Hacker News: fweimer

New comment by fweimer in "What I run on my $10 USD OpenStick homelab"

fweimer — Sun, 08 Nov 2026 17:32:06 +0000

Can you run AArch64 Linux on such a device, or would you better switch to 32-bit Arm due to the very limited RAM capacity?

New comment by fweimer in "ChatGPT silently passes data between connectors and the bounty wants an ID card"

fweimer — Fri, 06 Nov 2026 23:51:42 +0000

A common reporting path for bounties and non-bounty vulnerabilities has always been problematic. This leads to weird scenarios where the company cannot accept reports of denial-of-service vulnerabilities because they are out of scope for bounties, for example. Exclusions from bounties are one thing, but not acting on relevant problems is just a bad idea. The typical bounty program terms are also incompatible with multi-vendor disclosure, where no single vendor can take complete ownership over the disclosure process.

The close proximity to the bounty process can also be challenging to corporate reporters because it could give the impression that they are individually soliciting a gift of non-nominal value from a company that is a supplier or customer in other contexts.

New comment by fweimer in "Tell HN: Gemini trains on chat history, even for subscribers (personal accounts)"

fweimer — Wed, 04 Nov 2026 11:41:00 +0000

Even if they don't train on your content, Google reserves the right to use it for improving their services in other ways:

“ This license is for the limited purpose of operating the services, which means allowing the services to work as designed and creating new features and functionalities. ”

https://policies.google.com/terms?hl=en-US#toc-permission

The training exclusion has always been a distraction. These service providers have plenty of other uses for your data.

New comment by fweimer in "Novellas with Groff"

fweimer — Sun, 01 Nov 2026 17:36:27 +0000

O'Reilly used to typeset their books using troff/groff: http://www.lemis.com/grog/Documentation/PUS/tmac.Gbignuts.G (“O'Reilly & Associates book macros by Steve Talbott, Lenny Muellner”)

New comment by fweimer in "We're releasing the weights of Mistral Large 4"

fweimer — Thu, 29 Oct 2026 16:29:57 +0000

Allen AI imposes further restrictions on their model releases, I think: https://allenai.org/responsible-use

New comment by fweimer in "Anthropic S-1"

fweimer — Mon, 26 Oct 2026 13:38:16 +0000

If the user is held accountable when the AI goes wrong, it discourages too many applications. Blaming people is incompatible with the business goals, I think. That's very different from weapons and other dual-use tools.

New comment by fweimer in "Disney Raising Prices Days After Confirming Ads Will Now Show Up on Every Tier"

fweimer — Fri, 23 Oct 2026 16:06:42 +0000

The most charitable interpretation: The new wording includes all forms of covert advertisement (say embedded in the content itself). All their plans always had that.

New comment by fweimer in "OpenSSL 4.1.0"

fweimer — Fri, 23 Oct 2026 15:43:59 +0000

Not sure if the Stratus reference is correct. Big-endian x86-64 is a real thing on HPE NonStop:

“ The executable format that works on HPE Nonstop Compute's OSS is ELF64-bit, Big_Endian, X86_64. ”

https://inbox.sourceware.org/gcc-help/AM7PR02MB6483A12F6B98D...

New comment by fweimer in "Postgres recovery_prefetch can cause outages"

fweimer — Fri, 23 Oct 2026 08:06:30 +0000

Does the 16 come from maintenance_io_concurrency, which is also used to limit the prefetch depth?

I still think some details are missing for a true explanation of what is going on. Where does the random access pattern come from? Is prefetch unnecessarily performed for full-page images?

New comment by fweimer in "Tell HN: I unknowingly broke the PHP RCE embargo. I am sorry"

fweimer — Fri, 23 Oct 2026 07:09:14 +0000

It's not a press embargo in the sense that the whole package of information becomes public once the embargo expires. There will be a published record on the cve.org, but often those who initiated the embargo insist that key details related to the vulnerability remain private even after the existence of the vulnerability and its remediation have been published.

Consider this example:

https://www.cve.org/CVERecord?id=CVE-2026-113119

At this time, the corresponding Chromium bug tracker entry is still private:

https://issues.chromium.org/issues/557075191

(I just picked this one at random, I haven't checked if the information is publicly available from another source.)

