Anyway, as the guy who opened up the issue proposing that KDE explicitly allow slop PRs quite validly points out, it's not really up to anyone but KDE contributors. That's fair.
But what the same guy is a thousand percent communicating though, is that KDE is not for people like me. It's not for people who have something to lose to fascists. If KDE developers don't want to stand for KDE's users, there's no obligation for them to do so, but let's not mince words about it either.
This entry was edited (Monday, September 21, 2026, 11:09 PM)
This is not academic to me. It's very analogous to the situation with Calibre — not identical, but analogous. As lead developer of Calibre, Kovid had zero obligation to his users. He was *and is* free to betray Calibre users with impunity.
I don't hold his choice to do so in high regards, however, nor do I hold the guy opening that issue on KDE's AI policy in any kind of particular respect.
I hope that the rest of the KDE crowd tells him to take a hike and stop making things worse. We'll see.
I don't fully agree he has no obligation to his users.
He has no obligation to serve them, he doesn't work for them. He has no obligation to provide software to any standard, or to make decisions the way they want him to.
But generally we all have an obligation not to harm other people. The more direct and significant the harm, the greater the obligation to avoid the harm. Your right to swing your arms ends at my face.
What level of betrayal is acceptable? Not all levels.
@glyph @whitequark Perhaps I was being glib to elide that gnarly corner, but I meant something closer to "I cannot reasonably expect to compel said behavior through any means that looks roughly legal-ish, including project codes of conduct through to actual laws."
In a moral sense, though, yes, I agree, obligated.
@whitequark The jist was that, if the justification for having trademarks is so that consumers can be confident that a brand label or product name means what its reputation has come to mean, then the "owner" of that name repurposing it to mean something else in order to deceive consumers should be just as much a violation as some rando showing up making "counterfeit" products.
@dalias @whitequark Amazon did its best to ruin the concept of trademarks too.
To supposedly curb the absolute junk that gets peddled on their site, they insisted that every product *must* have a trademarked brand name. Which is how we ended up with Heqqqqqqi brand Android tablets that are carbon copies of Tuukwww brand Android tablets.
Great job there, Amazon. You really showed those low-effort crap resellers.
@dalias @whitequark This is why I prefer ordering from AliExpress, they make no effort to disguise that they're selling a buttload of garbage and scams with a few good things mixed in (that aren't actually all that hard to pick out of the muck once you're familiar with it). And they're cheaper.
[Amazon requires trademarked names and jibberish company names are the result]
This explains much as there is just so many unique meaningful or seemingly meaningful names for companies (or automobile models to cite a cognate) that could exist. However, I'd love a reference link or a publication name—or some info on how you discovered this very believable factoid.
@sfwrtr Honestly, I don't remember the article I read it in, but it was years ago, shortly after Amazon made that change and all of those random letter brands started showing up. It'll be buried in their enormous set of requirements and terms which vary from country to country and it only takes one reasonably large market with that requirement for that effect to occur.
@sfwrtr I did a quick search and this was the closest I could find: - identical products by different sellers result in combined listings (encouraging superficial distinctions like random brand names) - almost every product requires a GTIN (could be difficult to track down for unbranded electronics bought directly the factories, and you might simply opt to register your own brand for them) - Amazon's "IP Accelerator" tool that makes registering trademarks just a few clicks away.
What I gleaned is that you need a unique store name, however it is greatly encouraged that you register a trademark. Whilst it is practically a difference without distinction, in practical use getting that trademark prevents another seller from poaching it despite you using the name. In any case, a trademark doesn't appear to be a requisite, but without one you have to jump through an additional hoop of getting your store name okayed by someone at Amazon. At least that's what the doc seems to state; I've never become a seller to find out how it works IRL.
Regardless, the result is the same: lots of gibberish store names! Yes, it's annoying to search for something like a heating blanket and find the exact same product sold and branded by dozens of other stores, probably sold generically by one wholesaler and thus with the exact same specs and flaws. The result? Fifty different options, but maybe only three different competing products.
Thanks for posting this, however. Good food for thought.
@sfwrtr I'm not talking about the store name, but the products themselves. You can sell products with existing brand names all you want, as I understand it, the GTIN will be supplied to you by the manufacturer or distributor. But if you're reselling generic stock that may not have passed through totally legitimate channels, the GTIN might not be available. Since you're allowed to manufacture your own stuff and get a GTIN registered for it, you just *pretend* you made it and give it a brand name.
@StarkRG Labor is the most expensive part of most businesses. Too many sellers, and they don't want to add pennies to the cost of millions and millions of products.
[Sarcastic:] Maybe they could teach Rufus to catch these things. Finally, a worthwhile task for AI.
@dalias @whitequark Not entirely sure trademarks is the right comparison point for this concept though. I think I'd more compare it to takeovers, and we've had quite a lot of those in foss.
@dalias licensing is why we see so many crap products which are branded with the names of long dead companies which were known for their high quality products
Sorry, I'm a bit out of the loop- what exactly changed with calibre?
Searching around I'm seeing some stuff about an LLM 'discuss the book' (which...weird and crazy, but at least it seems to default to off?), but I am wondering if I missed something worse.
@Oggie In Calibre, you can right click on a book and "discuss selected book with AI."
In reailty though, it's just a completely bare-bones chat interface that can't even handle being resized correctly, and the only thing it does differently from any regular chatbot interface is it first sends a message saying something like, "I am currently reading the book Project Hail Mary, and I would like to ask some questions about it."
Even if you love all things AI, it's just incredibly bad.
@hadley @Oggie It's more than that. The latest version of Calibre has a "Write me a slop-story" feature. I can tell you from experience that, once a developer catches the slop-virus, they don't stop...and Kovid clearly has it bad.
"I'm fine with poisoning minority communities and funding a guy calling for ethnic cleansing, but I draw the line at arguments infused with human emotion."
@aud Mind you, they get to be as emotional as they want. Because they're *correct*, you see, not like all those whiny queers trying to say let's not do a fascism.
"Part of what we care about in KDE is providing an answer to what makes software and its development paradigm socially responsible."
me: "yes!"
me after reading more: "... jesus christ, he's arguing FOR the fucking LLMs."
How can you... possibly... POSSIBLY... use the words "socially responsible" in your *pro* comment. Jesus christ, it is clear that guy has no idea what that means.
newsflash: adding "poisoning minorities", "destroying collected knowledge", and "funding fascism" to your developer workflow *isn't socially responsible*
my brain has hit the "don't post something you'll regret" point so... which is kind of impressive when you consider what I *do* post, so if that tells you anything.
This is a continuation of this mailing list thread about "fully LLM-generated merge requests, preserved at https://mail.kde.org/pipermail/kde-devel/2026-September/004496.html There, I proposed...
I did find the *actual nazi* in the thread. Someone found their twitter account that is just straight up full of fash trash and they kept defending themselves with "I have been nothing but polite in this thread"
thing is... as a temperature check, I don't love that he seemed to feel comfortable sticking around after that was exposed.
I wish I could hop onto that thread and be like "hey has anyone noticed that actual mask off nazis agree with the LLM policy, do you want to maybe think about why that is"
This is a continuation of this mailing list thread about "fully LLM-generated merge requests, preserved at https://mail.kde.org/pipermail/kde-devel/2026-September/004496.html There, I proposed...
