Skull giver

Google’s “just so we can claim there’s competition in the marketplace” payment to Mozilla to stay the default search engine is what funds most of Firefox, even without the executive pay and unrelated nonprofit work. Building and maintaining a browser engine is not cheap, I don’t think mere individual donations are going to help here.

Without the reputation and contacts of Mozilla, those devs also wouldn’t have much of a say in the social side of browser development, like web standards and certificate authority programs.

I’d love a first party Firefox fork that’s not limited by Mozilla’s desperate attempts to stay afloat when Google decides not to keep them around anymore, but I don’t think a few developers are going to be enough to get it done. The current situation, “Firefox but with very minor tweaks”, is probably the best we can expect for now.

Everything but a few proprietary, business focused modules in the backend (like managing multiple organisations) is AGPL licensed. Unless you’re a business, you can probably make do with just the open source code. They’ve even included a compile flag to disable all proprietary code. The clients are all GPL-licensed as far as I can tell.

You can also run Vaultwarden as the backend, which is a third party server that takes a lot less RAM but isn’t suitable for hosting thousands of active users at once. I also don’t think it has been audited, unlike the Bitwarden code. Great option if you trust them as much as you trust the Bitwarden company to maintain security.

Because this is a response to a post calling NT an “obese slog”. Compared to Linux, it’s almost anorexic.

Plus, these drivers do cause plenty of issues, like that time a patch in an early kernel NFS driver caused PCs connected to a DisplayPort-to-HDMI-converter connected to certain Intel GPUs to hang while booting for several months. I’ve had to pin an unmaintained kernel for months until Intel finally patched their driver against that, it was a real pain, especially with everyone around me telling me to just install Windows like everyone else whenever this caused package conflicts again.

When this stuff becomes a problem, the kernel often becomes entirely useless and very few people will go through the necessary troubleshooting to get their computer working again. It’s easier to clean install another OS, be that a Linux distro with a different kernel version or Windows.

Pretty funny to quote the GDPR consent description when the website shares your visit with Google without your consent (through Google Fonts), as well as various other third parties.

One important addition to the considerations: don’t join ActivityPub if you’re successful or popular. The way the community responded to BridgyFed showed that large parts of the Fediverse want it to stay their own little obscure corner away from most of the world, and that doesn’t even touch the whole Threads thing.

There are exceptions to this rule, like being a Cool Internet Company such as WordPress/Tumblr, where forced login prompts and ever shittier paid subscriptions are tolerated because these companies have a historical cool factor that other companies don’t have.

As for legal consent, make sure you know the legal obligations that come with consent if you run a product or service as a company or organisation. You may be required to tell end users what profile information you’ve shared with what servers when, that you’ve encrypted your database, and may be required to enforce deletion of PII on remote servers as the primary responsible party for protecting your users’ privacy when your users withdraw consent.

Or you can just ignore the law, which most of the Fediverse seems to do, including several government servers.

Modprobe or not, my computers still scan for GPUs on the EISA bus. Not everything is loaded, but tons of unnecessary stuff is loaded just in case, like ancient PS/2 controller support and obscure filesystems. Installing usable drivers can even land you in a situation where two drivers fight for control over a device (Nvidia again), necessitating kernel flags or blacklists to prevent builtin drivers from loading.

Plus, even if they’re not loaded at runtime, drivers for hardware I’ll never encounter still take up space in the kernel. Impossible to prevent with Linux’ kernel architecture and barely a problem in practice (unless you want to boot a microcontroller or want to use Linux as a bootloader).

NT is actually pretty great. The Windows GUI may have gone to shit, but the underlying kernel is great. I’d even argue that it’s ahead of Linux in many respects.

Linux is a slog comparatively, coming with hundreds of packed drivers for machines that stopped being sold a decade ago, unless you’ve tweaked your custom kernel config to only include drivers for your specific system. Windows has its fallback drivers, but most of them are downloaded on the fly rather than being precompiled into the kernel. This is part of why Nvidia drivers are such a pain to deal with on Linux.

