You can also just sync your files. Koreader, by default, stores metadata in the same folder as the epub file. If you sync the directory with something like Syncthing, then you don’t need a server.

Thanks for posting the solution!

If you happen to be using a BTRFS or XFS file system, you might want to try duperemove. It will help you reclaim usable disk space without deleting any files, by using those filesystems’ built-in support for data deduplication and copy-on-write. In other words, it will make duplicate files point to the same data on disk, but still work as individual files. Files will appear and function exactly the same, and editing one copy will not change another (unlike with hard links, for example). That way it won’t interfere with cases like Flatpak or Python virtual environments where you really need multiple copies of the same files.

Yeah, I’d rather come at it from the opposite direction. “Everyone censors, so this exactly the kind of shit your government is going to try to force on you in the future.” Everyone should care about this, if only out of self-interest.

China’s writing the playbook. Other countries will follow it sooner than you might think. This is everybody’s problem.

Generally speaking, xz provides higher compression.

None of these are well optimized for images. Depending on your image format, you might be better off leaving those files alone or converting them to a more modern format like JPEG-XL. Supposedly JPEG-XL can further compress JPEG files with no additional loss of quality, and it also has an efficient lossless mode.

Do any of them have the ability to recover from a bit flip or at the very least detect with certainty whether the data is corrupted or not when extracting?

As far as I know, no common compression algorithms feature built-in error correction, nor does tar. This is something you can do with external tools, instead.

For validation, you can save a hash of the compressed output. md5 is a bad hashing algorithm but it’s still generally fine (and widely used) for this purpose. SHA256 is much more robust if you are worried about dedicated malicious forgery, and not just random corruption.

Usually, you’d just put hash files alongside your archive files with appropriate names, so you can manually check them later. Note that this will not provide you with information about which parts of the archive are corrupt, only that it is corrupt.

For error correction, consider par2. Same idea: you give it a file, and it creates a secondary file that can be used alongside the original for error correction later.

I also want the files to be extractable with just the Linux/Unix standard binutils

That is a key advantage of this method. Adding a hash file or par file does not change the basic archive, so you don’t need any special tools to work with it.

You should also consider your file system and media. Some file systems offer built-in error correction. And some media types are less susceptible to corruption than others, either due to physical durability or to baked-in error correction.

Unfortunately, you still need a level of trust with Proton. Even aside from trusting that they will not bend to pressure to terminate your service, you’re also trusting them with your network of contacts, because metadata (including the sender, recipient, and subject line) are not end-to-end encrypted in Proton.

Just wear disposable faces.

You humans wear the same face your entire life and then get upset when people recognize it?! Get over yourself! Aside from the obvious privacy issue, let’s be real: it’s also gross.

That can’t be good. But I guess it was inevitable. It never seemed like Arc had a sustainable business model.

It was obvious from the get-go that their ChatGPT integration was a money pit that would eventually need to be monetized, and…I just don’t see end users paying money for it. They’ve been giving it away for free hoping to get people hooked, I guess, but I know what the ChatGPT API costs and it’s never going to be viable. If they built a local-only backend then maybe. I mean, at least then they wouldn’t have costs that scale with usage.

For Atlassian, though? Maybe. Their enterprise customers are already paying out the nose. Usage-based pricing is a much easier sell. And they’re entrenched deeply enough to enshittify successfully.

Why? It’s Japanese and your browser should display it as マリウス. But I don’t know what that means.

Yeah, that’s true for a subset of code. But for others, the hardest parts happen in the brain, not in the files. Writing readable code is very very important, especially when you are working with larger teams. Lots of people cut corners here and elsewhere in coding, though. Including, like, every startup I’ve ever seen.

There’s a lot of gruntwork in coding, and LLMs are very good at the gruntwork. But coding is also an art and a science and they’re not good at that at high levels (same with visual art and “real” science; think of the code equivalent of seven deformed fingers).

I don’t mean to hand-wave the problems away. I know that people are going to push the limits far beyond reason, and I know it’s going to lead to monumental fuckups. I know that because it’s been true for my entire career.