@aud @cthos once again the "obviously terribly moderated" mastodon.social had already banned this person before the reply in question. (not saying anyone should move here, just feel like I need to call out the mod team's invisible wins since they get so much shit for their mistakes)
@glyph @aud @cthos Weirdly, wandering.shop didn't have him banned, but I already had him blocked with "annoying mansplainer" as my note to self, so I wasn't able to see much of any of the thread.
looks like they went one step further a couple of hours ago and not only deleted the posts quoting the nazi accounts twitter, they also blocked the account of the person who unmasked said "nothing but polite" nazi in that thread. great show of solidarity there, @kde
a fascist is never polite, by definition. They may be using "polite speech", but they are using polite speech specifically to provide cover for their hateful speech and actions. Much how like an anti-Black racist isn't "a good person" for being polite when they're talking to white people. That "politeness" is intimately linked to their bigotry.
this p much now looks like kde has no issue with fascists in their midst while actively removing posts exposing them, so yeah that thread is a complete disaster but not for the reasons nate thinks
I only have one PR pending merge to KWin, but I've gone ahead and asked that it be put on pause until this issue is resolved as I will not contribute to a project that allows this.
It's a small PR and a small change. My principles, unfortunately, are not at all small and I won't compromise them.
hell yeah, hell goddamn yeah! Yesterday evening, when I looked at the discussion, the comments were still there, didn't know they aren't at first, glad they're archived though. I don't understand much of what was above each link, because the words were censored weirdly and I didn't want to open that horrible thing called twitter, but it's good people can still see them.
bubububut we were told by 3 different guys that we have to allow for nuance or else we're a wokescold doing terminally online purity testing, surely these people aren't in bed with fascists! /s
@cthos Absolutely. But that "I'm taking my ball and going home" move at the end was priceless. It's his wont, of course, he should be free to walk away. It's still priceless, tho.
So sick of this manufacturing of consent. If you're gonna bend the knee and lick the boot, just fucking do it. Don't pretend to care about people's opinions and then just lock threads when you find out that most people don't quite have the same appreciation of shoe leather.
Just earlier this morning, I saw someone with some profile aesthetic signals of queer outcasthood making what was very likely a thinly-veiled post about me/my reporting (if you wanna elevate it to "reporting") on the KDE situation. How the stuff about the Akademy talks was overblown, by "the most annoying parts of the Anti-AI people".
I blocked them because like... if that's the quality of a person's judgment, I'd like less of their takes in my life, as a protection of my personal peace and sanity. But I keep finding myself wondering: "how's that feel now? Now that we know it was early steps to manufacture consent? That some KDE developers got called out for some shocking Twitter racism in that thread? That KDE really is presenting itself lubed up and ready for the fascist stereotype-repeater machine, and don't care if it makes the project hostile to minorities on the dev team and in the userbase?"
I dunno how a person with those traits feels stakeless in this. I don't know.
The funny thing about this is I was pretty clear that I wasn't jumping to conclusions and assuming the worst, just that KDE was raising questions that we as a community needed to demand answers to, because the worst-case answers were really bad, and the signals raising these questions were really alarming. I actually advised some other people to slow their roll, in case it was just like... one or two Akademy organizers who didn't speak for the dev team. I try to always assume the best I can of people, even if it puts me on the back foot in situations where I later learn I'm wrong.
@onepict Yeah. Thankfully I already use a very stable, pinnable tiling WM already, and just have KDE as my backup full-featured DE. This mostly hits me in the programs - I frequently use Kdenlive for editing, even for simple stuff like taking snippets of that Scott Jenson talk recently.
As much as I've enjoyed CachyOS, I do think I'll probably switch to AntiX for my Linux needs in the future, as an actively anti-fascist distro with good ergonomics in non-slop lite X11 WMs, and try to move most of my workflow to multiple flavors of Not Linux (probably partitions for GhostBSD, 9front, and Haiku). You know. When I somehow invent a zero-point module to pull the energy to do this out of the friggin' aether.
@MaddieM4 I saw that too and it felt very off to me as well, I’m impressed by your willing to call it out — you can’t brush off wanting to stop fascists taking control over the computing landscape we depend upon with FLOSS insularity and pearl-clutching.
Very specific KDE people are right now trying to force theough a change that violates every FLOSS principle, moral and legal, and they’re trying to say they hate all this misplaced anger. My dear, my anger is placed squarely where it belongs — in the face of AI pushing fascist fucks who know _exactly_ what they’re doing and use “civil discussion only” to enable themselves at every step
@ell1e @Kroc @MaddieM4 I have a feeling for the next few years, our approach to software is pretty much going to have to be: pin a distro/package versions from 2025, preferably one with LTS security updates, and occasionally pull newer packages or self-builds-from-aports/whatever of specific things we need.
@dalias @ell1e @Kroc I think that's going to be a really common path, as well as people trying more obscure slopless operating systems where its viable for their individual needs (see also: there is no universal Year of the Linux Desktop, only personal ones).
As for the way we do non-slopware, with some lessons learned... well I'll just speak for myself as a dev, and maybe other people will resonate with it. I talk about my far-off plans for developing and maintaining an OS, and people have gotten to watch the shape of those plans shape and refine over time. I once wanted to make a Linux distro around the Layover package manager, so, Layover Linux. Eventually I decided I wanted to not be Linux based, and I had some microkernel aspirations, and that was called Mercury Microkernel, and I thought maybe the desktop should be named after that - Mercury. But now I think, let these pieces have their own names. "LongevityOS" is maybe the name that ties it all into the right mission statement.
@dalias @ell1e @Kroc I dunno. I think I'm just really, very permanently "over" software that needs or expects regular updates, or especially software that feels entitled to push new versions onto your device using features or security as justifications.
Paradoxically, software that doesn't require updates later and can be securely and productively old-as-dirt feels like a flashy and exciting idea to me right now. Especially light stuff that runs on a toaster.
And they key to having that is to stop having attack surface.
The reason our phones supposedly need constant updates is because they come with a non-removable SMS client (that's grown to use new channels outside just SMS) that accepts media attachments from any random sender and attempts to decode them, and because they come with a non-removable browser that insists on letting random sites run code in a YOLO unsafe JIT engine for ZOMGFAST reasons.
If your stuff is not attack surface, it doesn't need updates. 💡
This entry was edited (Tuesday, September 22, 2026, 5:25 PM)
@dalias @MaddieM4 @Kroc Firefox still doesn't allow running WASM without JIT being enabled. Kind of frustrating. I agree all browsers should ship with JIT disabled. (Also WebGL/WebGPU and high performance timer, if you ask me.)
I'd love to see some opt-in "high end game" site permission...
My wife is obviously going through a lot of the same tired exasperation as the rest of us, just less terminally online than me. She's looking at getting a Linux laptop set up that she just never activates any internet connection for after installing a set of packages she cares about. Without that connection (or bluetooth etc.) as an attack surface, you actually can just enjoy a pre-slop Linux distro release as a stable appliance indefinitely.
@MaddieM4 @ell1e @Kroc Seems like a solid choice, more extreme than needed, but fair if it's for peace of mind. On a properly configured Linux install, just being connected to the internet should not be attack surface.
@MaddieM4 @dalias @ell1e As much as I know that we technically skilled users, hackers and tinkerers can solve this for ourselves, the lacunae surrounding hardware support and how friends and family are supposed to be included in this is too much to polyfill by ourselves. KDE is a well funded organisation, now with EU government backing
I'm all for slow software, but at some point it has to come head to head with government / banking apps that literally won't run on anything without a locked bootloader and Google spyware on it.