The Windows scheduler and the Linux scheduler deal with processes just fine. Windows deals with hitting memory limits way better, but Linux has more flexibility to control the CPU scheduler. I also find Linux to be less efficient with file system caches, but that’s probably because Windows takes forever to complete I/O operations because of NTFS. Windows will fill your RAM with stuff you may need, while Linux happily keeps gigabytes of RAM unassigned (and act all surprise Pikachu when you actually request the browser that you open literally every time you boot your PC).

Linux doesn’t do antivirus, that’s the biggest difference. You get infected more easily, but you get faster I/O in return. This is especially the case when accessing tons of tiny files, like when booting the computer or programming. The load is relatively small when loading games and such.

I find Windows to be a lot snappier with my iGPU in power save mode, while Gnome and KDE are snapper when the iGPU is enabled. Video acceleration make or break Linux DEs much more than Windows in my experience.

I also find Bitlocker to perform a lot better than standard LUKS2, especially during the early boot process. The Windows bootloader isn’t restricted in its access to encryption acceleration functions the same way Grub is, so unlocking disks with similar cryptographic strengths at boot time is just faster on Windows. Plus, hibernating with encryption is possible without hacks and disabling security features in Windows, which is why it boots so fast (shutting down hibernates the kernel unless you need updates).

Linux is generally faster at updating (though using Flatpak GUIs would have you think otherwise), which is the biggest speed concern I have with Windows today. Perhaps it’s to make System Restore actually usable (something Linux can improve on) but it takes forever to install minor updates. Maybe it’s related to NTFS as well, which isn’t too great compared to the Linux alternatives on offer.

Windows is also terrible if you’re still running from a hard drive. With Windows 11 I’m pretty sure the devs abandoned HDD support all together with how slow it boots on spinning rust. A real pain when using virtual machines.

If you notice an immediate difference between Windows and Linux, it’s probably because you’ve recently installed a fresh copy of Linux. My Ubuntu and my Windows partitions boot in about the same amount of time. Give it a few years of gathering cruft and you’ll probably have an equally slow Linux install.

Caches don’t count as free RAM. Caches are available RAM, but free RAM is just space waiting to be filled with caches.

If you clone the entire disk, the partition UUIDs remain the same.

You should probably change them and update the files so the unique identifiers are actually unique, but you don’t strictly need to.

Android does some estimations based on battery behaviour to make the percentage display more accurate.

This is just the user facing component, of course, but “50%” doesn’t mean much if the displayed percentages aren’t compensating for an older battery losing the last 25% of its charge in a few minutes because the cells are degraded.

I don’t know if there’s anything like that on desktop Linux, but I certainly wouldn’t say calibration isn’t a thing anymore. It’s just done automatically and hidden from the user.

If you’re buying a new SSD, you probably won’t need to worry about this. Just make sure to boot the installer in UEFI mode (usually the default). In the worst case, you’ll need to go into the motherboard settings and put Linux above Windows in the boot menu.

Only in MBR mode. On computers less than 10 years old, you can go into the UEFI settings and put Grub above the Windows bootloader in the boot order list.

Some broken computers only boot the “flash got wiped, let’s hope this one works” fallback bootloader. Windows and Grub will fight for that one, but that should only happen if your motherboard’s UEFI settings got reset or if the firmware is buggy.

Aside from Edge, the Netflix app will also work. Basically, Netflix doesn’t send you a 4k stream without PlayReady or an equivalent level of DRM, which native apps typically also will (though your experience may vary on Android if you buy cheap Chinese phones).

There are addons that will get you 1080p on Linux instead of the 720p that you get by default. They break every now and then, but at least full HD is possible without too much effort.

This is because of the way Lemmy solves the “missing responses” problem. ActivityPub doesn’t mandate a specific way to federate likes and responses, but the way Mastodon does this causes every server to have different replies and likes.

