@ekuber@hachyderm.io

@ekuber@hachyderm.io · Aug 06, 2026

Boosted by

Visiting CERN legit moved me. Huge peace-time undertaking to understand the Universe, away from petty squabbles. Physics are the same in Kazakhstan, Papua New Guinea and Bolivia. Our understanding is only obtained collaboratively. This place represents to me the best of us. Collaboration through science. Science through collaboration.

22

0

12

0

Esteban Küber

@ekuber@hachyderm.io · Aug 05, 2026

Boosted by

Trending Bot @trending@homestead.social

rust-lang/rust is adopting an LLM policy If you personally use LLMs and want to contribute to the Rust toolchain, you *really* should read it. The summary is "It's fine to use LLMs to answer questions, analyze, distill, refine, check, suggest, review. But not to *create*." Effectively, no other human should see LLM output, unless they ask for it, and LLM usage shall be disclosed. https://blog.rust-lang.org/inside-rust/2026/08/05/rust-langrust-is-adopting-an-llm-policy/ #RustLang #Rust

343

13

265

12

Esteban Küber

@ekuber@hachyderm.io · Jul 30, 2026

There are people who are good at rage baiting. People who enjoy it. People who make money off of it. These circles in the Venn diagram don't necessary completely overlap. But when you see a bonehead take/article that really riles you up, they are all three. They are not true believers, they are monetizing looking like an idiot both by taking money from idiots they pander to but, more importantly, by taking money from companies buying eyeballs.

8

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 26, 2026

RE: @GossiTheDog@cyberplace.social

But it wasn't a Seig Heil... 🙄
Fuck Elon Musk and his explicitly fascist agenda.

cyberplace.social

Kevin Beaumont: "Elon Musk, an immigrant, was interviewed recently…" - Cyberplace

8

1

3

0

Esteban Küber

@ekuber@hachyderm.io · Jul 23, 2026

Replying to @emilyskidsister@hachyderm.io

@emilyskidsister@hachyderm.io some years back I saw people who think that memory safety is bad because it enables hard to break trusted computing platforms, and that's bad because you then can't easily jailbreak your hardware to install your own software. I think they've identified a real problem (you should be able to control the software on hardware you own) and reached a conclusion I find untenable (software should be purposely vulnerable to allow tinkering). Since then I'm not really surprised by any takes.

6

0

Esteban Küber

@ekuber@hachyderm.io · Jul 23, 2026

Replying to @ekuber@hachyderm.io

FWIW, no one owes me anything, and me seeing the equivalent of two people shooting the shit at a bar between each other should be none of my business. But social media *isn't* a bar, and one can have a much wider audience than one thought, and reach people you wouldn't say that to their faces. Specially when the bar has 10 thousand of your closest friends, your pithy comment will escape the room.

12

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 23, 2026

Boosted by

Trending Bot @trending@homestead.social

I've seen several people express dislike towards Rust (the project or the ecosystem) due to being opposed to the use of LLMs and a perception that Rust is used by LLM users. It feels deeply unfair having spent the last decade trying to make the Rust compiler talk to people as humans, having that work be incidentally useful to LLMs, and that somehow reflecting poorly on the project. It's like if someone looked at the textual TTS optimized output we've been working on to improve accessibility, and criticized it because it could make it slightly easier on LLMs too.

122

10

63

0

Esteban Küber

@ekuber@hachyderm.io · Jul 22, 2026

Replying to @cwebber@social.coop

@cwebber@social.coop Just one nitpick: pretty much every GCd language has mechanisms to interact with pointers and muck around with types. The same capabilities from Rust's unsafe are generally available (except for fully sandboxed VMs, but that's not a property of the language), and depending on the prevalence of native modules in an ecosystem, their use can be quite extensive. Go and Java tend to reimplement natively, Python tends to FFI, for example. This is not to discount the benefits of a GC, but rather to point out that in general people haven't argued that those languages *aren't* memory safe despite the existence of torn reads or specialized APIs that can cause memory safety. Something like invisicaps/Fil-C or WASM sandboxing of an application and its dependencies would be a more complete solution than what most runtimes provide, of course.

