We are also changing how remote playback works for streaming personal media (that is, playback when not on the same local network as the server). The reality is that we need more resources to continue putting forth the best personal media experience, and as a result, we will no longer offer remote playback as a free feature. This—alongside the new Plex Pass pricing—will help provide those resources. This change will apply to the future release of our new Plex experience for mobile and other platforms.
Part of it is that Ubuntu/Canonical so aggressively pushed Snaps which became a huge culture war. So you have people who hate the idea of those style of packages because they hate Snap AND people who hate flatpak because they are Team Ubuntu for some reason.
And the other aspect is that it is incredibly space inefficient (by the very nature of bundling in dependencies) and is prone to “weirdness” when it comes to file system permissions and the like. And many software projects kind of went all in on them because it provides a single(-ish) target to build for rather than having a debian and an arch and a redhad and a…
Ah, I see. I’ve not tried Snaps, been avoiding Ubuntu because of Canonical’s weirdly corporate angle. Once they baked in Amazon into Ubuntu I was out.
I like the bundling of deps. Sure it’s inefficient, but it runs, and storage comes cheap nowadays anyway.
Storage is cheap until it isn’t.
On my desktop where I have something like 6 TB of NVME storage because I am a sicko? The only thing that makes me think twice about a flatpak is if I need to give it access to devices or significant parts of my filesystem (yay permissions weirdness).
On my laptop where I can have one drive and replacing it involves opening the entire laptop AND reinstalling Fedora (or dealing with clonezilla/
dd
)? Yeah… I very much care about just how much bloat I am dealing with. And, as the other person pointed out, flatpaks can balloon REAL fast.A lot of flatpaks early on wouldn’t survive a major point release upgrade or worst case would hold on to dependencies and the user would end up with an unbootable mess after an upgrade.
I haven’t seen that recently though.
However I regularly run appimages on my fedora silverblue system so take what I say with a grain of salt.
If dependencies are articulated (and maintained…) properly, it is very doable and is intrinsically tied to what semantic versioning is actually supposed to represent. So
appfoo
depends inlibbar@2:2.9
and so forth. Of course, the reality is thatlibbar
is poorly maintained and has massive API/header breaking changes every point release and was dependent on a bug inlibbar@2.1.3.4.5
anyway.Its one of the reasons why I like approaches like Portage or Spack that are specifically about breaking an application’s dependencies down and concretizing. Albeit, they also have the problem where they overconcretize and you have just as much, if not more, bloat. But it theoretically provides the best of both worlds… at the cost of making a single library take 50 minutes to install because you are compiling everything for the umpteenth time.
And yeah… I run way too many appimages too.
The space inefficiency is definitely there.
I find that clients, such as Jellyfin, Moonlight and Signal, works just fine as flatpaks but with those three apps my /var/lib/flatpak/ lands on 6.4GB.
When I temporarily had Discord installed it grew to 6.7GB, so the inefficiency is frontloaded and lessens the more of them you use.