Hacker News: rurban

New comment by rurban in "Feds suspect ransom negotiator ran ShinyHunters extortion site"

rurban — Wed, 21 Oct 2026 08:52:53 +0000

Killers must be a useful service, because the FBI, CIA and Department of War are all relying in their services.

"By any means" is a current movie with the FBI case.

New comment by rurban in "DeepSeek just open-sourced its inference engine"

rurban — Tue, 20 Oct 2026 16:01:01 +0000

Same for me. The cheap Claude subscription is just so much cheaper. Never had to buy more deepseek tokens for months (August), because Sonnet 5.5 is good enough

New comment by rurban in "DeepInfer: High-performance, large-scale inference framework for DeepSeek models"

rurban — Tue, 20 Oct 2026 15:58:14 +0000

You can buy such a PC off from ebay for only €150.000. We did.

But the architecture is similar to kimi's mixture of experts, and we already have such an optimization for different experts swapped in from disc by antirez.

New comment by rurban in "We fine-tuned a 235B model on 4 GPUs. Then we didn't need it"

rurban — Mon, 19 Oct 2026 17:49:27 +0000

We still fine-tune, because our problems aren't covered by existing models, proprietary tagged images. But not on mixture of experts models. This is only needed for text, not images

New comment by rurban in "Free Pascal compiler now supports dynamic packages"

rurban — Mon, 19 Oct 2026 03:47:08 +0000

Anthony: Whats "dynamioc library support"? Is that a typo in the commit message?

New comment by rurban in "Microsoft and CMU team up to train early career software engineers"

rurban — Sun, 18 Oct 2026 14:33:07 +0000

Wasn't Carnegie Mellon once a respected university, at least in mny times? When did it go downhill like this?

New comment by rurban in "Microsoft Teams is a red flag"

rurban — Sun, 18 Oct 2026 12:45:00 +0000

Not necessarily. We are a Linux company, which just happens to use github business and ms teams, because it's so much cheaper for smaller companies.

New comment by rurban in "We found a Tencent agent fleet"

rurban — Sat, 17 Oct 2026 14:38:52 +0000

Contrary to the Israeli team, this AI fleet behaved properly

We found a Tencent agent fleet

rurban — Sat, 17 Oct 2026 14:38:52 +0000

Article URL: https://swarmcha.se/posts/chinese-agent-fleet-full-report

Comments URL: https://news.ycombinator.com/item?id=50131584

Points: 5

# Comments: 1

New comment by rurban in "Claude reveals ECDSA curve origins"

rurban — Sat, 17 Oct 2026 09:01:33 +0000

Hello, the US government paid the IPSEC devs to include weak and backdoored algos back then. All revealed during the Snowden era

New comment by rurban in "C++26 Contracts: Introduction to Preconditions and Postconditions"

rurban — Fri, 16 Oct 2026 10:30:35 +0000

C++ like contracts. but without the ending semicolon, because that would cause backward compatibility problems. https://github.com/rurban/rcc/#additional-c-features

New comment by rurban in "GitHub Actions has added lock files"

rurban — Fri, 16 Oct 2026 05:53:53 +0000

Which moron community came up again with this wrong naming? Python, ruby or vbscript? I do have to use lock files for concurrency some times. They probably just want to pin versions.

New comment by rurban in "C++26 Contracts: Introduction to Preconditions and Postconditions"

rurban — Fri, 16 Oct 2026 05:18:07 +0000

Of course, that's the idea. At least my C compiler does so

New comment by rurban in "C++26 Contracts: Introduction to Preconditions and Postconditions"

rurban — Thu, 15 Oct 2026 17:58:37 +0000

In my C compiler rcc I throw compile-time errors of course. Not a full SMT solver like z3, just a self written solver, but it works fine and fast enough. Runtime errors are just glorified asserts, lame.

New comment by rurban in "AOSP Source Releases Are So Late Developers Disassemble Libraries to Report Bugs"

rurban — Thu, 15 Oct 2026 13:26:42 +0000

See, that's the problem with proprietary firmware. HarmonyOS would have been better. They also do have tests for such cases.

New comment by rurban in "QGIS enacts moratorium on AI contributions"

rurban — Wed, 14 Oct 2026 09:05:24 +0000

Then replace the maintainers with ones that can do that. Most can.

No need to paste something into a AI tool, just tell the tool which email or which ticket to fix or check. Maintainers who cannot do that, are not worth it.

New comment by rurban in "Ledger confirms spy chip implant behind wallet crypto thefts"

rurban — Wed, 14 Oct 2026 05:14:16 +0000

Likewise from a big coorp US or Swiss company.

The difference is that the western attackers go free, whilst the Chinese attackers will very likely be executed.

New comment by rurban in "QGIS enacts moratorium on AI contributions"

rurban — Tue, 13 Oct 2026 13:41:51 +0000

I have no idea why qgis is overwhelmed. I'm the maintainer of one of the main libs qgis wrote, libredwg. Without AI PR's and internal development usage I would have no chance to advance it and fix all the safety bugs. It's a huge momentum nowadays, and other CAD's also picked up libredwg to support DWG/DXF handling. And they are adding many features. Only Claude contributions though, not much GPT nor Deepseek/Kimi/GLM/zAI.

New comment by rurban in "11 year old unsoundness bugs found in Rust borrow checker"

rurban — Tue, 13 Oct 2026 13:27:33 +0000

And a fork fixed it before upstream rust :)

New comment by rurban in "Hackers activate automatic defrosting in supermarket fridges across Europe"

rurban — Tue, 13 Oct 2026 07:52:54 +0000

Be lucky when they send the DICOM reports via a secure socket. Mostly it's plain http.