5

0

2

0

Esteban Küber

@ekuber@hachyderm.io · Jul 21, 2026

#RustLang pro-tip: When wanting to inspect the value of a binding in a let-chain, you can take advantage of blocks being expressions as well and add one in the middle of the chain if let Some(next) = iter.peek() && { tracing::info!(?next); true } && let Some(...) = ... This is less disruptive than having to disentangle the chain every time you want to inspect something in the middle #Rust

12

0

7

0

Esteban Küber

@ekuber@hachyderm.io · Jul 17, 2026

Forgive me, Father, for I have posted.

11

0

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 14, 2026

Thanks to @muscraft@hachyderm.io, annotate-snippets-rs might have a new "no-graphics" mode in the short term (and in turn, add that support to rustc). We are still iterating on the specifics, but I would like to hear what people's reaction to the current state is. #Rust #RustLang #accessibility #a11y #programming

0

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 14, 2026

In my opinion, when it comes to level of maturity/evolution of these languages K&R C ≈ Java 1.1.1 ≈ Rust 0.x ANSI C/C89 ≈ Java 6 ≈ Rust 2015 edition C99 ≈ Java 8 ≈ Rust 2021 edition #Rust #RustLang #Java #C

7

0

Esteban Küber

@ekuber@hachyderm.io · Jul 13, 2026

The deep irony of all the time spent on making #RustLang easier for humans to use resulting on directly benefiting LLMs... #Rust

4

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 08, 2026

Replying to @ekuber@hachyderm.io

An article detailing that day from the point of view of the journalist and photographer, with more pictures: https://www.reuters.com/investigates/special-report/usa-250-washingtondc-picture/

2

0

3

0

Esteban Küber

@ekuber@hachyderm.io · Jul 05, 2026

Replying to @cthos@mastodon.cthos.dev

@cthos@mastodon.cthos.dev Misremembered. Thank you for the correction.

1

0

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Replying to @haferlasche@digitalcourage.social

@haferlasche@digitalcourage.social

23

0

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Replying to @ekuber@hachyderm.io

48

4

30

0

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Boosted by

Vee @VeroniqueB99@mastodon.social

Replying to @ekuber@hachyderm.io

79

14

72

1

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Replying to @ekuber@hachyderm.io

49

1

28

0

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Replying to @ekuber@hachyderm.io

77

2

40

3

Esteban Küber

@ekuber@hachyderm.io · Jul 04, 2026

Boosted by

Trending Bot @trending@homestead.social

This is a hell of a picture

503

58

432

8

Esteban Küber

@ekuber@hachyderm.io · Jun 30, 2026

Replying to @ekuber@hachyderm.io

Sometimes I feel like the people who most dislike strong type systems, safety features or what have you that restricts a subset of behaviors are people who just fundamentally hate being told "no". It doesn't matter if it is "no, but you can do this other thing".

22

0

8

0

Esteban Küber

@ekuber@hachyderm.io · Jun 30, 2026

"I don't like the compiler yelling at me." My sister in RISC, that's the compiler's main job.

77

5

29

0

Esteban Küber

@ekuber@hachyderm.io · Jun 25, 2026

Replying to @mttaggart@infosec.exchange

@mttaggart@infosec.exchange Open source projects tend to be (and Rust certainly is) a showupocracy. Shit gets done when people that care about that shit does it. This means that important stuff that everyone agrees is important but not important enough for *me* to do, doesn't get done. And that some things end up being 80% solutions that scratch the itch of the person driving a project and progress stalls beyond that.

