Hacker News: bayesnet

New comment by bayesnet in "Zed 1.0"

bayesnet — Wed, 29 Apr 2026 17:48:25 +0000

I'm beating a dead horse here but the challenge is a11y. Chromium wrappers get a11y for free; bespoke UI frameworks must implement accesskit (or something) which is a lot of work and something that (imo sadly) many small teams decide is not worth the investment.

New comment by bayesnet in "Zed 1.0"

bayesnet — Wed, 29 Apr 2026 17:47:02 +0000

I took a look at gpui-component a while ago when assessing GPUI for a project I was working on. IANAL but was dissuaded because it's almost certainly not compliant with the Zed license--gpui-component "borrows" gpui code patterns lifted straight from the main zed repo, which therefore must be AGPL/GPL (unlike the gpui-only which is Apache IIRC). Caveat emptor (caveat user?).

New comment by bayesnet in "Garbage collection without unsafe code"

bayesnet — Wed, 22 Apr 2026 10:42:01 +0000

The existence of a soundness bug in the typechecker doesn’t refute the value of soundness as a language design contract.

If anything it’s the opposite: issues demonstrated by cve-rs are _language bugs_ and are _fixable_ in principle. “Safe Rust should be memory-safe” is a well-defined, falsifiable contract that the compiler can be measured against. Meanwhile memory unsafety is a feature of the semantics of C++ and so it would be absurd to file a bug against gcc complaining that it compiled your faulty code.

New comment by bayesnet in "I'm Sick of AI Everything"

bayesnet — Wed, 22 Apr 2026 02:06:51 +0000

What’s wrong with printers? Imagine designing a laser that bounces off of mirror spinning at 20k+ rpm while coordinating with a paper feeder. Sounds pretty cool to me

New comment by bayesnet in "Claude Opus 4.7"

bayesnet — Thu, 16 Apr 2026 18:59:04 +0000

I don't think it's something special about my workflow and more the application area--I'm writing a lot of Lean lately and particularly knotty proofs can take quite a lot of time. Long thinking intervals are more of a bug than a feature IMO: Even if Claude can one-shot the proof in 40-60 minutes I'd rather have a partial proof in 15 and fill in the gaps myself.

New comment by bayesnet in "Claude Opus 4.7"

bayesnet — Thu, 16 Apr 2026 18:51:46 +0000

I used Gemini CLI for a while because it was free to me. The primary reason I stopped was because it wasn't very good, but their "thinking summaries" didn't help matters. They were model generated and just said things to the effect of "I'm thinking very hard about how to solve this problem" and "I'm laser-focused on the user objective". So I feel you: small things like this make a big difference to usability.

New comment by bayesnet in "Claude Opus 4.7"

bayesnet — Thu, 16 Apr 2026 15:57:09 +0000

This is a CC harness thing than a model thing but the "new" thinking messages ('hmm...', 'this one needs a moment...') are extraordinarily irritating. They're both entirely uninformative and strictly worse than a spinner. On my workflows CC often spends up to an hour thinking (which is fine if the result is good) and seeing these messages does not build confidence.

New comment by bayesnet in "iOS 26.3 and macOS 26.3 Fix Dozens of Vulnerabilities, Including Zero-Day"

bayesnet — Wed, 11 Feb 2026 20:38:50 +0000

Is there any word on whether these vulnerabilities were exploitable on devices with MIE[0]?

[0]: https://security.apple.com/blog/memory-integrity-enforcement...

New comment by bayesnet in "What's up with all those equals signs anyway?"

bayesnet — Tue, 03 Feb 2026 13:27:16 +0000

I know this is grumpy but this I’ve never liked this answer. It is a perfect encapsulation of the elitism in the SO community—if you’re new, your questions are closed and your answers are edited and downvoted. Meanwhile this is tolerated only because it’s posted by a member with high rep and username recognition.

New comment by bayesnet in "That's not how email works"

bayesnet — Thu, 29 Jan 2026 03:24:47 +0000

Small businesses accounts were/are also subject to ring fencing, and my recollection is that large banks sought to recover the costs of ringfencing rules via charges on large clients.

Come to think of it this was all also at the time of very low rates which was more likely to be the issue.

New comment by bayesnet in "That's not how email works"

bayesnet — Wed, 28 Jan 2026 19:53:18 +0000

This is arguable for HSBC (in the UK at least). Ringfencing laws post 2008 have made customer deposits in the UK very difficult to invest profitably, to the point where (at least last time I cared about this) they were charging commercial customers to have UK domiciled accounts.

