rossy

Tbf I was talking about first-party Google TV boxes (Google TV Streamer and Chromecast with Google TV.) Likely not perfect for privacy, but not pure evil like built-in smart TV firmware, they support sideloading without time limits, alternative app stores (Flicky/F-Droid) and they're friends/family/roommate/partner friendly with support for DRM-ed streaming apps, Chromecast and pairing with the official YouTube app.

I chose one over the Apple TV because I knew I wouldn't able to stand any limitations on sideloading. For a while I was running a custom build of Wholphin with patched libmpv/libplacebo to work around a bug before the fix was upstreamed. A device without the freedom to do that sort of thing would drive me nuts.

Obviously, as an advertising company, Google should have your utmost scrutiny. But as far as I know, their TV boxes don't ACR your content. If they were the only two options, sticking with your smart TV software instead of using a Google TV box would be a privacy mistake.

> I remember a socialistic coworker of mine—that was bragging about his -arr setup—talking about the end of the culture because authors aren't earning money.

To be fair, you can have it both ways. If you buy-and-pirate, and you buy 4 UHDs per year or 10 albums per year, you've put more money into the industry than the average streamer. Admittedly, buying-and-pirating isn't as commonly done for movies and TV as it is for music, but that can be explained by the increased DRM and lock-in.

I don't see the issue. The thing I can't move past with generative AI is that it wouldn't exist without the labour of people who post things for free on the internet, but it's displacing those same people from their day jobs. Legal or not, I think having your own work used against your will to harm you is something to be justifiably angry about.

And many pro-AI communities sweep the issue under the rug, consider it invalid in some way, or just over-discussed and boring. They have to, because it's uncomfortable and it has no solution. There is no generative AI without scraped datasets. But the consequence of ignoring the issue is that you might be surprised when you encounter a large group who are still very (justifiably) angry about it. Bluesky has a large community of people who post art for free on the internet.

You can't expect them to be civil either, when so many countries are in a cost of living crisis. No one faces losing their home or health insurance and thinks "I'd better be polite to the people doing it," and a high profile developer posting publicly about implementing a long-asked-for feature with AI is exactly what the AI marketing machine needs. A lot of devs are wondering whether AI-generated code is good or slop, and it's the opinion of other respected devs that sways them.

Also, let's not forget that PDN isn't FOSS. Linux platform knowledge was scraped from FOSS Linux software, trained into AI, and is being used to port a closed source program to Linux where it will displace FOSS alternatives. And the dev expects a pat on the back from the community? Why?

Holding the threat of death over people isn't the way to a "high-trust society." Maybe you think it's the way to something similar, but you have to give up the word "trust" because it doesn't apply here. If the only reason I think a man isn't going to rob me is because I will shoot him if he does, I don't trust that man.

A society which only has low crime because everyone is able to use lethal force to defend themselves and their property isn't one I would call higher "trust" than one that maintains the same crime levels without that threat.

> Honesty section:

This one gets to me. Not only is it useless (paragraph would have the same meaning if it was removed) but wdym "honesty section," Claude? The whole readme should be honest! One of the many causes of psychological damage when reading AI-generated technical writing is the reminder that one of its tics is that declares when it's being honest or real, casting doubt on the rest.

I read the comment as "allowing wealth to access the skill of the people who created the contents of the training set." I don't think anything prior has allowed such direct access to the skill of other people while denying them access to wealth. At least the creators of stock photos and templates get paid.

Let's Encrypt can issue wildcard certs too

> It’s a targeted advertising machine that is extremely effective on kids and teenagers. It preys on them so, so efficiently. It’s a technical work of art. A young mind is extremely susceptible to the algorithms on those platforms. Much more than adults are, and adults are already really susceptible.

Sure, but the Australian government's definition of an age-restricted social media platform doesn't mention advertising or algorithms at all. Technically, their definition also covers algorithm-free social media like Mastodon, which I'd argue isn't nearly as harmful.

