Ubuntu Is a Canonical Product
utcc.utoronto.ca
utcc.utoronto.ca
I’ve been running a couple of Fedora versions as desktops on fully supported hardware and, oh my, they were quite bad in terms of stability (bluetooth not working, video calls breaking), speed, and general UX when compared with Ubuntu.
Sure, distro X, _when properly tuned_, will be better than Ubuntu. But a lot of users, even power users, are in Linux to do their job, which is NOT tuning linux distros.
Ubuntu OOB experience right now is still the best for a Linux imho (pop os actually, but it’s an ubuntu spinoff). I tried Debian about 1y ago and its desktop defaults were very cheap, requiring a lot of work as a user.
Open to changing my mind. But please, let’a discuss a default experience, not a ton-of-tuning-patches-installs experience .
> Fedora versions as desktops on fully supported hardware
Is there a list somewhere, as I have not seen this. (Seriously).
> oh my, they were quite bad in terms of stability (bluetooth not working,
Fedora has a practiced behavior of fixes getting upstream first before shipping, (rising tide lifts all boats). If bluetooth was broken for fedora, it likely would be broken across multiple distribution, and the ubuntus would have had to do something that wasn't upstream.
If ubuntu ship patches different than upstream, then it just come down to a timing issue of when your specific hardware has been fixed.
I can talk more about the other points you mentioned, They seem more anecdotal than provable (Which is fine, perception is 9/10ths of the law).
hardware is usually some Lenovo Thinkpad (X and P series) that work with RH. Now that you tell me, it seems that Fedora has no official compatibility list. But Ubuntu has it. Isn’t that a Fedora problem?
> Fedora has a practiced behavior of fixes getting upstream first before shipping, (rising tide lifts all boats). If bluetooth was broken for fedora, it likely would be broken across multiple distribution, and the ubuntus would have had to do something that wasn't upstream.
That's exactly what I mean about Ubuntu being a desktop-oriented _product_. In that instance, it was a kernel problem. Ubuntu clearly has this idea of testing for desktop features, and probably not releasing (or patching) something if it's a problem for desktop users. Fedora just pushed the new kernel through, and probably helped fixing the issue within the kernel. But it's not a great end-user experience.
Btw, of course this is an opinion. I don’t have hard long-term, wide-population data on the subject.
Switch to Ubuntu and you will start hating it just as much.
Work laptop had a problem with a broken kernel delivered. I also always lag one major fedora version behind for safety, which goes against the philosophy, but this is mostly because gnome api for plugins is highly unstable and I really depend on some of those to do my work. I wish I went with KDE.
On the personal laptop I use Kubuntu (ubuntu with KDE). The UI has been consistent and stable, for mostly 6 years. I love that aspect. Yesterday I upgraded it and I got a broken kernel (panic on boot).
Overall I faced the same issues across the two in 6 years (ubuntu) and 2 years (fedora).
With Fedora I have the latest software usually, which is a blessing or a curse.
With ubuntu i have slightly outdated software, which is a blessing or a curse.
I like both distro, I'm also used at the rough edges of desktop linux.
Thinkpads seems to work out of the box with Fedora.
The Ubuntu laptop is a system76 laptop, before popos was a thing.
Has system76 never realized how unfortunate their OS name is?
I dunno, I just didnt know how you came to the conclusion that it was officially supported or on any list, when I have never seen any list.
Should fedora do a 'hardware compatability 'list', would it change developer behavior, I doubt it.;
These things have little to do with the distro. They should be identical, barring version differences. Maybe you had breakage with Fedora because of bleeding edge stuff, but the other way round is equally likely, where Ubuntu LTS will be too old for new hardware. Once your hardware is 1-2 years old, all distros become largely identical. And if your hardware is already old, RHEL/Rocky/Alma will work too.
- There are also configuration differences, ubuntu can blacklist some hardware for certain features.
- Ubuntu also gets bug reports from users and customers, the developers will apply the fix in Ubuntu but it might take time until the upstream project applies the fix, or even if upstream applies the fix it will only appear in the other distros in the new release unless the distro has the energy to backport non security stuff.
Fedora releases every 6 months.
Just keep using $latest-1 and enjoy a ton of fixes.
Windows is there for the odd games Steam didn't port to Linux, and to compartamentalize my son's use from mine. Heck, I hacked around less in Ubuntu than I did back the day installing Windows 2000, or to figure out how to make a Win 10 Pro instance run with local user accounts only.
