proofs anyone can independently check
→ Validate RAG claims against their sources
Works with Claude Code, Cursor, Windsurf, VS Code and more via MCP.
One Rust binary. Runs locally. Fully open source.
100% Free and open-source.
Repo link in 🧵↓
37
115
1,098
92,423
Vasanth retweeted
I'm looking for someone with a PhD (or decently into one), or 10+ years in systems programming, anywhere in the world Remote allows me to pay, to publish a monthly column for @theconsensusdev selecting and putting a CS research paper in context. $600 per article.
No particular length requirement as long as it's well done. Use AI for whatever you want but you must write by hand and own correctness (I'm checking too of course).
Please pass along.
editor@theconsensus.dev
No particular length requirement as long as it's well done. Use AI for whatever you want but you must write by hand and own correctness (I'm checking too of course).
Please pass along.
editor@theconsensus.dev
9
56
278
22,695
Vasanth retweeted
Rust in agentic workflows sure consumes a lot of disk space, right?
If you want this fixed, here's something you and your employer should fund 👇
github.com/rust-lang/goal…
If you want this fixed, here's something you and your employer should fund 👇
github.com/rust-lang/goal…
8
23
181
25,947
Vasanth retweeted
Agaram 2.5B is out. It's a Tamil-first language model we trained from scratch at @MuraiLabsHQ , our small non-profit lab. As far as we know, it's the first Tamil model trained from scratch that can call tools.
It chats in Tamil, Tanglish and English, translates, and can call the base model, chat model and GGUF files under Apache-2.0. You can download it and run it offline on your own laptop.
The first chat model had a four-day public preview. Seeing where it fell short with real users helped us decide what to work on. We rebuilt the fine-tuning around what we saw, using 222,662 records.
Thank you to everyone who spent time with that early version.
The video walks through the model. You can try it here:
hf.co/spaces/mechram…
நன்றி 🙏
It chats in Tamil, Tanglish and English, translates, and can call the base model, chat model and GGUF files under Apache-2.0. You can download it and run it offline on your own laptop.
The first chat model had a four-day public preview. Seeing where it fell short with real users helped us decide what to work on. We rebuilt the fine-tuning around what we saw, using 222,662 records.
Thank you to everyone who spent time with that early version.
The video walks through the model. You can try it here:
hf.co/spaces/mechram…
நன்றி 🙏
113
421
2,272
93,086
Vasanth retweeted
We finally have a fully open-source ESP32 emulator for the Xtensa L6
It's fully upstream brought to you by @schematikio
mikroverk.github.io/esp32sim/run.h…
It's fully upstream brought to you by @schematikio
mikroverk.github.io/esp32sim/run.h…
5
4
51
2,120