The framing of social media as something that's inherently bad no matter how you do it is a framing that helps social media giants like YouTube, Instagram and Facebook to continue to "do it" in a way that harms people. I'm sure they love the idea that the ills of social media can be solved by banning their least profitable users while doing nothing to regulate what they do with the others. They're probably thrilled that their healthier algorithm-free competitors haven't even entered the conversation. They want to be the tobacco companies of the future, because making addictive things for adults is incredibly profitable.

> They are mostly on the wrong side in the war on general-purpose computing.

All modern Chromebooks can be put into developer mode without opening the case, which gives you root access.

> (though this isn't the case for Anubis) the cliche inappropriately dressed inappropriately young anime characters dawned as mascots in an ever increasing number of projects

I think the fact that people bring up things that the Anubis mascot isn't when talking about Anubis is more telling of their own harmful (and potentially racist) biases against Japanese-styled media than it is about the idea of having anime-styled mascots for free software projects.

To be fair there are a lot of games on Steam that don't have DRM, which means you can just drag them out of the steamapps folder to a computer that doesn't have Steam and they work fine. The decision to add DRM comes from the developer/publisher, not Valve.

Did you forget about Vulkan? Valve and AMD are Khronos members and active contributors to the Vulkan spec. Games like Baldur's Gate 3 and Civ VII use Vulkan on Deck. There's a complete graphics ecosystem with full participation from the games industry that doesn't have Microsoft as the gatekeeper.

The more I think about it, the harder it is to recommend anything else for the average Windows gamer/prosumer but first-time Linux user.

- Rolling release, so you don't have to do a major upgrade twice a year - which would otherwise be much more often than Windows.

- Latest kernel and graphics drivers, so it works with newly released hardware with the best performance.

- Steam, NVIDIA drivers, H.264/H.265 codecs, Gamescope, GameMode, MangoHud, etc. all in the default repos - a huge boon for new Linux users compared to having them in an external repo like RPM Fusion or having to install them manually, which can otherwise cause confusing dependency problems over the life of the installation.

- Nothing unusual about it that would be confusing or cause compatibility problems. It's just a normal mutable binary distro with a normal package manager, upstream packages, glibc and systemd.

The biggest issue is the lack of an official graphical installer, but while the install process is intimidating, it's not very difficult for people who are patient, can follow detailed instructions, and have a vague idea of what a partition and a bootloader is.

I don't think being able to migrate your account addresses the rugpull concern. The rugpull scenario is that one day, in five years or so, bsky.app drops all AT Protocol support and transforms into a Twitter-like centralized social media website. The problem isn't that the account will stop "existing" but that Bluesky users will stop seeing it. The average non-techie Bluesky user who doesn't know about the AT Protocol won't even notice the change, except that, from their perspective, a tiny percentage of nerdy users have stopped posting. For you, "migrating" your account away is effectively just deleting it from the now-centralized Bluesky and willfully decreasing your audience by 100-fold or more.

The problem is a social not a technical one. It doesn't matter how good AT Protocol is at account migration. The vast majority of AT Protocol users think of themselves as Bluesky users and don't even know what the AT Protocol is. If the official Bluesky clients move away from the AT Protocol, the majority of users are moving with Bluesky.

For all the UX concerns people have with Mastodon/ActivityPub, at least they make it obvious that different users are hosted on different instances, and no one instance has more to gain than it does to lose by defederating.

It worked on my device. The page you linked looks very outdated and doesn't have my device's board or any device made in the past 5 years. The lists of unsupported devices also look pretty reasonable - old kernels, CPUs that don't support virtualisation and 32-bit ARM. Since modern ChromeOS uses the same virtualisation to run Android apps, I doubt there's a modern device where it doesn't work.

Sorry, but "CLI stuff" is not "as far as it goes" with desktop Linux apps on ChromeOS. ChromeOS provides Wayland and PulseAudio servers to the apps as well so GUI and audio works too. It even synchronises file associations and installs a ChromeOS-like GTK theme into the container. The Linux GUI apps I had installed back when I used it felt completely native.

I got a ChromeOS device a few years ago and it was great. I think they get an underserved bad reputation from being the locked-down devices you're forced to use in schools, but a personal ChromeOS device is a capable computer that can run any Android app or desktop Linux app.