And battery life on my X1 Extreme Gen 2 is bad under either OS, I deactivated hybrid graphics and only use the GeForce card. Otherwise, I just never got dual / external monitors to work through Thunderbolt. Plus, it is used as a gaming device, so I don't care about on-board Intel graphics at all.
So let people whose job does include 'tuning' operating systems handle that. Does your company not have an IT department or something?
> Open to changing my mind. But please, let’a discuss a default experience, not a ton-of-tuning-patches-installs experience.
Again, where is this double-standard coming from? Since when do fleets of Windows laptops run without any 'tuning'? The work it takes to get a decent UX on Linux given known hardware, selected for the purpose, is extremely trivial for anyone who knows Linux at all.
It's not hard to master your own install media for most distros either. Buy your shit from a decent Linux hardware vendor and they'll even install whatever custom OS you want before they send you the machines.
This idea that a whole-ass company should be somehow helplessly captive to distro defaults is pitiful. Hire someone with decent Linux competency and prepare a suitable OS image! It's what you should be doing regardless of the Ubuntu question.
Yes. It's me. And I'd rather spend my time doing my actual job rather than tuning linux distros.
My Ubuntu systems runs fine without tuning. My MacOS laptop runs fine without tuning, my company just installs its fleet management tools, but there's no actually tuning involved, just a couple of policies.
I don't really know what to make of large Windows fleets and I don't use Windows since eons, but most laptops work OOB nowadays (barring removing some adware).
BUT: I was comparing Linux-to-Linux, other OSes may be a bit different.
I agree that finely tuning a Linux distro, in a large company, could be feasible. But it can be tricky: you need to override repositories or pin packages, otherwise things can break... and it's not a job anybody would like to do.
And yes, some people used to do manual patch management for Windows workstation deployments, once upon a time. I don't know if it's still a thing, but it was very messy and the famous "mixed-patch environments" were a boon for attackers (some vulnerabilities could only be exploited in those unpredictable scenarios).
I want my basic OS experience to be managed by whomever creates it.
Last week I reinstalled my laptop and only restored my /home folder. All good.
Maybe I don't understand what you mean by 'tuning' here. The only tuning I can really see required if one chooses, e.g., Debian over Ubuntu, is the installation of proprietary and patent-encumbered codecs and drivers that are not legally redistributable for community distros.
And can it really even be said that Ubuntu works free of tuning? I support a baby boomer relative on a Kubuntu system I built for them years ago. Upon upgrading to the latest release, I had to do some clunky apt pinning shit to purge Snap from their system to get their browsers working normally again after Canonical snapified them.
> you need to override repositories or pin packages, otherwise things can break...
This is not hard. I learned how to maintain a Linux repository in a few weeks at the age of 18 because I decided it was lame to be restricted to choice of distro by package availability. The first repo was up in a weekend and by the end of the year I was comfortable packaging for Gentoo, Arch, Ubuntu, and openSUSE. I bet most people reading this post could learn to do it competently in a similar timeframe.
(Ubuntu and Debian here actually make such things harder, because apt is pretty much worst-in-class when it comes to managing multiple repositories.)
More importantly, don't you already need to do this just for things like security agents and monitoring tools anyway?
> and it's not a job anybody would like to do.
Why not? I would.
Maintaining an in-house repository for desktop Linux support is easier, more interesting, more satisfying, and potentially higher impact than many things IT and DevOps people do every day. Idgi.
> BUT: I was comparing Linux-to-Linux, other OSes may be a bit different.
The industry expectation is clearly that Linux users should do their own tech support... but that no one else has to. With a standard like that, of course many people are going to come away with impressions like 'I don't have time to run Linux on my workstation', 'only Ubuntu is really viable in a corporate environment', etc.— even when they are totally, totally false.
Strongly disagree. Installing Ubuntu is still a bewildering process for non-Linux-experts. It actually seems to have gone downhill in the last 10 years. Mint is quite a bit more understandable.
Debian Stable had fewer regressions in security updates, Ubuntu really dropped the ball on one regression that would make some devices unbootable (Ubuntu strangely silent on the bricking, and I had to confirm it with a contact at another distro company), getting rid of Ubuntu's Snap changing things outside of APT control was a huge relief, and we don't have to worry as much about privacy&security problems of Debian phoning-home.
The biggest headache with Debian was installing laptop workstations, due to WiFi usually needing firmware blobs. It required complicated instructions and contortions that made it a million times harder. But that's now solved: https://cdimage.debian.org/cdimage/unofficial/non-free/cd-in...