Our digital space has been occupied by aggressors. I'm of the opinion that we should have a French-resistance attitude to what we make rather than trying to outcompete them. We won't be able to live without slop in our lives until the bubble bursts. Until then, software vigilantism is the way to protect ourselves and to fight the bastards.
@dalias @MaddieM4 @ell1e That will work right up until the point you're faced with starvation. Disabled people and people on benefits etc. don't get the choice to not use government apps. That's not a fight that can be won by a boycott
@Kroc @MaddieM4 @ell1e Surely there are people who need these services who can't afford phones that can run them, or who can't retain a phone (abusive partners, camp raids, etc.), or who can't figure out how to use a phone, etc.
If there really is a problem of governments trying to force people to use apps that need DRM phones, maybe public libraries should have burner phones where you can go in use it to login and do whatever they're making you do, with assistance if needed.
you would think the government would keep that in mind.
but it currently looks less and less like it... otherwise, why are banks allowed to require an app for any sort of usable electronic access? the EU doesn't really seem to care
(I use a Linux phone and a bank that offers QrTAN. But many banks now appear to require either some custom weirdo USB driver and/or an Android/iOS app.)
This entry was edited (Tuesday, September 22, 2026, 5:32 PM)
@MaddieM4 @dalias @ell1e @Kroc The "Move fast and break things" techbros have no way to comprehend or interact with the thought of you moving at tectonic speed and fixing shit. Good.
@BinChicken @MaddieM4 @dalias @Kroc one of the things that seem to shock tech people I talk to the most, is 1. I turn off JIT/WebGL/WebGPU for browsers, 2. I turn off hyper threading and I run with mitigations=full, and 3. I like to develop software on devices with 16GB RAM or less. Sometimes I use a Raspberry Pi 5 as my work desktop during the summer.
Why did everything need to be a bit faster and a bit more 3D and give us all these problems? I don't get it. We really don't need half the features that browsers added.
if this comes to fruition I'd love to pitch in. I've worked at the application layer my whole career but am starting to learn about how compilers and OSes work out of necessity
@wbrawner @dalias @ell1e @Kroc Nice! While I do really want to have rump kernel compatibility to share code with NetBSD (I think freeing drivers from specific kernels is literally vital to kernel diversity) and that will help a lot for the hardware support side, there is a *lot* to do in the userspace, at a scope where I am just not enough people for the job. People like you will be critical to making something usable and human-friendly. I know I'll need a good graphic designer at some point, too.
Having a compiler expert in particular would be a big deal. I'm currently learning a lot by implementing and designing a programming language, but it cheats some of the hard stuff by being an interpreted language that transpiles to C source code. I am very intimidated by the idea of writing a C compiler, and that's just a thing that's gonna be needed at some point.
In case you haven't seen it, libfirm.github.io/ is somewhat-abandoned (plus side: that means no AI risk) but has a C compiler that's fairly well architected and small/simple. I'd love to see someone pick it back up.
@ell1e @MaddieM4 @wbrawner @dalias With respect to [All of the Above], there is already plenty of anti/no-Ai compilers like Zig (which compiles C) and OSes like Haiku / Redox.
If you want to rebuild the universe from first principles you will end up only solve theoretical problems rather than real ones.
still, a fresh, lightweight, independent no-AI C compiler that supports C99 with the lower-hanging fruit from C11 (that would be e.g. atomic ints that seem to be pretty essential for ARM64 targets, since there a 64bit int doesn't seem guaranteed atomic by the architecture) would be nice.
@etchedpixels @MaddieM4 @wbrawner @ell1e @Kroc Given how many bugs the mainstream compilers have had over the years misinterpreting the language specification, I question the claim that C compilers are easy to write. Unless you consider correctness as part of a stretch goal of "good" rather than just a baseline requirement.
"C89 without bitfields" is not a C compiler. A very minimal baseline is full C99 (at least long long, designated initializers, and compound literals are essential), and to be serious we should be aiming for C23.
is C23 common? I rarely see any of the commonly used libraries use much more than C99 with at best C11 atomics. even C11 threading I rarely ever see used
This entry was edited (Tuesday, September 22, 2026, 8:14 PM)
@ell1e @etchedpixels @MaddieM4 @wbrawner @Kroc No, not yet, but if we're talking about a future that's not a dead-end, we should be *aiming for* (I used those words specifically) C23 support.
Otherwise what's bound to happen is a few years down the line, there's good software folks want to use, but it's written in C23, and now they're stuck getting another compiler for that. 🤦
@ell1e @etchedpixels @MaddieM4 @wbrawner @Kroc BTW I generally consider C11 not terribly useful, but someone did remind me recently that C11 is where we finally got anonymous structs and unions standardized. These are needed to get rid of some really nasty old macro hacks for "struct members that share the same storage".
@dalias @etchedpixels @MaddieM4 @wbrawner @Kroc interesting! i didn't know that. i guess gcc must have simply provided them as a non-standard extension for C99 code as well, before C11 came around.
That's largely inaccurate, and I suspect whoever made that table didn't know what they were doing.
The deficiencies cited for GCC, for example, are parts of Annex F (alignment with IEEE floating point) which is officially optional.
I'm pretty sure cparser/FIRM has at least as many things incorrect in floating point as GCC does. Especially on 32-bit x86 where it's really hard to get them right, yet it's colored green/perfect in the table.
This entry was edited (Tuesday, September 22, 2026, 8:35 PM)
Yep. That one I agree was a good thing, although it was in every real compiler far before.
I actually implemented them by accident originally because in a normal C parser it takes effort to actually spot and block anonymous structs in the type parsing.
And I added unicode identifiers because I firmly believe you should be able to write code in any language or have a variable that's 🐏
This entry was edited (Tuesday, September 22, 2026, 8:33 PM)
@lispi314 @dalias @ell1e @wbrawner There are two standards to any programming language, the one written on paper and that one that's the implicit contract between developers and their tools.
C has a mindnumbing set of "undefined" cases, rules about pointers that nobody understands, and missing things everyone implements the same way because the language is really defined by convention too
I think it's fair to say that is true of pretty much any programming language.
@ell1e @etchedpixels @lispi314 @wbrawner That's the price paid for having it be an International Standard (TM) rather than a bunch of techbros doing YOLO spec-by-implementation shenanigans.
But seriously, everyone just gets the draft documents which are freely distributed and say the exact same thing.
@dalias @ell1e @etchedpixels @wbrawner > That's the price paid for having it be an International Standard (TM) rather than a bunch of techbros doing YOLO spec-by-implementation shenanigans.
Hardly, it's pure capitalism-brained nonsense to have it published as something other than an Open Standard, or indeed for those organizations to be allowed to publish anything else without losing their accreditation.
I'd rather be warezing standard documents from the ISO than going along with whatever the FOSS bros inviting AI and Palantir into their projects publish for free.
@ell1e @dalias @lispi314 @etchedpixels @wbrawner Could you untag me, please? I don't think C is any kind of way forward, and as such, I don't think I have anything to contribute here.
@ell1e @etchedpixels @MaddieM4 @wbrawner @Kroc That's how C99 looked until maybe 2007 or so. C moves at a glacial pace and this is a very very very good thing.
@dalias @etchedpixels @MaddieM4 @wbrawner @Kroc to me, C is mainly useful to 1. run old code that is still maintained and worth using and 2. as an interface language (e.g. to share parseable FFI headers for other languages to call into).
so i guess as a standalone language, C to me actually is a bit of a dead end.