New comment by fweimer in "C# now beats Rust on WASM size"

fweimer — Wed, 21 Oct 2026 17:01:35 +0000

I thought that C# does not have escaping interior pointers, unlike Go. Last time I looked at the CLR, I couldn't find support for interior pointers.

New comment by fweimer in "C# now beats Rust on WASM size"

fweimer — Wed, 21 Oct 2026 09:25:34 +0000

Is the support gap mostly related to weak references and finalizers? Or is something else missing?

(Obviously those are already pretty significant gaps.)

New comment by fweimer in "What makes Lisp difficult to read?"

fweimer — Mon, 28 Sep 2026 18:54:20 +0000

I think with CLOS, you can go back to using single verbs for function names. However, this has a performance cost. Encoding the receiver type in the function names makes the call much easier to optimize.

What remains is field access. Curiously, first-edition K&R C actually required similar type-encoding in field names because field names were global:

https://usenet.trashworldnews.com/?thread=288637

Same for ML, which is roughly of the same age, except that it took much longer for ML-derived languages to lift this restriction.

On the other hand, every time I wrote down examples that I think should look really bad in prefix notation, like this one:

    p3.x := (p1.x + p2.x) / 2;

It is actually not so terrible with a sufficiently streamlined prefix syntax:

    (setf x p3 (/ (+ (. x p1) (. x p2)) 2))

But maybe there should be a mechanism that the reader transforms p1.x to (. x p1), in the same way that it transforms 's into (quote s). Then we get this:

    (setf x p3 (/ (+ p1.x p2.x 2)))

Or if there is a concept of a place:

    (set p3.x (/ (+ p1.x p2.x 2)))

But that probably needs static typing for an efficient implementation (or set needs to deconstruct its first argument during compilation).

New comment by fweimer in "Italian parliament votes for return to nuclear energy"

fweimer — Wed, 23 Sep 2026 20:51:06 +0000

Where in Italy would you put reactors?

https://maps.eu-risk.eucentre.it/map/european-seismic-risk-i...

Sardinia?

New comment by fweimer in "I asked Meta’s Muse for its filesystem and it sent me 6.8GB"

fweimer — Tue, 22 Sep 2026 18:25:17 +0000

When emulated on today's hardware. But the hardware actually available at the time was pretty unreliable by modern standards, I think.

New comment by fweimer in "I asked Meta’s Muse for its filesystem and it sent me 6.8GB"

fweimer — Tue, 22 Sep 2026 17:20:49 +0000

Exfiltrating many binaries gives you the right to their source code, or at least triggers attribution requirements for licensing compliance.

But perhaps Meta did the smart thing and put the source code into the VM, too. That would be a very reliable indicator that they expected exfiltration, and this is in fact working as intended.

New comment by fweimer in "I asked Meta’s Muse for its filesystem and it sent me 6.8GB"

fweimer — Tue, 22 Sep 2026 17:17:45 +0000

I'm pretty sure people said this about the early COBOL compilers, too. They were buggy, the API had terrible uptime, and was slow to respond.

Overall, this whole approach to programming seems to align really well with the original premise of COBOL. I wonder when people will start putting

# Identification Division

into their Markdown files.

New comment by fweimer in "Why MCP Was Always a Bad Idea?"

fweimer — Sun, 20 Sep 2026 20:14:59 +0000

There's probably still value (if you want to call it that) in it as a proxy, both to bypass IP address rate limits and to add necessarily credentials.

There's also another aspect Quite a few API providers provide automatic renewal for MCP server registrations, but not for personal access tokens. This may be less relevant when models just drive the user's browser.

New comment by fweimer in "ChatGPT now knows what you do on other websites via ad collector"

fweimer — Sun, 20 Sep 2026 19:34:49 +0000

Likely a reference to ELIZA: https://en.wikipedia.org/wiki/ELIZA

New comment by fweimer in "Jemalloc 5.4.0"

fweimer — Fri, 18 Sep 2026 17:39:15 +0000

It's just a compiler barrier (signal fence), not a memory barrier that concerns the CPU. The CPU is free to reorder loads and stores.