3

0

Esteban Küber

@ekuber@hachyderm.io · Jun 23, 2026

I'm very interested in systemic solutions to problems. Blaming an individual when they make a mistake *feels* reasonable. When a significant percentage of people make the same mistake (70%, 2%, it doesn't matter as long as it is measurable, predictable and has a negative impact), then it's *not* a persons fault. We know how to tackle those kind of problems: we change the environment to minimize the likelihood of mistakes. Traffic and pedestrian safety is an example of a well explored space where we know what works. Bollards, narrower streets, slower speeds, raised pedestrian crossings, providing good alternatives to driving to reduce traffic... Similar in industrial spaces: clear signaling, segregate heavy machinery from human limbs, clear and practiced process to get out of dodge when things go poorly... It's always about looking for solutions to eliminate mistakes, and when they can't be eliminated, minimize the blast radius. This is why I like #Rust: it's a language and tooling that applies the industrial approach of "if two people make this mistake, figure out how to mitigate it". We can't "individual responsibility" our way out of systemic problems. I want more software with bollards.

29

0

18

0

Esteban Küber

@ekuber@hachyderm.io · Jun 21, 2026

Boosted by

Trending Bot @trending@homestead.social

For any project that relies on collaboration, it doesn't matter how smart an individual is if they drive other people away. The underlying technical points might have a lot of merit, but I would not want to use a tool that would have me interact with people like that that would berate me for filing a bug. I would certainly not want to contribute code to such a project. I am amazed how many people confuse "being a dick" with "having a personality". I hate the "asshole genius" trope because all it's done is convince assholes that they are geniuses, and non-geniuses to think the way to appear smart is to be an asshole. If you can't communicate effectively ("play well with others"), you're not nearly as smart as you think you are. Smart people know how to get shit done, and the only way to get shit done at a scale beyond what any given human can do on their own is collaboration. And no one wants to collaborate with an asshole.

27

0

22

1

Esteban Küber

@ekuber@hachyderm.io · May 29, 2026

"Everyone is stupid. I've thought about this problem for 30 seconds and this is the obvious solution. Why didn't they do that? They must be stupid. They haven't even talked about the problem! They must be stupid."

The obvious solution doesn't work.
The problem was discussed to death a decade ago, and every few months since.
You haven't seen a recent discussion because the last 3 discussions ended 3 comments in after linking to the previous ones.
Don't come into a room yelling and swinging your ego unless you've bothered to look into the problem more than 30 seconds if you don't want to be confronted with your own ignorance, lest you are forced to reassess your self image.

33

7

5

0

Esteban Küber

@ekuber@hachyderm.io · May 27, 2026

Learned about the Frequency Illusion, and now I'm seeing it everywhere.

23

4

9

0

Esteban Küber

@ekuber@hachyderm.io · May 12, 2026

Replying to @lcamtuf@infosec.exchange

@lcamtuf@infosec.exchange @2something@transfem.social https://github.com/rust-lang/rust/issues/120426 The discussion in the tracking issue gives an idea of why the std didn't ship with it in 1.0. I hope they can be resolved quickly, but I would lean on https://docs.rs/crate/cap-std/latest instead.

0

Esteban Küber

@ekuber@hachyderm.io · May 12, 2026

Replying to @soph@grrl.me

@soph@grrl.me It is exhausting that all of a sudden some people care about documentation and diagnostics not because it is good for humans/workers, but rather to aid their automatons. I hope that the effort put to make it easier for LLMs is done in such a way that it *also* aids humans. But I'm not holding my breath.

1

0

1

0

Esteban Küber

@ekuber@hachyderm.io · May 04, 2026

Replying to @ekuber@hachyderm.io

"May the project you started for fun become popular" and other ancient curses.

32

2

3

0

Esteban Küber

@ekuber@hachyderm.io · May 04, 2026

Every now I then I see a rhetorical "Who asked for this? Why does this project exist?" where the OP clearly doesn't think the project should even *exist* and it always makes me wonder if I just didn't receive the memo of the list of people we need to ask for permission first before opening a text editor and start writing code.