Also, especially for startups, there are often efficiencies to using Debian Stable for everything for which it can be used. For example, it's not the smallest base of a bespoke Docker container, but there are development and operations (personnel) efficiencies when containers, host servers, and workstations are all running the same thing. Fewer surprises in production, less time and complexity trying to support multiple environments, and you can rapidly debug some things more easily on bare metal on workstation when necessary.
And the official installer is going to provide the non-free firmware now (and non-free firmware will also be in a separate non-free-firmware section of the repositories so you won't need to enable the whole non-free section anymore)
Curiously, I've had regular Debian updates also mess up my install, to where it cannot boot: https://blog.kronis.dev/everything%20is%20broken/debian-and-...
That said, both Debian and Ubuntu have been passable in my experience, both on the desktop and for servers, basically the whole DEB group of distros including something like Linux Mint and others on the desktop as a daily driver.
I might be an outlier, but they've generally let me get things done more quickly and easily when compared to something like CentOS, RHEL, Oracle Linux and so on (though Fedora on the desktop was generally pretty good). On RPM distros things tended to go wrong, everything from using Docker instead of Podman for Docker Swarm in some smaller projects not being entirely supported (even when Podman didn't have feature parity), some oddness when setting up K3s clusters because I didn't have resources for OpenShift or whatever, regular Docker networking failing and needing to add masquerading to firewall configuration, issues with SELinux not playing nicely with some software, kswapd utterly killing the CPU when the system is running low on swap and others.
> Also, especially for startups, there are often efficiencies to using Debian Stable for everything for which it can be used. For example, it's not the smallest base of a bespoke Docker container, but there are development and operations (personnel) efficiencies when containers, host servers, and workstations are all running the same thing. Fewer surprises in production, less time and complexity trying to support multiple environments, and you can rapidly debug some things more easily on bare metal on workstation when necessary.
I actually moved over my container images to Ubuntu as well, so that I can setup dev environments on the desktop directly the same way I would for a container (sometimes debugging inside of containers is a pain): https://blog.kronis.dev/articles/using-ubuntu-as-the-base-fo...
So now I'm basically running the setup you're describing (the same OS locally, on servers and also in containers) even if I've settled on Ubuntu LTS instead because of the longer EOL and it's been a great experience so far! That said, the whole snaps/Flatpak/AppImage situation could be a bit better.
It is not that any employee can just purchase things in academia (at least in 3 different countries I worked at). Most money and purchasing power, approvals go to official-administrators (even if Professors take efforts to apply and receive grants). Spending is highly regulated. There were times I was denied buying RHEL telling me to use a 'free linux'. There is a wide belief (among official-administrators) Linux is free so paying for it is the employee wasting money. To even encourage Dell's linux server offerings I once purchased a few preinstalled RHEL devices but was later asked why not use free ones (or the windows dell offerings as they were strangely cheaper - thanks to bloatware).
At the same if you say you need to buy a windows server or zoom license this will be approved as these people know that it is not free. Their society is conditioned in this way.
In the rare case if this was approved, usually there are external auditors that analyse spent money and come back at you for not using free linux.
Ideally Canonical and RedHat (via IBM) goes to these administrators (or meetings in Ministry of Education) and advise them that making software is $$.
I have never come across that is more reluctant to actually pay for the tools to do their job than devs. Hell we've even built a quasi religious cult in foss that revolves around not paying. Yes yes I know the "free" In foss is supposed to be free as in freedom, but in practice it ain't, it's all about free as in beer. See the gp and the offence taken at actually being asked to pay for the foundational tool they use to do their job.
Both sides should calm down. It is OK to criticise Ubuntu/Debian/RHEL. A restaurant critic does not have a restaurant. CKS anyway says he/they will move to Debian.
The entire tone of the article is kind of weird given how conspiratorial in bold letters it states:
>"Which is to say, Ubuntu exists to make money for Canonical"
Yes, it's a company that pays developers and provides the long term support that the author likes, of course they exist to make money.
A quick web search returns articles claiming the company has been profitable since 2018 and that an IPO may be in the works. I can understand how something like Red Hat has managed to survive over the past twenty years, given the name it has built for itself in the "enterprise" Linux world (for better or worse), but Canonical continuing to operate off the back of Ubuntu has always baffled me.
It isn’t to me. People keep using what they are familiar with on their servers. Plus it’s one of the default option at most hosting provider.