If I’m verifying anyway, why am I using the LLM?

Validating output should be much easier than generating it yourself. P≠NP.

This is especially true in contexts where the LLM provides citations. If the AI is good, then all you need to do is check the citations. (Most AI tools are shit, though; avoid any that can’t provide good, accurate citations when applicable.)

Consider that all scientific papers go through peer review, and any decent-sized org will have regular code reviews as well.

From the perspective of a senior software engineer, validating code that could very well be ruinously bad is nothing new. Validation and testing is required whether it was written by an LLM or some dude who spent two weeks at a coding “boot camp”.

XMPP is still around! Despite Google’s best efforts to kill it.

The “free market” solution is for malpractice suits to be so ruinously expensive that insurance companies will apply sufficient pressure to medical practices to actually do their fucking jobs.

Same in the legal field, plus we should see a wave of disbarments already.

I’m not holding my breath. AI is shaping up to be history’s greatest accountability sink and I’ve yet to see any meaningful pushback.

All major browser engines are FOSS.

Chrome and Edge are proprietary wrappers around Chromium (BSD license). Firefox and derivatives are FOSS (Mozilla Public License). Safari is built around WebKit (LGPL/BSD).

The problem, however, is governance. These projects are all too big for anyone to realistically fork and maintain independently. So in practice, they are under control of Google, Mozilla, and Apple — all of which have questionable priorities (especially Google).

Perhaps this goes without saying, but if you want to do that, be sure you are installing Firefox through similar means across distros. This will not work with the Flatpak, for example.

Manual installation of Firefox is quite simple, and it self-updates, so that’s always an option.

Just gave it a try. I couldn’t get coherent results from img-to-svg with a few different tests of low-res pixel art and high-res cartoons. txt-to-svg also gave me incoherent blobs even with simple prompts. Something must be wrong there. Is it working for anyone else?

I might just try installing it locally when I get home.

Hard to judge quality when what we’re seeing is practically a pixel-perfect recreation. The tricky part of automated vectorization is detecting and plotting curves in such a way that it scales correctly. Bad implementations will use too many elements, or include straight lines that should be parts of curves, etc. Those errors would not be visible in those low-res rasterizations.

Wireless card readers are relatively new tech. I see them more and more as time goes on. New places usually give their waitstaff mobile readers, but there’s little motivation for older restaurants to upgrade their whole POS systems. POS systems have pretty long life expectancy. At least the older ones do.

Absolute nothingburger.

From F-Droid’s statement (emphasis mine):

We respect Tusky’s decision to block mentioned website; it’s their right to introduce restrictions like these into their software. We also respect Fedilab’s decision not to hardcode a login block; instead they are actively working on making it easier to block certain domains in the app itself and thus giving users more power to moderate which content they’ll see. If people disagree with F-Droid’s decision not to flag Fedilab, a idea is to develop a decentralized tag system based on package IDs which allow third-party servers to share their own warnings with their community.

They’re not blocking any fediverse apps that you can use to connect to any server.

Feel free to use any Lemmy or Mastodon client to connect to any hate-filled instances your hateful little heart desires. If you don’t like that some apps have built-in blocklists, use different apps, or modify them. That’s the beauty of open source.

Do what you want, just don’t expect the rest of the world to amorally do it for you. This is perfectly in keeping with open-source philosophy.

Note that even GrapheneOS no longer supports the 5a, since it is no longer getting firmware security patches. You’ll need to install an archived version of GrapheneOS, or go for something else like LineageOS.

With LineageOS, you’ll continue to get OS patches, but you’ll be stuck with an unlocked bootloader, so there’s a security tradeoff there.

I mean that an individual folder will always look the same (consistent), and also look distinctly different from any other folder (unique) if that’s how you arranged it. So you could identify a folder instantly.

Everything in list view looks the same at a glance, and most file managers don’t retain a folder window’s size and placement. Modern macOS kiiiind of does but you have to fight it if you don’t want a single-window browsing UI.