Though having said that, in the past year I've replaced ChromeOS with desktop Linux (postmarketOS) and I love it even more now. 4GB of RAM was a bit slim for running everything in micro-VMs for "security," which is what ChromeOS does. I've had no trouble with battery life or Android emulation (Waydroid) since switching.

If the creator of this app doesn't know their stuff, they're putting other people at risk.

I only need to look at one file[1] (context[2], and it used to be worse[3]) to decide that I don't really want Hyprland, or anything else from that quick-and-dirty culture of software dev, on my system. Encouraging plugins to hook any function or method in a C++ program is insane. I'd be surprised if a Hyprland setup with multiple plugins could ever not be alpha-quality.

[1]: https://github.com/hyprwm/Hyprland/blob/e999ad664da9/src/plu... [2]: https://wiki.hypr.land/Plugins/Development/Advanced/ [3]: https://github.com/hyprwm/Hyprland/blob/a5a648091760/src/plu...

I don't see how AI can become democratized. (I don't follow this stuff too closely, but) it seems like larger models with less quantization and more parameters always outperform smaller models of the same type, and that trend isn't stopping, so if/when we get consumer hardware and local models that equal today's SotA SaaS models, the SotA SaaS models of that time will be even better, and even more impossible to run on consumer hardware. Not to mention that local AI is reliant on handouts from big business - both in base models that the community could never afford to train themselves, and in high-VRAM GPUs that can run big models, so if SaaS AI is more profitable, I don't think we'll be "allowed" to run the SotA at home.

Human skill was already democratized in that anyone can obtain skills, and businesses have to be good at managing those people if they want to profit from those skills - ultimately the power is in the hands of the skilled individuals. But in the hypothetical AI future, where AI has superhuman skill, and human skills are devalued, it seems like there will be a more cynical, direct conversion between the money you can spend and the quality of your output, and local/self-hosted AI will never be able to compete with the resources of big business.

As I understand, the AGPLv3 requires the corresponding source code to be provided under the same license, so the Arch guys wanting an AGPLv3 source package isn't just a niche Arch-specific concern or a "packaging issue," but a licensing requirement that can't be ignored or delayed.

> All you need to do is omit the Enterprise feature flag during compilation, and what you get is a 100% AGPL-3.0 build.

Maybe I'm misunderstanding, but my interpretation of this issue[1] is that Stalwart contains AGPLv3 licensed functions that call into the SEL licensed `has_tenant_access` function, among others, and that the affected functions are not conditionally compiled out of the AGPLv3 binaries. @afontenot says on that issue that they don't believe it's "possible to use Stalwart under the AGPL at present." Are they wrong and can that issue be closed?

I am also concerned about the webadmin. A free software program that downloads proprietary code on first start isn't free software in practice, and since there aren't two separate SEL and AGPLv3 licensed builds of the webadmin on GitHub, that must be the case.

> So while the optics of this situation may look rough from the outside, I promise it’s not some “open source in name only” kind of thing. It’s just one of those painful balance acts between building features, maintaining packages, and paying the bills.

I get it, but it's disappointing that AGPLv3 compliance is so low in the list of priorities that this licensing issue has been known about but not solved in 8 months, all while receiving grants intended for free software projects. That balancing act must have included the consideration that the free software community is regularly burnt by rug-pulls (Redis) and trust isn't easily won back once its lost.

> And hey, if you're heading back to Maddy, no hard feelings. But the door’s always open if you want to give Stalwart another shot down the road.

I might. Sorry if I've been harsh, but it's only because Stalwart is a very cool project. A FOSS all-in-one mail server written in a safe language is exactly what email needs, and since learning about it, I've been worried that it's too good to be true. Please don't let it be. I don't think it will gain the momentum to replace Postfix if it can't be packaged in Linux distros due to licensing issues.

[1]: https://github.com/stalwartlabs/stalwart/issues/783

I checked my repo to see which Stalwart version I was running and if I could update, and I was surprised to find that the Arch package has been deleted due to FOSS licensing concerns[1], the most severe of which seems to be that Stalwart can no longer build without proprietary code[2]. Other smaller issues include the fact that the web admin interface isn't included in the source distribution, but is downloaded from GitHub on first run, and _also_ seems to contain proprietary code[3].