@ell1e @dalias @MaddieM4 @wbrawner @Kroc That's a really good point. If you are trying to write something new then except for small embedded stuff or squashing stuff into tiny spaces C is probably the wrong answer.
@dalias @MaddieM4 @wbrawner @ell1e @Kroc C23 is nothing to do with C as K&R designed it. But that's really a side issue to the fact that parsing C code, even nasty bloated C23 code is not difficult.
@etchedpixels @wbrawner @dalias @ell1e @Kroc hey, you can take my C99 designated initializers and explicit width integers from my cold dead hands! I'm half kidding, but agree about codegen being the hard part. I haven't tried qbe but I like its documented mentality (every optimizer pass that can be reasonably implemented in a small library).
Without them it's impossible to write portable code initializing structs defined by the system headers or other API-level-only contracts that document members but don't mandate an order for them (or lack of named padding between them).
@MaddieM4 @etchedpixels @wbrawner @dalias @Kroc QBE looks nice, but i have neither seen it commit to no AI nor does it seem to support 32bit (which on the one hand I understand, on the other hand that seems like a bit of a portability headache)
@ell1e @MaddieM4 @etchedpixels @wbrawner @Kroc Last I checked it also lacks support for inline asm, making it a non-solution for systems programming.
Not really a problem for application-level though, and if you can compile libc and your kernel with pcc or firm or something, maybe a QBE-based application compiler is fine.
@dalias @MaddieM4 @etchedpixels @wbrawner @Kroc hmmm lack of any inline assembly also seems a little painful to be honest. i could see myself wanting to use inline assembly for some corner cases
anyway, QBE is impressive and it's a small project. i applaud them for their work!
@lispi314 @ell1e I mentioned aports because Alpine works fine that way. You can install a stable version, clone the aports repo, and build some newer packages you need against whatever libs your stable distro has.
@dalias @ell1e So keep apk to main, no community (second-tier official support, if I understood the documentation correctly), and aports everything else?
(Assuming the core doesn't get slopped.)
With no support for anything convenient like Gentoo overlays, is there something comparable to yay & such for Alpine apk?
Avoiding the slackware (or raw dpkg, as it does at least detect conflicts) experience matters to me.
Something I can't stop thinking about after reading your post earlier: how did the techbro gain the power to shut down community dialog? We've seen this happen enough times that it seems like a common-ish governance failure.
A concluding hot take: software developers should not have majority say over software-centric communities involving significant numbers of non-developers. I don't know how realistic that is but there are many perverse incentives involved.
Ai Bros keep asking for consent to embrace the fascism they already planned to embrace and being shocked when people say no. Weird how that keeps happening, it's almost as if people with no financial incentive to push it don't want AI, and those whose lives are actively being harmed by AI and the politics the pushers support definitely don't want it.
if you keep throwing the word fascism around to designate everything you dislike, it will lose meaning eventually. Also you're not a community, you're a gang of harassers that lash out at people doing a hard and thankless job instead of going to therapy. As a maintainer myself, I feel sorry for KDE folks.
four.php
in reply to four.php • • •reshared this
👻 Furbland's Very Spooky Account™ 👻, Dźwiedziu and Ódio e Apatia reshared this.
Unbro
in reply to four.php • • •reshared this
Ódio e Apatia and webhat🔜#40c3 reshared this.
four.php
in reply to four.php • • •Anyway, as the guy who opened up the issue proposing that KDE explicitly allow slop PRs quite validly points out, it's not really up to anyone but KDE contributors. That's fair.
But what the same guy is a thousand percent communicating though, is that KDE is not for people like me. It's not for people who have something to lose to fascists. If KDE developers don't want to stand for KDE's users, there's no obligation for them to do so, but let's not mince words about it either.
reshared this
👻 Furbland's Very Spooky Account™ 👻, Cassandrich, David Gerard, Techokami, Dźwiedziu, [object Object], Simon Zerafa, Ódio e Apatia, papiris🏴 🇵🇸 ⁂ and webhat🔜#40c3 reshared this.
four.php
in reply to four.php • • •This is not academic to me. It's very analogous to the situation with Calibre — not identical, but analogous. As lead developer of Calibre, Kovid had zero obligation to his users. He was *and is* free to betray Calibre users with impunity.
I don't hold his choice to do so in high regards, however, nor do I hold the guy opening that issue on KDE's AI policy in any kind of particular respect.
I hope that the rest of the KDE crowd tells him to take a hike and stop making things worse. We'll see.
reshared this
👻 Furbland's Very Spooky Account™ 👻, Ozzelot, Dźwiedziu, Ódio e Apatia and Elyse M Grasso reshared this.
Nelson
in reply to four.php • • •I don't fully agree he has no obligation to his users.
He has no obligation to serve them, he doesn't work for them. He has no obligation to provide software to any standard, or to make decisions the way they want him to.
But generally we all have an obligation not to harm other people. The more direct and significant the harm, the greater the obligation to avoid the harm. Your right to swing your arms ends at my face.
What level of betrayal is acceptable? Not all levels.
reshared this
👻 Furbland's Very Spooky Account™ 👻 and Cassandrich reshared this.
✧✦Catherine✦✧
in reply to four.php • • •Cassandrich reshared this.
Glyph
in reply to ✧✦Catherine✦✧ • • •four.php
in reply to Glyph • • •@glyph @whitequark Perhaps I was being glib to elide that gnarly corner, but I meant something closer to "I cannot reasonably expect to compel said behavior through any means that looks roughly legal-ish, including project codes of conduct through to actual laws."
In a moral sense, though, yes, I agree, obligated.
✧✦Catherine✦✧
in reply to four.php • • •four.php
in reply to ✧✦Catherine✦✧ • • •four.php
in reply to four.php • • •Cassandrich
in reply to ✧✦Catherine✦✧ • • •Cassandrich
in reply to Cassandrich • • •rakoo likes this.
webhat🔜#40c3 reshared this.
StarkRG
in reply to Cassandrich • • •@dalias @whitequark
Amazon did its best to ruin the concept of trademarks too.
To supposedly curb the absolute junk that gets peddled on their site, they insisted that every product *must* have a trademarked brand name. Which is how we ended up with Heqqqqqqi brand Android tablets that are carbon copies of Tuukwww brand Android tablets.
Great job there, Amazon. You really showed those low-effort crap resellers.
StarkRG
in reply to StarkRG • • •This is why I prefer ordering from AliExpress, they make no effort to disguise that they're selling a buttload of garbage and scams with a few good things mixed in (that aren't actually all that hard to pick out of the muck once you're familiar with it). And they're cheaper.
RS, Author, Novelist, Prosaist
in reply to StarkRG • • •@StarkRG
This explains much as there is just so many unique meaningful or seemingly meaningful names for companies (or automobile models to cite a cognate) that could exist. However, I'd love a reference link or a publication name—or some info on how you discovered this very believable factoid.
StarkRG
in reply to RS, Author, Novelist, Prosaist • • •@sfwrtr Honestly, I don't remember the article I read it in, but it was years ago, shortly after Amazon made that change and all of those random letter brands started showing up. It'll be buried in their enormous set of requirements and terms which vary from country to country and it only takes one reasonably large market with that requirement for that effect to occur.
(1/3)
StarkRG
in reply to StarkRG • • •@sfwrtr
I did a quick search and this was the closest I could find:
- identical products by different sellers result in combined listings (encouraging superficial distinctions like random brand names)
- almost every product requires a GTIN (could be difficult to track down for unbranded electronics bought directly the factories, and you might simply opt to register your own brand for them)
- Amazon's "IP Accelerator" tool that makes registering trademarks just a few clicks away.