Ubuntu is wildly popular on desktops and in the cloud (and largely for good reasons, all the phone-home and marketing etc. aside, it's a solid and polished distro), but I expect a vanishingly small fraction of Ubuntu users actually pay a dime to Canonical. Now Canonical has been trying to monetize Ubuntu in various ways over the years, some better and some worse, but I do hope they succeed so that Ubuntu is long term sustainable.
There is no AIX or HP-UX in the wild anymore, at least not outside of niche or legacy deployments, because RHEL ate it up.
Every enterprise I've ever been at required official licenses and official escalation and support SLAs. By offering enterprise support and reasonable response times they became the de facto winner of the Unix enterprise world.
Ubuntu still reigns supreme for individual users, but they're a drop in the bucket compared to F500 companies.
Nobody learning Linux for the first time today is excited about Ubuntu.
I wonder if this will seriously erode Canonical's position in the Enterprise market, or if they're sufficiently entrenched to just coast for another generation.
This post seems to imply them at canonical is some sort of evil overlord.
If you don’t like the ads, go to a different distro - that’s the good thing about Linux, theres a distro to suit everyone.
One thing I would like: Ubuntu Declarative Donkey, come on, get some inspiration from NixOS, make it simpler, it's time.
Not every project can be as amazing as debian and run strong for so many years on volonteers alone.
Apple is a commercial entity that purely has the intention of making profit and as a customer you can atleast guess where their incentives are(but can only blindly trust). They also dont benefit from the massive coomunity of developers developing features for them. Airdrop, Continuity thing, display management are stuff they had to pay tons of money to engineers working in house to develop. Hence, they get to dictate the terms.
Linux desktop by virtue of being developed by open community doesnt get that commericial dictator. Unless, some company threw their weight behind a custom linux distrbution, and developed a feature set similar to what Apple does on top of base Linux system. But, such commercial entity cannot label themselves as the "free & open" desktop operating system like Canonical does. I see System76 doing something with their own Rust based Desktop thing. If they can bring that cohesion, they get some game.
Servers are something else entirely - I'm an older UNIX beard, so I stick with the BSDs - not a fan of the GNU tool chain. Here, I'm very definitely in the minority.
I, personally, would have to spend quite a bit of time tweaking my Karabiner Elements and Hammerspoon configurations before I'm off to the races. Until then, I'm off to a crawl...
Can I put the config in git and just clone it on a new machine? Yes. But I can do that on a Linux installation, too.
that's not how this works. They choose apple and get macOS or they choose pc and get windows with it.
IE was based around gatekeeping your choices. The same can't be said of Ubuntu. It's not the 'default' Linux, its just the one that a majority choose to download and install, because most other distros dont give a damn about user experience or decent documentation.
You only have to look at the sites for Slackware, Debian and Gentoo to feel like you've been ported back to the 90s - very offputting to anyone researching which OS to pick.
what are you referring to? do you mean packages shipped on your distro are broken because the maintainers didn't change the paths or do you mean that pre-compiled binaries don't work?
because, I mean, the whole point of open source is that you can personally get your hands dirty and change the stuff. if you use binaries instead of compiling then that's what you get unfortunately
If I start an AWS windows, rhel or suse EC2 I pay far more than with ubuntu. I'm not sure if the Ubuntu AMI has any "software" charge, but RHEL on a t3.large is 14.32c/hour in US-East-2, Ubuntu is 8.32c/hour.
The thing with testing is that security wise it's the worst of both unstable and stable. It does not get timely security fixes from upstream, nor from security backports to stable.
Most importantly that would be running a VM for MS Office, Adobe and Autodesk without all the friction of having a complete desktop inside another window, managing shared folders etc etc. I'm already paying VMWare $80/yr for a shitty experience and would be glad to switch.
Alternativley, just charge for premium support for individuals that actually helps out.
I'm not a financial advisor, but is running ads and bothering people with all this snap BS really the best strategy they can came up with?
I've been around in Free Software since the nineties. At that time, it seemed very much a hobbyist effort and entirely outside the commercial world. I liked the idea of empowering users, but as much as I wanted the model to be successful, it was unclear if it was sustainable. There were plenty of naysayers.
Back then there were a bunch of new companies trying to find a business model around Free Software. Many folded. Canonical is one of the few still standing.
There's a huge bunch of work that's needed to maintain a distribution that people tend not to volunteer for, but that still needs doing. Having a functional business model allows this work to get done.