Lemmy forces the comments and likes to federate by boosting every one of them from the community actor, therefore ensuring that every server sees every comment. On Mastodon, you only see comments made by people someone on your server follows.

This leads to another rather annoying Mastodon problem, where someone will ask a question or pose a problem and will get a million of the same replies because everyone replying thinks they’re the first to come up with an answer (as their server shows a small subset of replies). It can look like a storm of reply-guys and probably doesn’t help Mastodon gain any popularity.

Lemmy and Mastodon both do this correctly from a spec point of view, but it’s an example of how two correct implementations of the same spec can still have trouble interoperating.

I don’t think they exist. The drivers that don’t load firmware blobs into the WiFi device just come pre-packaged with (probably outdated) firmware blobs. Very few devices work without firmware.

You can add a layer of isolation but hooking your device up to a random access point over ethernet, though the experience certainly won’t be as nice.

I think there are also (incomplete) attempts to write fully open-source firmware for WiFi chips like the ESP32, but I don’t know if anyone ever wrote a fast interconnect for the standard dev boards for that. You may need to set up your own PCB to turn those into a fully open source WiFi chip. Performance will be very limited, of course (10-20mbps) because these IoT oriented boards lack hardware processing.

Could be hardware (chips desoldering themselves, data cable shorting out when heating up, broken RAM), could be software (driver issue, IOMMU configuration, etc).

I would start with testing the memory chips. If those pass, I’d try a much newer/older kernel to see if that makes the problem disappear. If that doesn’t, maybe try running the OS that came with the laptop and verify that it’s a hardware issue. If it doesn’t happen on another OS, it’s probably some form. Of software issue; in that case you’d need to catch the crash somehow, like by plugging a USB-to-serial converter into a USB port, dumping the kernel output to that, and having a second machine monitor the kernel output while you try to trigger the crash.

If all operating systems have the issue, reseating RAM and perhaps any important cables may also help.

Librewolf would need to ask permission to a folder (for the standard downloads folder for instance) or it would need to show two save prompts when downloading two files (isn’t that what it does already?)

The “two files” thing only applies to applications that ask access for one file (say, an mp4) and also want a second file in that same directory (say, a matching .srt). That can be worked around by selecting multiple files in the file picker, but that does pose for an annoying restriction. I don’t see how a browser would be affected by this, though, as browsers don’t tend to also send secondary files when you upload something.

but if I want to use it to open a file that isn’t in “downloads” I have to use flatseal to give it extra permissions

There has been a portal to prevent this issue for years now. The fix isn’t to patch around issues in Flatseal, it’s for developers or Flatpak packagers to fix their security policies and code.

As an added benefit, KDE users get thumbnails in their file picker because they’re no longer stuck with the old GTK one but instead can use their native file picker portal. A win for everyone!

I think they’re a move in the right direction.

Just looking at the weird scaremongering around Signal from the past few days ("a chat app stores keys as files that you can read) shows a trend that I’ve been seeing more the past years: people have gotten so used to the Android/iOS sandboxing system that they’ve either never been taught or have forgotten how normal programs work.

Flatpak and the necessary desktop portals are very much a work in progress when it comes to user friendliness, but they’re what the world has been moving towards for a while now.

I don’t know why a journaling app needs full system access and access to system settings, and the permission Flatseal requests is a dangerous one if you pay attention to these things. Looks like they’re doing their job to me.

I think this is the more worrying part if true. The backend is licensed under the AGPL, so this would technically be a violation of their terms

The AGPL doesn’t require you, the author, to do anything. As the copyright holder, you decide the license your code falls under. You publish code with a license so others can use it. You can always do with your own work on your own computers as you wish, assuming you don’t also use other (A)GPL code that forces you to release your own.

Many companies sell GPL software this way; the (A)GPL version is free to use, but if you don’t want to share your alterations and any code you integrate the (A)GPL code with, you pay money to get a non-AGPL licensed copy. Qt does this, for instance, so car manufacturers can design their closed source vehicle dashboards and open source projects can use Qt to build a Linux desktop.