(2/3)
StarkRG
in reply to StarkRG • • •@sfwrtr
Sources:
sell.amazon.com/blog/amazon-pr…
sell.amazon.com/blog/amazon-gt…
sellercentral.amazon.com/help/…
sell.amazon.com/brand-registry
(3/3)
What are product IDs or GTINs?
Amazon Selling Partner BlogStarkRG
in reply to StarkRG • • •RS, Author, Novelist, Prosaist
in reply to StarkRG • • •@StarkRG
I appreciate the effort!
What I gleaned is that you need a unique store name, however it is greatly encouraged that you register a trademark. Whilst it is practically a difference without distinction, in practical use getting that trademark prevents another seller from poaching it despite you using the name. In any case, a trademark doesn't appear to be a requisite, but without one you have to jump through an additional hoop of getting your store name okayed by someone at Amazon. At least that's what the doc seems to state; I've never become a seller to find out how it works IRL.
Regardless, the result is the same: lots of gibberish store names! Yes, it's annoying to search for something like a heating blanket and find the exact same product sold and branded by dozens of other stores, probably sold generically by one wholesaler and thus with the exact same specs and flaws. The result? Fifty different options, but maybe only three different competing products.
Thanks for posting this, however. Good food for thought.
StarkRG
in reply to RS, Author, Novelist, Prosaist • • •RS, Author, Novelist, Prosaist
in reply to StarkRG • • •StarkRG
in reply to RS, Author, Novelist, Prosaist • • •RS, Author, Novelist, Prosaist
in reply to StarkRG • • •@StarkRG
Labor is the most expensive part of most businesses. Too many sellers, and they don't want to add pennies to the cost of millions and millions of products.
[Sarcastic:] Maybe they could teach Rufus to catch these things. Finally, a worthwhile task for AI.
Juho Mäntysalo
in reply to Cassandrich • • •Oh, that's a great argument! Thank you for (re)sharing!
Haelwenn /элвэн/
in reply to Cassandrich • • •I think I'd more compare it to takeovers, and we've had quite a lot of those in foss.
Cassandrich
in reply to Haelwenn /элвэн/ • • •webhat🔜#40c3
in reply to Cassandrich • • •@dalias licensing is why we see so many crap products which are branded with the names of long dead companies which were known for their high quality products
cc: @lanodan @whitequark @xgranade
Cassandrich reshared this.
Oggie
in reply to four.php • • •Sorry, I'm a bit out of the loop- what exactly changed with calibre?
Searching around I'm seeing some stuff about an LLM 'discuss the book' (which...weird and crazy, but at least it seems to default to off?), but I am wondering if I missed something worse.
Hadley
in reply to Oggie • • •@Oggie In Calibre, you can right click on a book and "discuss selected book with AI."
In reailty though, it's just a completely bare-bones chat interface that can't even handle being resized correctly, and the only thing it does differently from any regular chatbot interface is it first sends a message saying something like, "I am currently reading the book Project Hail Mary, and I would like to ask some questions about it."
Even if you love all things AI, it's just incredibly bad.
Darby Lines
in reply to Hadley • • •manchicken
in reply to four.php • • •I don’t want any of this AI bullshit on my equipment.
Is this how I discover Plan9?
Tim Ch-AI-se (now with bonus spooky) 👻
in reply to manchicken • • •@manchicken
and/or NetBSD 😁
@xgranade
:: Asta ::
in reply to four.php • • •reshared this
David Gerard, Dźwiedziu and [object Object] reshared this.
four.php
in reply to :: Asta :: • • •:: Asta ::
in reply to four.php • • •Sensitive content
knew I shouldn't have clicked it
"Part of what we care about in KDE is providing an answer to what makes software and its development paradigm socially responsible."
me: "yes!"
me after reading more: "... jesus christ, he's arguing FOR the fucking LLMs."
How can you... possibly... POSSIBLY... use the words "socially responsible" in your *pro* comment. Jesus christ, it is clear that guy has no idea what that means.
newsflash: adding "poisoning minorities", "destroying collected knowledge", and "funding fascism" to your developer workflow *isn't socially responsible*
cthos 🐱
in reply to :: Asta :: • • •Sensitive content
:: Asta ::
in reply to cthos 🐱 • • •Sensitive content
:: Asta ::
in reply to :: Asta :: • • •four.php
in reply to :: Asta :: • • •cthos 🐱
in reply to :: Asta :: • • •Sensitive content
Proposed KDE LLM guidelines (#187) · Issues · Plasma / Plasma Workspace · GitLab
GitLabpapiris🏴 🇵🇸 ⁂ reshared this.
four.php
in reply to cthos 🐱 • • •Sensitive content
:: Asta ::
in reply to four.php • • •Sensitive content
yep yep, time to log off, audrey
I did find the *actual nazi* in the thread. Someone found their twitter account that is just straight up full of fash trash and they kept defending themselves with "I have been nothing but polite in this thread"
thing is... as a temperature check, I don't love that he seemed to feel comfortable sticking around after that was exposed.
:: Asta ::
in reply to :: Asta :: • • •Sensitive content
I wish I could hop onto that thread and be like "hey has anyone noticed that actual mask off nazis agree with the LLM policy, do you want to maybe think about why that is"
but no... "too heated"...
Fall smell is leaf mold 🍁
in reply to :: Asta :: • • •:: Asta ::
in reply to Fall smell is leaf mold 🍁 • • •Sensitive content
Fall smell is leaf mold 🍁
in reply to :: Asta :: • • •@aud @cthos regardless, the most important thing is to actually pay attention to the talk that was given which clearly states
WE DO NOT PROPOSE PUTTING AI INTO KDE
:: Asta ::
in reply to Fall smell is leaf mold 🍁 • • •Sensitive content
Hey, maybe PAY ATTENTION TO WHAT THE FUCK I'M TALKING ABOUT
Which is the thread, about the proposal, to allow LLM usage. Either read what I'm saying or get fucked.
:: Asta ::
in reply to :: Asta :: • • •Sensitive content
Proposed KDE LLM guidelines (#187) · Issues · Plasma / Plasma Workspace · GitLab
GitLabFall smell is leaf mold 🍁
in reply to :: Asta :: • • •:: Asta ::
in reply to Fall smell is leaf mold 🍁 • • •Sensitive content
cthos 🐱
in reply to :: Asta :: • • •Sensitive content
:: Asta ::
in reply to cthos 🐱 • • •Glyph
in reply to :: Asta :: • • •Sensitive content
four.php
in reply to Glyph • • •Sensitive content
:: Asta ::
in reply to Glyph • • •Sensitive content
Glyph
in reply to :: Asta :: • • •Sensitive content
Glyph
in reply to Glyph • • •Sensitive content
David Gerard
in reply to :: Asta :: • • •Sensitive content
Dźwiedziu
in reply to David Gerard • • •Facts don't care about his feelings \s
@aud @cthos @xgranade
1-bit machine goth
in reply to David Gerard • • •Sensitive content
reshared this
Cassandrich, David Gerard and Oriel Jutty reshared this.
:: Asta ::
in reply to 1-bit machine goth • • •Sensitive content
are you SERIOUS??
a fascist is never polite, by definition. They may be using "polite speech", but they are using polite speech specifically to provide cover for their hateful speech and actions. Much how like an anti-Black racist isn't "a good person" for being polite when they're talking to white people. That "politeness" is intimately linked to their bigotry.