These issues, which would be showstoppers for a real free software project, and pretty easy to fix if you were the rightsholder of the code, were promised to be fixed "in a few weeks" in September last year, and "in a few months" in January this year, however they're still not fixed, which means I can't upgrade - not that I probably want to anymore. I truly believe in free software, so I find the idea of using "open source" as an empty marketing bullet-point for at least eight months to be fairly distasteful. Might be time to switch to Maddy.

[1]: https://gitlab.archlinux.org/archlinux/packaging/packages/st... [2]: https://github.com/stalwartlabs/stalwart/issues/783 [3]: https://gitlab.archlinux.org/archlinux/packaging/packages/st...

Looks like it includes the stuff for running OpenGL, OpenCL, Vulkan and VA-API on top of D3D12 as well.

Microsoft made their own Mesa builds for this once, but they don't appear to be updating them? https://apps.microsoft.com/detail/9nqpsl29bfff

If I'm already compromising by using non-free software, does it matter that much? How do I know what a native app is sending back in its telemetry?

No, it's definitely a win for Linux. I get it. I've dabbled in software minimalism. I love native dev. I know the web "sucks." But the range of mainstream software available for Linux has exploded now that software is moving to the web (including Electron) and I can't see how that's a bad thing from the perspective of a Linux user. Of course I'd rather open a web browser to run an app than change my entire operating system to run an app.

There are even games like Infinity Nikki with anti-cheat that allows the Steam Deck but specifically detects and blocks desktop Linux. You have to wonder if that gets them any real security since the method they use to detect the Deck is probably spoofable.

> Art is something that cannot be generated like synthetic text so it will have to be nearly forever powered by human artists or else you will continue to end up with artifacting.

The rise of GPT slop is making it increasingly clear to me that this distinction doesn't exist, and it's just an under-appreciation of the skill that goes into good writing. That thing where LLMs generate overly-wordy mealy-mouthed text is just what bad writing looks like: the writing equivalent of a bad drawing. Subtle inaccuracies and ill-fitting metaphors are just the text version of visual artifacts.

Not to diminish the plight of art and artists, but it's the same as the plight of writers and writing. Writers are also having their copyrighted works used against their will to destroy their own industry. LLMs also need big human-written datasets to keep the magic running, that are drying up as they get poisoned by their own output.

In my opinion, the Win32 era only ended recently (and I don't think that's a good thing!)

Back in Windows 10, you could use C and the Win32 API to make a lightweight dependency-free GUI app that looked exactly like the built-in apps like Notepad, File Explorer, Task Manager, and so on. You could even get the same ribbon component they used, which was implemented in a public DLL.

In Windows 11, all those built-in apps have been rewritten with Mica-styled widgets in XAML Islands, and if you want to build your own lightweight dependency-free C app that looks the same way, you'll find out it's not possible. For the first time ever, the Win32 common controls library implements a widget style that's completely different to what the built-in apps look like. Even if you use XAML Islands, if you just use the system XAML, you'll find your app looks like Windows 10, because the Windows 11 look-and-feel is implemented in WinUI, a DLL that you're supposed to vendor with your application.

The era of Windows shipping with a C API that you can use to build perfectly native-looking apps ended when Windows 11 came out 3 years ago, and it's a real shame.

Iceshrimp, or at least this JavaScript version of it, is a fork of Misskey, which has been mentioned on HN a few times. If you've heard of Mastodon, Misskey and its forks participate in the same decentralized social network - Mastodon users can like and comment on the tweet-like posts that a Misskey user makes and vice-versa.

The Iceshrimp devs are also working on a "port" to .NET that's basically a brand new social media app, but with an upgrade path from the JavaScript version.

I also use Iceshrimp to self-host my own fediverse account and I think it's pretty good software. I think it has a better UI that Mastodon and it has some cutesy features that Mastodon lacks, like being able to emoji-react to posts.