Ubuntu Desktop: charting a course for the future
ubuntu.com
ubuntu.com
I miss the dreams of two decades ago. It was like a cause to fight for, a bunch of hackers against a billion dollar industry. Actually, we already had support from part of that "billion dollar industry" but there was still a fight against antagonists. Every contribution was a small step, every new version, new improvements, a new program covering a feature we missed from some evil proprietary software... The points where we were ahead like security, stability and performance made every enthusiast user feel like part of great opus, a soldier in a somewhat silent war.
Fast forward 20 years, linux is now "the kernel that powers billions of end-users devices". We're still way behind on the desktop, but the "war for the end-user" has a clear leader. Many other markets were also taken entirely or partially: HPC, embedded, auto, entertainment, routers, servers, cloud... The thing matured. If you want to get a contribution in the kernel, you'll have to convince the "billion dollar industry" we seemed to be fighting against in another era. You could say they are on our side now, but I really don't know what difference it makes. Those teens of two decades ago should be proud of what was achieved and, I'm sure, many are.
The situation is good, stable, mature. Always a bit better. But the feeling, the enthusiasm, the hope... all that seems gone, and I sorely miss that. I'd love to have that feeling again. The feeling that heavy investments will be made, the feeling that the desktop is a good cause to fight for, that a small circle of hackers are willing to make something BIG.
Yes, I miss that.
What if this framing was holding us back the whole time?
The "other side" did indeed act in probably unethical ways that hold us back. Legal threats, defamation campaigns, lobby... All that was used by "the other side" and that really did hold us back. The SCO scandal and Halloween documents are just very good examples.
With reagard to the "bunch of hackers"... well, those were just trying to write awesome software. And they did.
User-respecting mobile OSes and hardware are still as unusable as the desktop Linux of 20 years ago, the incumbents are not helping at all, and the stakes are even higher (users get their freedom violated 24/7 instead of only when they sit down to their desktop).
I went looking around the web to understand why people dislike Snap in the first place. The top reasons cited on Reddit are:
(1) "Controlled by Canonical, not an open format" (This seems to be a mis-explanation of the complaint that the Snap server itself, which Canonical runs, is closed-source. As though a Snap server would be hard to create, it's a simple CRUD webapp) Anyway, this in no way causes problems for the "casual users" I mentioned, so irrelevant (unless the people who claim that closed-source software is invariably hopelessly buggy compared to open-source are right, that is).
(2) "Ubuntu reinvented the wheel, Flatpak was fine" But does that affect the "casual user" in any way? People are still going to package software for Snap. The casual user isn't going to miss a minute of sleep over their packages coming from a store that existed after a perfectly-good store already existed. It's just not top of mind at all.
(3) "Ubuntu did a bad job designing the system. You have to close Snap apps to update them, but the updater doesn't have a friendly UI for killing apps so they can be updated, which just confuses the average user" (This seems valid. Flatpak updates the app in the background and the new container isn't launched until you quit the app and reopen it normally).
(4) "Forced updates" (This is actually a positive for most casual users, who don't want to think about updates, they just want it to happen).
For instance if it works like channels its virtually impossible to charge a percentage like Apple's app store because people will just publish to a free channel, every distro will include the free channel, and users will just ignore the paid channel. Consider the paucity of software on the Mac app store vs the iphone app store.
1B. There will never be an open source Snap store backend because it would both be more work than you imagine and continued work to retain compatibility of both front and back. If you try you will most likely end up with something that sometimes work sometimes fails and takes a full time maintenance. You may as well call it "snap but shittier how I burned 200k figuring out how moats work"
2. The problem isn't merely re-inventing the wheel its reinventing it for no reason and doing it poorly then not integrating it in a fashion that makes it easy on the average user. The casual user of ubuntu may end up end up needing traditional packages, snaps, flatpaks, ppas. If I understand correctly the current state of software management they can easily install both traditional and snap packages but will have to turn to additional cli tools to integrate ppas and flatpaks.
Arch with the AUR enabled provides funny enough an easier and simpler experience.
3.Agreed
4. The is a positive wherein new versions never fuck things up or remove features that users rely on. This doesn't seem to be the current state of play and pinning a version or launching an older version seems to be table stakes for dealing with flawed software.
