fault v. error v. failure

In common language they can be synonyms, but often aren’t in places (like computing) where there’s a useful difference. A page fault isn’t an error; a handled error doesn’t result in a failure; fault-tolerant systems fail less often than their components fault.

NaN is one of the ways IEEE maths signals an erroneous calculation when the only outbound communication channel is the returned IEEE value. Reading the history and rationale of the floating-point standard will likely convince you this was and probably is a good compromise.

Because IEEE values often have hardware acceleration that is a significant factor in performance, languages rarely (if ever) choose to map all IEEE error values to language-specific error reporting (e.g. a thrown exception or error-tracking algebraic data type).

JS is hardly an outlier here.

I do think it probably violates the principle of least surprise, but I don’t think I understand the scale of “momentum” change that would need to take place across society (industry, academia, etc.) to align it with naive expectations.

https://www.destroyallsoftware.com/talks/wat is much better. This post is pretty low-effort.

Why is nothing ever done? How is there still something to add to a C compiler in 2026?

The C standard is routinely updated. Both of the compilers support multiple languages, which are also receiving new versions. Finally, even if the language was unambiguously defined and never changed, new in-compiler optimization strategies are still being discovered and implemented.

Also, every compiler I’ve ever used for every language I’ve ever used includes at least some features that are not part of the language standard/specification (if it exists), and those pieces are also refined, sometimes even with the goal of getting them into a later version of the language.

That said, if the code you are compiling was really C2010 conforming, it will almost certainly work in a compiler fork from more than a decade ago. Unfortunately, most projects and developers either don’t care to or don’t know how to keep their source code strictly conforming.

Sure, “LLMs and GPTs are just tools” is a potentially defensible take.

But, Copilot and Claude and Gemini and Apertus and every other model, “open weight” or not, that has been studied enough will (unknown the the user) sometimes generate output that infringes on the copyrights of the training data AND most, if not all, of the models with high-quality output involved human rights abuses as part of their training process (per Amnesty International).

Maybe, in the future, we’ll have models that are trained only on freely redistribute content without human rights abuses and without setting environmental goals back 10-15 YEARS. Maybe.

If that happens, I might be willing to use those tools myself, and I’ll certainly stomach their output as part of my won workflows. Until we have those ethical models, FUCK all of them and their output, they aren’t just tools; using there output is apologia for immoral acts and using the models is collaboration immoral people.

EDIT: Reworded for clarity: fails to generate output that will not infringe the copyright terms of the training data

As a Linux user, we might not, but most projects do. It’s actually hard to find a project that has a strong rejection of the plagiarism machines. I’ve been trying to rebuild my workflow without commits from them, and I’ve had to jump to a BSD and start using wip source builds and send compatibility patches for other things (luckily, I’m a developer, too). And, EVEN THEN there are some generative AI code that I haven’t been able to excise – getting an “AI”-free C compiler means using a fork from an old version.

It all makes me sad. I’ve written code for fun since I was in kindergarden; I don’t want to stop using computers, but FUCK all this unavoidable “AI” getting everywhere.

I can usually assuage the consumerists by implementing champagne socialism, but the authoritarians (and usually the accelerationsists and Malthusians) never get along with me.

But, depending which disasters hit and how quickly I have also been deposed several times, even once after restoring the climate to pre-industrial quality.

https://play.half.earth/ ?

Maybe also the price to install a charger at home?

I have owned an EV for 7+ years and do not have a home charger, but rarely charge anywhere else. You can charge them at 8A 120V overnight and still get more range than most people use most days. I mostly charge at 12A 120V.

If car manufacturers would make more affordable EVs I bet more people would happily get one when they need a new car

Maybe. The lack of BEVs (or even PHEVs) in the US market is certainly woeful.

Capitalism only exists because of the idea of profit, but there is no such thing. Pursuit of it leads to externalization of costs, denial of liability, deceit of the worker, and destruction of everything not owned and assigned a value.

Odd, I see one. They are using the inline image markdown. If you’d just like a URL it is: https://infosec.pub/pictrs/image/b18fe7f2-0c3c-4597-9bd5-fe100a6428e3.png (or at least that’s the one I see; maybe re-written by my instance).

I don’t think we have to be a monolith. But, I am personally glad when gas prices get closer to matching their environmental impact, even if that causes prices (of everything) to go up. I believe there are a LOT of things that can be done by the government to address any amount of “we’re all fucked” that is currently ameliorated by the price of fuel.

While I don’t know how it gets allocated (probably all into the general fund and then out to military contractors), U.S. fuel also has existing taxes that are supposed to go toward road maintenance. At least that’s the justification my state of Arkansas gives for charging me extra fees to register my Volt, which they treat as an EV: it doesn’t use (as much) fuel, but it does require (just as) maintained roadways.

This is why a carbon tax on fuel would help, and why ending fossil fuel subsidies would also help. Even both isn’t the end of government action, but they would both be good way to incentivize using alternatives.

Might I suggest the Working Families Party instead of forming a new one? Just make sure to take into account the realities of FPtP voting, if your district hasn’t implemented a better system.

This is NOT cute. Batteries only do this when they are in distress.

(Sorry if I got the wording wrong, I couldn’t find the copypasta.)

As an American, I was still a little self-shamed that it wasn’t until my second watch-through of “Ted Lasso” that I realized Jamie’s fan song is “Baby Shark” but with the titular phrase changed to “Jamie Tart”. :/

Pineapple Wellington, but with pizza dough instead of puff pastry? Or did you mean the rings (still cut, just not quite as much)?

Before “The Big Lie” was specifically Trump won the 2020 election, it was a strategy pursued by authoritarians, including Putin, the Kim family, Mussolini, and Hitler: https://en.wikipedia.org/wiki/Big_lie

The truth is whatever we say the truth is: do not trust your lying eyes. To speak otherwise is to betray Friend Computer.

Cory Doctrow has a much more articulate argument than me and I encourage you to read it.

But, no. Training is almost certainly fair use. It’s basically just consuming and doing statistics, the same actions are done everyday both locally and remotely for all kinds of services from search and beyond. While they were “in the lab” the outputs were not widely distributed and what distribution did occur didn’t impact any commercialization of any of the inputs.