:: Asta ::
in reply to :: Asta :: • • •Sensitive content
David Gerard reshared this.
1-bit machine goth
in reply to :: Asta :: • • •Sensitive content
reshared this
AlisonW ♿🏳️🌈♾️ and David Gerard reshared this.
:: Asta ::
in reply to 1-bit machine goth • • •Sensitive content
I only have one PR pending merge to KWin, but I've gone ahead and asked that it be put on pause until this issue is resolved as I will not contribute to a project that allows this.
It's a small PR and a small change. My principles, unfortunately, are not at all small and I won't compromise them.
the esoteric programmer
in reply to :: Asta :: • • •Shadow, Ruler of Darkness
in reply to the esoteric programmer • • •Snapshot: archive.ph/6YccJ#note_1648724
(you might need to switch DNS as Cloudflare is blocking the link)
the esoteric programmer
in reply to Shadow, Ruler of Darkness • • •mx alex tax1a - 2020 (6)
in reply to :: Asta :: • • •Sensitive content
Dźwiedziu
in reply to four.php • • •It comes down to who they get to treat as human and who a 3/5ths of a human at most \s
@aud
cthos 🐱
in reply to four.php • • •four.php
in reply to cthos 🐱 • • •cthos 🐱
in reply to four.php • • •Dźwiedziu reshared this.
four.php
in reply to cthos 🐱 • • •The Orange Theme
in reply to four.php • • •Dźwiedziu reshared this.
MaddieM4
in reply to four.php • • •Just earlier this morning, I saw someone with some profile aesthetic signals of queer outcasthood making what was very likely a thinly-veiled post about me/my reporting (if you wanna elevate it to "reporting") on the KDE situation. How the stuff about the Akademy talks was overblown, by "the most annoying parts of the Anti-AI people".
I blocked them because like... if that's the quality of a person's judgment, I'd like less of their takes in my life, as a protection of my personal peace and sanity. But I keep finding myself wondering: "how's that feel now? Now that we know it was early steps to manufacture consent? That some KDE developers got called out for some shocking Twitter racism in that thread? That KDE really is presenting itself lubed up and ready for the fascist stereotype-repeater machine, and don't care if it makes the project hostile to minorities on the dev team and in the userbase?"
I dunno how a person with those traits feels stakeless in this. I don't know.
reshared this
Dźwiedziu and Cassandrich reshared this.
MaddieM4
in reply to MaddieM4 • • •The funny thing about this is I was pretty clear that I wasn't jumping to conclusions and assuming the worst, just that KDE was raising questions that we as a community needed to demand answers to, because the worst-case answers were really bad, and the signals raising these questions were really alarming. I actually advised some other people to slow their roll, in case it was just like... one or two Akademy organizers who didn't speak for the dev team. I try to always assume the best I can of people, even if it puts me on the back foot in situations where I later learn I'm wrong.
I'm tired, boss. I'm so, so tired.
reshared this
Dźwiedziu reshared this.
Esther Payne
in reply to MaddieM4 • • •@MaddieM4 I think there's some folks who really are very privileged and don't get consent.
Nor do they seem to understand the concept of intersectionality.
I do not have the brain space to switch my gui, but if KDE are going that way, rebuilding my environment it is.
I'm already playing with tiling managers and I switched my long term distro last year. I will make the time.
MaddieM4
in reply to Esther Payne • • •@onepict Yeah. Thankfully I already use a very stable, pinnable tiling WM already, and just have KDE as my backup full-featured DE. This mostly hits me in the programs - I frequently use Kdenlive for editing, even for simple stuff like taking snippets of that Scott Jenson talk recently.
As much as I've enjoyed CachyOS, I do think I'll probably switch to AntiX for my Linux needs in the future, as an actively anti-fascist distro with good ergonomics in non-slop lite X11 WMs, and try to move most of my workflow to multiple flavors of Not Linux (probably partitions for GhostBSD, 9front, and Haiku). You know. When I somehow invent a zero-point module to pull the energy to do this out of the friggin' aether.
Kroc Camen
in reply to MaddieM4 • • •@MaddieM4 I saw that too and it felt very off to me as well, I’m impressed by your willing to call it out — you can’t brush off wanting to stop fascists taking control over the computing landscape we depend upon with FLOSS insularity and pearl-clutching.
Very specific KDE people are right now trying to force theough a change that violates every FLOSS principle, moral and legal, and they’re trying to say they hate all this misplaced anger. My dear, my anger is placed squarely where it belongs — in the face of AI pushing fascist fucks who know _exactly_ what they’re doing and use “civil discussion only” to enable themselves at every step
ell1e coding things
in reply to Kroc Camen • • •@Kroc @MaddieM4 the Alpine Linux discussion seems to involve similar arguments of LLM users being upset people care, it's a bit sad
(Alpine seems lost to the slop, sadly)
Cassandrich
in reply to ell1e coding things • • •MaddieM4
in reply to Cassandrich • • •@dalias @ell1e @Kroc I think that's going to be a really common path, as well as people trying more obscure slopless operating systems where its viable for their individual needs (see also: there is no universal Year of the Linux Desktop, only personal ones).
As for the way we do non-slopware, with some lessons learned... well I'll just speak for myself as a dev, and maybe other people will resonate with it. I talk about my far-off plans for developing and maintaining an OS, and people have gotten to watch the shape of those plans shape and refine over time. I once wanted to make a Linux distro around the Layover package manager, so, Layover Linux. Eventually I decided I wanted to not be Linux based, and I had some microkernel aspirations, and that was called Mercury Microkernel, and I thought maybe the desktop should be named after that - Mercury. But now I think, let these pieces have their own names. "LongevityOS" is maybe the name that ties it all into the right mission statement.
MaddieM4
in reply to MaddieM4 • • •@dalias @ell1e @Kroc I dunno. I think I'm just really, very permanently "over" software that needs or expects regular updates, or especially software that feels entitled to push new versions onto your device using features or security as justifications.
Paradoxically, software that doesn't require updates later and can be securely and productively old-as-dirt feels like a flashy and exciting idea to me right now. Especially light stuff that runs on a toaster.
reshared this
Space Hobo reshared this.
Cassandrich
in reply to MaddieM4 • • •And they key to having that is to stop having attack surface.
The reason our phones supposedly need constant updates is because they come with a non-removable SMS client (that's grown to use new channels outside just SMS) that accepts media attachments from any random sender and attempts to decode them, and because they come with a non-removable browser that insists on letting random sites run code in a YOLO unsafe JIT engine for ZOMGFAST reasons.
If your stuff is not attack surface, it doesn't need updates. 💡
ell1e coding things
in reply to Cassandrich • • •@dalias @MaddieM4 @Kroc Firefox still doesn't allow running WASM without JIT being enabled. Kind of frustrating. I agree all browsers should ship with JIT disabled. (Also WebGL/WebGPU and high performance timer, if you ask me.)
I'd love to see some opt-in "high end game" site permission...
Cassandrich reshared this.
MaddieM4
in reply to Cassandrich • • •@dalias @ell1e @Kroc So true.
My wife is obviously going through a lot of the same tired exasperation as the rest of us, just less terminally online than me. She's looking at getting a Linux laptop set up that she just never activates any internet connection for after installing a set of packages she cares about. Without that connection (or bluetooth etc.) as an attack surface, you actually can just enjoy a pre-slop Linux distro release as a stable appliance indefinitely.