23

6

0

Esteban Küber

@ekuber@hachyderm.io · May 04, 2026

Replying to @AdrianVovk@fosstodon.org

@AdrianVovk @zeenix @pid_eins ignoring the problems with and current limitations of the current Rust std APIs is bad, as it doesn't leave an open path to improve them. Ignoring the real limitations for why those APIs haven't yet been changed, given that Rust's std must support platforms that don't have the necessary primitives, doesn't help improve things either. There are crates that implement the desired APIs, supporting only Windows, modern Linux and macOS, and that's what most people should likely be using. A crate can also make the decision to support platforms on a best effort manner and "lie" about what the underlying platform supports (like the win9x Rust port does), but that is not suitable for the std.

6

0

Esteban Küber

@ekuber@hachyderm.io · May 03, 2026

Replying to @lizardbill@hachyderm.io

@lizardbill compilers must be written for providing feedback to deverlopers first, and only incidentally for generating object code. Or, as I said a couple of years ago: "Compilers are error reporting tools, with a code generation side-gig." 😉

3

0

Esteban Küber

@ekuber@hachyderm.io · Apr 28, 2026

"Fuck you, I've got mine" is not only sad as a position in general, it is also a bad tactical choice because over long enough timeframes you can't assure that you can *keep* yours if others are deprived.

21

0

5

0

Esteban Küber

@ekuber@hachyderm.io · Apr 21, 2026

Replying to @HelenBranswell@scicomm.xyz

@HelenBranswell@scicomm.xyz When I was younger, when I thought there were only a handful of them out there, I used to find antivaxxers as amusingly naive. After COVID I realize now that they are aggressively dangerous.

0

2

0

Esteban Küber

@ekuber@hachyderm.io · Apr 19, 2026

7

1

0

Esteban Küber

@ekuber@hachyderm.io · Apr 18, 2026

Replying to @jam@tilde.zone

@jam I've gone out of my way to add finishing touches to pieces in places that no person is ever going to see unless they dissasemble it. The underside of a chair, the inside of an opaque computer case, the behind of a picture frame. And I wouldn't have it any other way.

5

0

Esteban Küber

@ekuber@hachyderm.io · Apr 18, 2026

Replying to @ekuber@hachyderm.io

I think that a lot of the talks about art and LLMs shows that there's no consensus on what "art" is (shocking!—I know). Some of us think it is a process. Others that it is a result.

9

5

0

Esteban Küber

@ekuber@hachyderm.io · Apr 18, 2026

Replying to @ekuber@hachyderm.io

I guess the point being made in the original (alluded) thread is that "thinking about the problem and writing the code to apply it" is fundamentally the same creative process as "thinking about the problem and writing a prompt for an LLM", and the main difference is that a lot of the final shape of a widget is the result of cascading problem solving that only happens in the process of doing. The constraints and how we interpret them changes the result. It's Schroedinger's innovation, you can't tell if you need to change things in your original thoughts until you attempt to apply them. If the problem space is constrained enough, sure, you'll end up with one of a handful of options (there's only so many ways of splitting a string) and that process or the result is not creative. But that's focusing on the screw and not the chair.

18

1

0

Esteban Küber

@ekuber@hachyderm.io · Apr 18, 2026

> My code is not intended to evoke emotion. It does not change people think about the world.

The actual *code* I write for rustc is unimportant and shitty, but the *result*... it absolutely is meant to evoke an emotion (cognition and relief) and change how people think (I try to teach them a programming language!). Even APIs that I write I try to make them so misuse is unlikely, which in turn teaches people how to use them correctly. I am not an artist, just some moron with a tool. And I won't claim some high minded goals. But there's absolutely artistry involved. It's impossible for there not to be when people are involved. A string concatenation function isn't going to be hung in the Louvre, a lot of code is functional widgets. Very rarely anyone thinks about the "artistry" of a screw or a nail, but one absolutely thinks about the artistry of a chair. It might be that the perspective of someone who works on an end user application is different to that of someone working on a standard library, but even in the latter I'd say there's a lot to think about and innovate.

