• 4 Posts
  • 7 Comments
Joined 2 months ago
cake
Cake day: February 5th, 2026

help-circle
  • If you have all the AppArmor patches and use a custom snap store, I believe so. There’s some inefficiencies with flatpak that are currently ignored. For example, every flatpak app has its own bubblewrap processing running, though they are light on resource usage. However, inter process communication is really inefficient, there’s a lot of context switching. You have the app talking to the dbus proxy and the proxy talks the real dbus (there might even be a step between the dbus proxy and real dbus).

    Meanwhile, for snap, this security stuff is handled by AppArmor security profiles. There’s no need for a dbus proxy.


  • That’s part of what I mean. Snap could be so much more interesting and useful if not for Canonical doing stuff like only allowing one store and slacking on proper support for non-AppArmor distros.

    One of the more bizarre experiences I’ve had is that a Canonical employee packaged a version of a Minecraft launcher. It was absolutely garbage, didn’t even start. The first thing that comes to mind is that snap is just garbage. But for fun, I made my own package of it, and it just worked perfectly. Which just leaves me the question of why a Canonical employee who works on snap can’t create a good snap package.

    There’s also the weird fact that Ubuntu dropped the ball with its core24 runtime. For some reason, Canonical’s own snaps stuck to core22 up until this month. Like, why wouldn’t they upgrade to their latest runtime? If there was an issue with it, why has it been broken for 2 years? Doesn’t inspire trust.


  • I can understand MIT being an issue in some cases. For example, VSCode is a proprietary fork of the MIT open-source Code. If Microsoft wanted, they could stop publishing the MIT open source version. Of course that code would still exist as MIT, but development would slow down without Microsoft.

    But I don’t see uutils being MIT as an issue. It’s primary goal is to be compatible with GNU coreutils. You can’t really rug pull a project with a goal like that. And permissively licensed utils have been around thanks to BSD and it’s never been an issue. You don’t see companies like Apple using proprietary forked versions as benefit. The “value” they add is higher up the tech stack with their own truly proprietary stuff or open stuff that encourages lock-in to its ecosystem, like Swift.







  • It certainly has it’s issues. Takes up quite a bit of space since each app tends ships its own copy of electron (though distros like Arch do try to make apps share a single Electron build). Apps may ship out of date versions that may have security vulnerabilities, though it’s not always the end of the world since they tend not to access outside of their own domains. As for slowness and resource usages, it’s bit of a tricky subject; an Electron app can be optimized, but will always use quite a bit of RAM.

    Though undeniably they have been beneficial for Linux if only because it allows some companies to support Linux without too much extra work.