Cassandrich
in reply to MaddieM4 • • •Kroc Camen
in reply to MaddieM4 • • •@MaddieM4 @dalias @ell1e As much as I know that we technically skilled users, hackers and tinkerers can solve this for ourselves, the lacunae surrounding hardware support and how friends and family are supposed to be included in this is too much to polyfill by ourselves. KDE is a well funded organisation, now with EU government backing
I'm all for slow software, but at some point it has to come head to head with government / banking apps that literally won't run on anything without a locked bootloader and Google spyware on it.
Our digital space has been occupied by aggressors. I'm of the opinion that we should have a French-resistance attitude to what we make rather than trying to outcompete them. We won't be able to live without slop in our lives until the bubble bursts. Until then, software vigilantism is the way to protect ourselves and to fight the bastards.
Cassandrich
in reply to Kroc Camen • • •Kroc Camen
in reply to Cassandrich • • •Cassandrich
in reply to Kroc Camen • • •@Kroc @MaddieM4 @ell1e Surely there are people who need these services who can't afford phones that can run them, or who can't retain a phone (abusive partners, camp raids, etc.), or who can't figure out how to use a phone, etc.
If there really is a problem of governments trying to force people to use apps that need DRM phones, maybe public libraries should have burner phones where you can go in use it to login and do whatever they're making you do, with assistance if needed.
ell1e coding things
in reply to Cassandrich • • •you would think the government would keep that in mind.
but it currently looks less and less like it... otherwise, why are banks allowed to require an app for any sort of usable electronic access? the EU doesn't really seem to care
(I use a Linux phone and a bank that offers QrTAN. But many banks now appear to require either some custom weirdo USB driver and/or an Android/iOS app.)
Bin Chicken
in reply to MaddieM4 • • •ell1e coding things
in reply to Bin Chicken • • •@BinChicken @MaddieM4 @dalias @Kroc one of the things that seem to shock tech people I talk to the most, is 1. I turn off JIT/WebGL/WebGPU for browsers, 2. I turn off hyper threading and I run with mitigations=full, and 3. I like to develop software on devices with 16GB RAM or less. Sometimes I use a Raspberry Pi 5 as my work desktop during the summer.
Why did everything need to be a bit faster and a bit more 3D and give us all these problems? I don't get it. We really don't need half the features that browsers added.
William Brawner
in reply to MaddieM4 • • •MaddieM4
in reply to William Brawner • • •@wbrawner @dalias @ell1e @Kroc Nice! While I do really want to have rump kernel compatibility to share code with NetBSD (I think freeing drivers from specific kernels is literally vital to kernel diversity) and that will help a lot for the hardware support side, there is a *lot* to do in the userspace, at a scope where I am just not enough people for the job. People like you will be critical to making something usable and human-friendly. I know I'll need a good graphic designer at some point, too.
Having a compiler expert in particular would be a big deal. I'm currently learning a lot by implementing and designing a programming language, but it cheats some of the hard stuff by being an interpreted language that transpiles to C source code. I am very intimidated by the idea of writing a C compiler, and that's just a thing that's gonna be needed at some point.
ell1e coding things
in reply to MaddieM4 • • •@MaddieM4 @wbrawner @dalias @Kroc for what it's worth, I'm working on a programming language too.
But I think a C compiler that is no AI and supports a reasonably wide subset of it, would be a big deal for sure. If only for compatibility.
(And for now, my compiler uses C as a universal assembler, too.)
I'm just not interested enough in C myself to commit to writing a C compiler on top of things. I have too many projects already...
Cassandrich
in reply to ell1e coding things • • •@ell1e @MaddieM4 @wbrawner @Kroc Very pleased to see the goal of using C like that!
In case you haven't seen it, libfirm.github.io/ is somewhat-abandoned (plus side: that means no AI risk) but has a C compiler that's fairly well architected and small/simple. I'd love to see someone pick it back up.
Firm - Optimization and Machine Code Generation
libfirm.github.ioKroc Camen
in reply to ell1e coding things • • •@ell1e @MaddieM4 @wbrawner @dalias With respect to [All of the Above], there is already plenty of anti/no-Ai compilers like Zig (which compiles C) and OSes like Haiku / Redox.
If you want to rebuild the universe from first principles you will end up only solve theoretical problems rather than real ones.
Cassandrich
in reply to Kroc Camen • • •LisPi
in reply to Cassandrich • • •@dalias @ell1e @wbrawner From taking ages building it with Gentoo, I can confirm it does.
"Fast builds", yeah sure, if you don't care about rebuilding your toolchain I guess.
ell1e coding things
in reply to LisPi • • •@lispi314 @dalias @wbrawner in their defense, they're trying to ditch LLVM I think.
still, a fresh, lightweight, independent no-AI C compiler that supports C99 with the lower-hanging fruit from C11 (that would be e.g. atomic ints that seem to be pretty essential for ARM64 targets, since there a 64bit int doesn't seem guaranteed atomic by the architecture) would be nice.
ValorZard
in reply to ell1e coding things • • •there is a c23 compiler someone made by themselves without ai
kefir.protopopov.lv/posts/anno…
(it has sadly gone into private mode, but the latest public branch might be useful)
also i found this c like language written in 10k LOC (and it has a hard cap on that)
github.com/robertmuth/Cwerg/tr…
GitHub - robertmuth/Cwerg: The best C-like language that can be implemented in 10kLOC.
GitHubThe Penguin of Evil
in reply to MaddieM4 • • •@MaddieM4 @wbrawner @dalias @ell1e @Kroc C compilers are easy to write. Good C compilers are a different story.
For Fuzix I wrote my own (in part because nobody else fitted an ANSIish C compiler in to a 64K address space machine).
It's C89 without bitfields basically, because bitfields are evil and everything added to C post C89 except "..." is wrong 8)
You could add C99isms to it but either way C compilers are not tricky, codegen is.
codeberg.org/EtchedPixels/Fuzi…
Fuzix-Compiler-Kit
Codeberg.orgSpace Hobo reshared this.
Cassandrich
in reply to The Penguin of Evil • • •@etchedpixels @MaddieM4 @wbrawner @ell1e @Kroc Given how many bugs the mainstream compilers have had over the years misinterpreting the language specification, I question the claim that C compilers are easy to write. Unless you consider correctness as part of a stretch goal of "good" rather than just a baseline requirement.
"C89 without bitfields" is not a C compiler. A very minimal baseline is full C99 (at least long long, designated initializers, and compound literals are essential), and to be serious we should be aiming for C23.
ell1e coding things
in reply to Cassandrich • • •Cassandrich
in reply to ell1e coding things • • •@ell1e @etchedpixels @MaddieM4 @wbrawner @Kroc No, not yet, but if we're talking about a future that's not a dead-end, we should be *aiming for* (I used those words specifically) C23 support.
Otherwise what's bound to happen is a few years down the line, there's good software folks want to use, but it's written in C23, and now they're stuck getting another compiler for that. 🤦
Cassandrich
in reply to Cassandrich • • •ell1e coding things
in reply to Cassandrich • • •Kroc Camen
in reply to ell1e coding things • • •C99 - Wikipedia
Contributors to Wikimedia projects (Wikimedia Foundation, Inc.)Cassandrich
in reply to Kroc Camen • • •That's largely inaccurate, and I suspect whoever made that table didn't know what they were doing.
The deficiencies cited for GCC, for example, are parts of Annex F (alignment with IEEE floating point) which is officially optional.