It's like discounting the artistry of a wedding photographer: most of the job is mechanical and repeatable, but the person behind the shutter is doing something beyond pattern matching, even if unconsciously! They are reasoning about consequences, and problem solving on the spot.

63

10

29

0

Esteban Küber

@ekuber@hachyderm.io · Apr 15, 2026

Replying to @wffl@im-in.space

@wffl @rustnl congratulations!

3

1

0

Esteban Küber

@ekuber@hachyderm.io · Apr 08, 2026

Replying to @skinnylatte@hachyderm.io

@skinnylatte foiled by my mediocrity once again

1

0

Esteban Küber

@ekuber@hachyderm.io · Mar 31, 2026

Replying to @sgued@pouet.chapril.org

@sgued@pouet.chapril.org looking at https://github.com/trussed-dev/littlefs2-sys/blob/45e25a4dfd381d96ccfc3712bab9ad3630cdc57e/build.rs I can't really suggest a better alternative because as you say littlefs2 is not something that your system's package manager is gonna pull in. We need to improve our tooling to programmatically tag non-rust dependencies. A git submodule as you use it at least gives a chance to analyze what you're doing and compare that against known issues.

0

Esteban Küber

@ekuber@hachyderm.io · Mar 31, 2026

Replying to @piecritic@hachyderm.io

@piecritic@hachyderm.io this is a lack in cargo, yes. The path forward is to cater to the use-case.

1

0

Esteban Küber

@ekuber@hachyderm.io · Mar 31, 2026

Replying to @ekuber@hachyderm.io

#Rust crates that vendor C code and build it in their build.rs should *not* make them the default. Instead they should attempt to use the system one and bail if that fails, with an error asking people to enable a feature flag to use the vendored dependency. "It just works" invisibly means that people can be unaware that they are including vendored code or that their platform's system packages aren't properly set up. #RustLang

16

5

0

Esteban Küber

@ekuber@hachyderm.io · Mar 31, 2026

Another go-around of the discourse around dependency counts.

Vendoring is a terrible idea, it gets in the way of automated dependency scanners that would help you find out about vulnerabilities. Having to vendor is a bug report to you package manager, as it clearly is lacking some feature.

Counting 100 100 line long crates all by the same authors as inherently more dangerous than 1 10000 line long crate makes no sense to me.

12

2

4

0

Esteban Küber

@ekuber@hachyderm.io · Mar 20, 2026

Replying to @YesJustWolf@hachyderm.io

@YesJustWolf I try my best to always stop myself when I want to say "I dislike X" and reframe it as "I like Y". Talking about why something is nice, even if you still mention X to compare and contrast is much more productive and results in nicer (if shorter) conversations and a much Kore reasoned thought. Its easy to say "this is shit", it's more effort to go into the nuance for why the tradeoffs of your preferred choice are reasonable.

0

Esteban Küber

@ekuber@hachyderm.io · Mar 20, 2026

Replying to @benjamineskola@hachyderm.io

@benjamineskola @YesJustWolf I'm *some* cases it is rage farming: they know how hashtags work, but want to troll the people who follow them

0

Esteban Küber

@ekuber@hachyderm.io · Mar 16, 2026

Replying to @timmc@better.boston

@timmc @SnoopJ @cliffle @bitprophet If nothing else, emoji have been instrumental in making proper international script support close to ubiquitous by forcing people who deal with text that users want to read to put in at least a half assed effort into supporting most corners of Unicode. For that reason alone I find it hard to get mad about them.

9

2

3

0

Esteban Küber

Posts

Kevin Beaumont: "Elon Musk, an immigrant, was interviewed recently…" - Cyberplace

Remote instance

Media