Putting lxd/lxc in snap also baffles me. Layers upon layers of abstraction and virtualization for no reason. But that's a different story.
And yeah, it was because of Snap.
Ubuntu suffers something that effects windows really badly. In an attempt to polish, you obfuscate a lot of moving parts. When one of those parts break down, it becomes far harder to repair.
I also understand the irony of wanting people to switch to Linux then being against a polished distro.
Ubuntu Snap wasn't the only reason, but one of them.
I recently gave the new Debian a try on my laptop, and I've been happy with it so far. I've also been pleasantly surprised that it's noticeably snappier than Ubuntu for some reason.
My main machine is still running Ubuntu, but when the time comes to reinstall or upgrade, especially if the snap situation becomes even worse, I will probably switch to Debian entirely.
I gave away my gaming computer to a teen whose mainboard broke. I don't miss Windows one bit.
Interestingly, this doesn't show in live mode, only after install.
Honestly, this whole "if it's not a .deb it's a nightmare" mentality is baffling to me.
1. keeping the store part of it closed. 2. forcing the use of snaps on me.
Those two feel really off to me, like a poison-pill.
Where do they see that? Opening the link they reference gives the following table:
"SteamOS Holo" 64 bit 42.07% +2.74%
"Arch Linux" 64 bit 7.94% -0.39%
Ubuntu 22.04.2 LTS 64 bit 7.38% -0.49%
Freedesktop.org SDK 22.08 (Flatpak runtime) 64 bit 5.99% -0.03%
"Manjaro Linux" 64 bit 4.29% -0.08%
Linux Mint 21.1 64 bit 3.84% -0.71%
Pop!_OS 22.04 LTS 64 bit 2.97% +2.97%
Other 25.52% -4.02%
* My wifi used to slow down multiple times a day, and speed up again after I turned it off and on.
* Right-click context menu in Firefox often failed to copy links to the clipboard.
* The file dialogs in Firefox became weird after an upgrade a year or so ago. I could no longer hit Esc to dismiss them. 50% of the time, pressing ctrl+o to bring up the file open dialog and continue spamming o's into it until I killed Firefox. Or it would spam ctrl+o's spawning ever more file open dialogs until I killed Firefox.
I blamed Firefox for many of these. But now I'm not so sure. Manjaro has been great.
I didn't realize the extent to which software can vary at all levels between distributions. Now I think of distributions like different places with different weather. If you don't like the climate in one, try another one (maybe just off a USB stick). It might be very different.
Why does Canonical always reinvent the wheel. Just ship ZFS or BTRFS please.
as an option in the Ubuntu installer
So yeah they'll probably be using systemd-cryptenroll under the hood, but just make it more accessible for end users.
Also Snap push is huge tun off for me. My next install will be Debian.
- HUD/search via global-menu - You could find any option in menus by typing them (https://www.omgubuntu.co.uk/2012/01/hud-new-unity-feature).
- Menus in the top bar (like KDE/macOS), but also in the title bar of application (quite unique approach - https://ubunlog.com/en/ubuntu-14-04-menus-in-the-title-bar/).
I guess they need to say something other than "we're the systemd/dpkg/GNOME distro that everyone's familiar with".
ZFS has been an installer option since 20.04 and it's been working great for me.
https://ubuntu.com/blog/zfs-focus-on-ubuntu-20-04-lts-whats-...
Maybe they will improve the LUKS user experience.
So Ubuntu now is just a "goto" for newbies not anymore "something good for both expert and newcomer" so have a fate already depicted. Sorry.
NisOS probably will be the future Ubuntu in 4-6 years.
I'm glad to see the Ubuntu Desktop project move in this direction - it strikes me as a mark of maturity.
Congrats!
What is the desktop going to be?
Seems like just a list of empty feelgood to me.
* Looking at you, Nvidia
Also, to be clear, my asterisk is that Nvidia drivers seem to be no easier on Ubuntu than on Debian (you have to seek them out), and for me they've often caused problems once installed.
It always felt like going through the distro's included stuff was a waste of time.
There is firmware in the main ISO, which means your WiFi might work. The Nvidia driver is an `apt install` away.
Either way, Nvidia needs to be installed by you on Debian or Ubuntu, and GUI vs CLI is no different to me.
And yes, I was calling out Nvidia as the exception to "it just works" even on Ubuntu. I realize that wasn't clear.