I really despise the idea/existence of app stores.
Why can't a publishers website be their store front, then from that site, you can run (or "download"/"install" if you must) the app?
Technically it looks like that would actually work [1]. They could host the `.snap` file. Although I'm sure snap will add friction to this (as in, not simple let you open and click "Install") from a downloaded snap file, as they probably want you to use their app store.
For me, it's updates... I don't want popups telling me to update software every few days, and then opening a webpage, downloading a snap/deb/whatever, installing it again, with sudo passwords and all that,... and then open another app, another popup, another site, another .deb... It feels like windows.
I want a centralized "update" feature, that will do all that for me.
This can be done by signal adding its own repo, but it's a pain to add a per-app repo, if the app can be on the central (distribution-managed) one, especially if it's a popular app.
Not sure if your post was satire, but isn’t everything you mentioned doable with apt already? Signal already has an official apt repo so all you need to do is the standard apt upgrade and everything including signal would be updated.
The problem with that is now you've given a 3rd party power to update your system in any way that they want. Which is fine when it's just signal but less ideal when it's 20-30 randomish companies of varying degrees of trustworthiness. The point behind flatpaks et. al. is sandboxing applications so that they can be safely updated automatically.
I don't use apt but I think you can use the pinning feature in apt and only allow the Signal application from Signal's repo. It doesn't solve all problems since they could add dependencies from other repos, but at least it is partly stops them from adding their own dependencies in their repo.
Are they, technically, any different than APT repositories?
Don't get me wrong. I hate Snap much more than the next guy, but the idea of keeping a repository so you can go look for, and discover, stuff that is supposedly also vetted by someone is nice.
The issue is, when an app store is a monopoly and not standardized.
APT repositories is, in my eyes, an example of "the good" type of app store.
Technically not really, instead of packaging the software into a .deb it’s shipping a container and metadata to get it working.
I think the rub here is that the maintainers of the signal snap can't reuse the work they’ve already put in to this and offer it elsewhere in another snap repository as there’s no other snap repository possible
I always say that one of the problems in the Canonical's strategy with Snap is not to provide the backend allowing everyone to set their own "store". Then, Canonical would focus in the added value of their one, with things such as a payment processor, malware scanning, ci/cd integration, stats, etc making it appealing.
Flatpak allows this, but, in the other hand, despite I exclusively use Flatpak these days, I dislike the fact that is not focused in CLI apps but desktop ones instead.
There is a proprietary piece of software that is released on Snap Store only. They used to release a .deb file. When they did that, it was as easy as extracting the files from the .deb file and run the executable. With Snap, however, it is much more of a hassle to do the same thing. It is possible (at least for the one I have in mind), but yeah, no.
Do Linux users installing software actually have a "discovery problem"? The usual solution is to go to google.com and find in-depth information on the options available for a given class of software, with user reviews, notes on strengths and shortcomings, etc. I have never visited a store first on a desktop OS, and even average users of mobile OS' that I know will fire up Google first if they're actually trying to find the best implementation of X app. I don't know a single person that has anything good to say about the Windows app store.
The Year of the Linux Desktop isn't now, and it may never happen. Even if it does, I don't think any of the hypothetical "normie" Linux users would really care for it all that much. Trying to be the devil's advocate for things like the Snap store benefits a group of users that doesn't exist in any timeline at the expense of current users in this timeline.
Absolutely, when I want to find something, the first thing I will do is DNF search. It's amazing how much you can find and so quickly. It sorts in a smart way as well, where exact matches will appear first, then partial matches on the package name, and then metadata. This makes it quick and easy to see if what I want is there.
When searching on Google, I have to look through a bunch of crappy results from SEO optimized crap. Then when I find something that looks interesting, I have to see if it's packaged for Fedora or not. If not, then I have to hunt down either a GitHub page or look for a RPM file or a flat pack or app image. I won't use a snap even if it's the only distribution method.
There's also a very convenient way of checking for popularity, which doesn't necessarily mean it's a bad pic, but if the package is popular enough that somebody has packaged it for Fedora and maintains the package, there's a very high chance that it will work as needed.
If you don't currently check the apt or DNF repos first, I would definitely recommend doing so.
I do need t think discovery is the same problem any more. You can search for self hosted snaps about as well as you can within the snap store. However, trust can be an issue which has solutions that come at the cost of speed and storage.
Yea, don't bucket Linux users in with typical Google and Apple store users. I need no help "discovering" software through some companies arbitrary filter. I can use google, github, hacker news, etc.
That's a different problem to where the packages are hosted. You can host your own packages and get added to an official aggregator. Said aggregator can function as an automatic verification system to assure people that your software is not compromised, has been updated recently, etc.
Ideally it would be handled like BitTorrent/Magnet/IPFS:
- Links resolve, and data can be fetched, as long as anyone is hosting it (regardless of who)
- Anyone (including big orgs/corps) can provide a curated lists/search for links they vouch for
- Removing something from a list doesn't affect anyone's ability to fetch and use it; only whether they see the link in that particular search/list UI
You don't already do this for Linux? :( A new Linux setup already requires me to visit a bunch of sites and find .debs. Chrome, Signal, Edge, Visual Studio Code, Slack, etc. I don't really trust random unofficial maintainers in snap store and all of the above is required for my work.
It isn't impossible to have both. An app store that is a registry pointing to packages/files hosted by the developer. And since you mentioned Windows, that's exactly how choco and winget work.
Why can't a publishers website be their store front, then from that site, you can run (or "download"/"install" if you must) the app?
Technically it looks like that would actually work [1]. They could host the `.snap` file. Although I'm sure snap will add friction to this (as in, not simple let you open and click "Install") from a downloaded snap file, as they probably want you to use their app store.
1. https://github.com/snapcrafters/signal-desktop/issues/70#iss...