I'm pretty sure cparser/FIRM has at least as many things incorrect in floating point as GCC does. Especially on 32-bit x86 where it's really hard to get them right, yet it's colored green/perfect in the table.
LisPi
in reply to Cassandrich • • •@dalias @ell1e @etchedpixels @wbrawner > unions standardized
Those were nonstandard for that long? Huh.
Cassandrich
in reply to LisPi • • •@lispi314 @ell1e @etchedpixels @wbrawner Anonymous unions and structs, so you can do things like
struct foo {
union {
int i;
unsigned u;
};
void *p;
} bar;
Where bar.i and bar.u have the same storage.
The Penguin of Evil
in reply to Cassandrich • • •Yep. That one I agree was a good thing, although it was in every real compiler far before.
I actually implemented them by accident originally because in a normal C parser it takes effort to actually spot and block anonymous structs in the type parsing.
And I added unicode identifiers because I firmly believe you should be able to write code in any language or have a variable that's 🐏
The Penguin of Evil
in reply to LisPi • • •@lispi314 @dalias @ell1e @wbrawner
There are two standards to any programming language, the one written on paper and that one that's the implicit contract between developers and their tools.
C has a mindnumbing set of "undefined" cases, rules about pointers that nobody understands, and missing things everyone implements the same way because the language is really defined by convention too
I think it's fair to say that is true of pretty much any programming language.
ell1e coding things
in reply to The Penguin of Evil • • •Cassandrich
in reply to ell1e coding things • • •@ell1e @etchedpixels @lispi314 @wbrawner That's the price paid for having it be an International Standard (TM) rather than a bunch of techbros doing YOLO spec-by-implementation shenanigans.
But seriously, everyone just gets the draft documents which are freely distributed and say the exact same thing.
LisPi
in reply to Cassandrich • • •@dalias @ell1e @etchedpixels @wbrawner > That's the price paid for having it be an International Standard (TM) rather than a bunch of techbros doing YOLO spec-by-implementation shenanigans.
Hardly, it's pure capitalism-brained nonsense to have it published as something other than an Open Standard, or indeed for those organizations to be allowed to publish anything else without losing their accreditation.
Cassandrich
in reply to LisPi • • •@lispi314 @ell1e @etchedpixels @wbrawner I don't mean it has to be that way, just that that's the system we presently live in.
I'd rather be warezing standard documents from the ISO than going along with whatever the FOSS bros inviting AI and Palantir into their projects publish for free.
ell1e coding things
in reply to Cassandrich • • •i'm not quite following how one is conencted to the other. (Sorry if I missed something!)
My guess would be the ISO C AI bro ratio is pretty high.
four.php
in reply to ell1e coding things • • •ell1e coding things
in reply to Cassandrich • • •@dalias @etchedpixels @MaddieM4 @wbrawner @Kroc i mean i suppose. i'm actually not sure though that C23 will ever gain wide adoption.
we'll see i guess!
Cassandrich
in reply to ell1e coding things • • •ell1e coding things
in reply to ell1e coding things • • •@dalias @etchedpixels @MaddieM4 @wbrawner @Kroc to me, C is mainly useful to 1. run old code that is still maintained and worth using and 2. as an interface language (e.g. to share parseable FFI headers for other languages to call into).
so i guess as a standalone language, C to me actually is a bit of a dead end.
The Penguin of Evil
in reply to ell1e coding things • • •ell1e coding things
in reply to The Penguin of Evil • • •The Penguin of Evil
in reply to ell1e coding things • • •@ell1e @dalias @MaddieM4 @wbrawner @Kroc
Good chance to get rid of perversions like
switch(c) {
case 0:
foo()
bar();
if (0)
case 1:
case1onyl();
shared();
}
and other gems of language misdesign like not trapping numeric overflows unless you use a special type defined to allow overflows.
The Penguin of Evil
in reply to Cassandrich • • •MaddieM4
in reply to The Penguin of Evil • • •Cassandrich
in reply to MaddieM4 • • •@MaddieM4 @etchedpixels @wbrawner @ell1e @Kroc Yeah lack of designated initializers is really the show-stopper.
Without them it's impossible to write portable code initializing structs defined by the system headers or other API-level-only contracts that document members but don't mandate an order for them (or lack of named padding between them).
ell1e coding things
in reply to MaddieM4 • • •Cassandrich
in reply to ell1e coding things • • •@ell1e @MaddieM4 @etchedpixels @wbrawner @Kroc Last I checked it also lacks support for inline asm, making it a non-solution for systems programming.
Not really a problem for application-level though, and if you can compile libc and your kernel with pcc or firm or something, maybe a QBE-based application compiler is fine.
ell1e coding things
in reply to Cassandrich • • •@dalias @MaddieM4 @etchedpixels @wbrawner @Kroc hmmm lack of any inline assembly also seems a little painful to be honest. i could see myself wanting to use inline assembly for some corner cases
anyway, QBE is impressive and it's a small project. i applaud them for their work!
LisPi
in reply to Cassandrich • • •@dalias @ell1e Besides Gentoo and NetBSD, what distros are there to facilitate this?
Gentoo doesn't really suit the "easy VMs" kind of use-case that Alpine did.
Cassandrich
in reply to LisPi • • •LisPi
in reply to Cassandrich • • •@dalias @ell1e So keep apk to main, no community (second-tier official support, if I understood the documentation correctly), and aports everything else?
(Assuming the core doesn't get slopped.)
With no support for anything convenient like Gentoo overlays, is there something comparable to yay & such for Alpine apk?
Avoiding the slackware (or raw dpkg, as it does at least detect conflicts) experience matters to me.
AUR helpers - ArchWiki
wiki.archlinux.orgCassandrich
in reply to LisPi • • •MaddieM4
Unknown parent • • •@kaidenshi @onepict oh for FUCK'S SAKE
Ahem. Sorry. Thank you, I had not previously been informed.
ltning
in reply to four.php • • •Lou.
in reply to four.php • • •No-AI isn't enough. No suggestion that AI is fine.
R.L. Dane 🍵
in reply to four.php • • •*sigh*
every. flipping. time.
Techbros be like, "Yo, relax dawg, it's not like we're talking about systematically enslaving you and maybe hunting you down for sport."
"That's literally your endgame."
"Ok, but that's not what we're talking about TODAY. Calm down!"
reshared this
Jonathan Schofield, Ozzelot and Dźwiedziu reshared this.
Anthony
in reply to four.php • • •Something I can't stop thinking about after reading your post earlier: how did the techbro gain the power to shut down community dialog? We've seen this happen enough times that it seems like a common-ish governance failure.
A concluding hot take: software developers should not have majority say over software-centric communities involving significant numbers of non-developers. I don't know how realistic that is but there are many perverse incentives involved.
Mike P
in reply to four.php • • •- Yes (0 votes)
- Maybe Later (0 votes)
Poll end: Tuesday, September 29, 2026, 10:51 PMRycochet
in reply to four.php • • •Jean 🐧 🇺🇸 🏳️⚧️ 🏳️🌈
in reply to four.php • • •four.php
in reply to Jean 🐧 🇺🇸 🏳️⚧️ 🏳️🌈 • • •Dmitry Tantsur
in reply to four.php • • •four.php
in reply to Dmitry Tantsur • • •@creepy_owlet If we can't accurately call LLMs fascist, then what in the fuck are words even for?
Quit trolling and start fighting, good grief.
Ravioli
in reply to four.php • • •it's more like
Should we do the fascist thing? Y/n
Press nothing for yes
Press n for no (sudo command only)