New comment by bayesnet in "Airfoil (2024)"

bayesnet — Wed, 28 Jan 2026 14:49:12 +0000

It’s kind of sad IMO. Bartosz has made a ton of these super interesting and meticulously designed explainers. Something thrown together with AI is much more likely to be made by someone who doesn’t know what they’re talking about, and I’m worried that the sheer volume will crowd out actually quality content like this.

New comment by bayesnet in "Why my Rust benchmarks were wrong, or how to use std::hint::black_box? (2022)"

bayesnet — Wed, 28 Jan 2026 04:03:14 +0000

> I have no idea what the mov rcx, rsp instruction is for. The rcx register is not read anywhere after that.

Anyone have any ideas? I’m surprised that the compiler emits a “useless” instruction at -o3.

New comment by bayesnet in "Text Is King"

bayesnet — Mon, 26 Jan 2026 23:24:19 +0000

Ok I just had to look this up. There’s a kernel of truth but I didn’t find evidence of genital mutilation: “Sometimes the boys were whipped or violently bumped on the boundary stones to make them remember”( “Beating the bounds”: https://en.wikipedia.org/wiki/Beating_the_bounds?wprov=sfti1 )

New comment by bayesnet in "Windows 11's Patch Tuesday nightmare gets worse"

bayesnet — Mon, 26 Jan 2026 16:03:05 +0000

Genuinely curious—what parts of Windows 11 do you like? I can’t find a single redeeming quality compared to W10, but admittedly I daily drive arch + macOS and only occasionally use my windows machine.

New comment by bayesnet in "Show HN: A small programming language where everything is pass-by-value"

bayesnet — Mon, 26 Jan 2026 03:41:52 +0000

IMHO this is both unnecessarily pedantic and not really quite right. Let’s say we accept the premise that “everything is a value” means reduction is impossible. But a value is just the result of reducing a term until it is irreducible (a normal form). So if there is no reduction there can’t really be values either—there is just “prose” (syntax) and you might as well read a book.

New comment by bayesnet in "Waiting for dawn in search: Search index, Google rulings and impact on Kagi"

bayesnet — Thu, 22 Jan 2026 12:00:05 +0000

It should actually be “Layer 3: Paid, ad-free, asbestos free, subscription-based search”. Come on. I don’t think it’s productive to make decisive and conspiratorial declarations on the future of Kagi search because they didn’t use the magic words you like. [ And since I know freediver is active here I want to state plainly that I would cancel immediately if there is so much as a hint of an ad in the Kagi results ;) ]

edit: h/t to https://xkcd.com/641

New comment by bayesnet in "eBay explicitly bans AI "buy for me" agents in user agreement update"

bayesnet — Thu, 22 Jan 2026 11:55:20 +0000

This is exactly how eBay bidding works now. Sniping still works because your satisfaction with the outcome of an auction isn’t just determined by “I got the item below my price ceiling” but by _how much_ below my price ceiling I got the item.

Early bids make you commit to matching other bidders’ exploratory bids. You lose out on the (naive) dream of a “great deal”. Sniping (without paid-for bot assistance) is a costless way of not revealing your ceiling until the last moment (and it commits you to actually sticking to your ceiling because there isn’t time to rebid later).

If everyone bid rationally, this wouldn’t matter, but it’s very easy to convince yourself that you can stomach bidding just a little more than your ceiling just to win the item. This cuts two ways: last-minute bids prevent this behavior from others while also stopping it in yourself.

New comment by bayesnet in "Brands upset Buy For Me is featuring their products on Amazon without permission"

bayesnet — Sun, 11 Jan 2026 05:43:36 +0000

But this is literally the point of MCF: third-party logistics for selling off-Amazon. All these brands chose not to sell on Amazon for one reason or another and yet, without explicit opt-in, were being surfaced on a marketplace they didn’t want to be on. If I send a video through gmail that doesn’t mean I want it on YouTube.

New comment by bayesnet in "Show HN: Tylax – A bidirectional LaTeX to Typst converter in Rust"

bayesnet — Fri, 09 Jan 2026 07:27:44 +0000

I haven’t tried this out yet but my gripe with pandoc is that it produces latex (and typst) that no human would ever write. It looks messy and is annoying to share with coauthors.

This is not to say that pandoc isn’t a fantastic tool.