So yes, of course Ubuntu is a commercial product. The commercial side is needed to fund the free side. How this is done is a delicate balance, but how Canonical has negotiated this seems to be OK because it's one of the few still left standing. Better that than nothing at all.
Debian is a bit of an exception, but it doesn't properly work as a counterexample because a number of Debian Developers are employed by Canonical whose work also goes into Debian, and the non-DDs at Canonical working on Ubuntu also send a firehose of patches to Debian. I wouldn't claim that Debian couldn't stand on its without this contribution, but equally one can't really discount it either.
Speculation that Ubuntu will move what it has provided for free to a paywall or a subscription wall seems unhelpful to me, given that this has never actually happened in its history. The speculation seems to arise, and Canonical/Ubuntu is getting this current flak, for having dared to provide more.
I don't know what the right answer is. Clearly some sort of balance has worked for both Red Hat (although that is shifting) and Canonical, it's just a question of whether pulling too hard into things people feel is scummy will stand the test of time.
Telling users about updates that are available to them for free seems like a reasonable thing to do when they run the CLI tool designed for humans that checks for updates ("apt", vs. "apt-get" which is for scripting).
It's only speculation that this will be a general advertising channel. But that speculation seems to have turned into (false) fact through the grapevine.
Because Mark Shuttleworth (Ubuntu's leader, and Canonical's CEO) has publicly committed the following: "I do think it would be reasonable to expect the news to be related to updates and changes in the archive that are interesting for someone using apt, rather than general news, so can happily agree to set that policy and expectation amongst the folk who have write access to the news." (https://bugs.launchpad.net/ubuntu/+source/ubuntu-advantage-t...)
So now, when someone complains about "advertising", I'm never sure if they're complaining about what's actually present, or this speculation about the future that Mark has committed to not doing anyway.
Arguably the Amazon search thing Canonical had in Unity some years ago was sleazy, but the above isn't AFAIU doing any data mining on users or such.
That said, what I read was clearly an ad, just a benign one[0]. May be we are talking about different things, but this is what I saw. I reiterate, I think this ad is rather benign, and people worrying this means Ubuntu will start selling general targeted text ads or whatever in terminal space is just a bit much.
That Ubuntu Pro has a "free tier" doesn't really matter in that front. For example, I'd consider plastering notices about a heritage walk around your city where you'd may be ask for donations (using a local example in singapore) is still something I'd call "advertising." I'd call old classifieds in newspapers ads, ads need not only be for things you make money directly from or at all.
Anyway, I'll stop nit-picking here. The point I want to make is this reminds me for example about freeware or shareware notices in DOS programs. Again, rather benign, and I am not that upset about it. That said, it's not hard to foresee how it would not have elicited the response it would have given the attitude in FOSS, and that itself is an unforced error on Canonical's part.
[0] https://twitter.com/omgubuntu/status/1574759306544701443/pho...
CentOS was acquihired because it had serious sustainability issues and releases had sometimes been delayed by many months over RHEL. Red Hat couldn't base those upstream projects on a distro that could stop producing releases any time.
It was a platform, it didn't have to bring money just like the people who manage a kubernetes cluster don't bring money (the money is brought by those that write the software that runs on said cluster).
The problem is that, after a few years, those same products could not afford even the 6-month delay until changes went from RHEL to CentOS. It would basically be a waterfall model, with a huge fixed delay in the critical path before things could be developed and tested. Hence the "left shift" that is CentOS Stream, where OKD and friends can develop on the next minor release of RHEL instead of the last one.
Personally over the past year I have migrated a few CentOS 7 machines to CentOS Stream 9 and found it very stable. But if people want to use Alma, Red Hat won't complain.
But this is RH's doing, right ? RH controls the delays since they own both RHEL and CentOS. So now CentOS stream doesn't have delays, but why couldn't the same process be applied to CentOS ? Is the argument that shipping a RHEL binary compatible clone suffers from delays, but shipping something not compatible doesn't ? This seems like a business decision more than a technical one.
If you're a more technical user, no reason to run it since you can have something leaner and more fine-tuned and customized to you're needed.
But if you're not a technical user, it's probably perfect as it gets less in the way from whatever spreadsheet or document or web work you have to do.
The compromise is of course the bloat and all the stuff running under the surface to make all that work.
Different distros for different use cases.
Slick, consistent UX. Smooth upgrades. Didn't have an issue for years.
Tried Manjaro but that has shown to create considerably more work than Ubuntu.