Maybe beaten to death at this point, but Android is a Linux. Anything running on the Linux kernel is a Linux. You can be a Linux distro without using a GNU environment. There are plenty of buildroot/musl embedded environments out there with no GNU software in them that we call Linux.
Sometimes, but I'd wager we lack terminology for "Linux on the desktop" and we just cram everything into one box.
FWIW you can run Android on a desktop; you can run ChromeOS too.. those are both linux.
But you also have; GNOME+Linux, KDE+Linux, i3wm/sway+Linux.
Each of those are so wildly different as to be effectively different operating systems; people are not interacting with the kernel directly and normal users are not interacting with the Command-line in an effective way either. (and even if they were the the differences between distros and window managers would make it different enough that it could only be tangially related really).
Yeah, Android isn't a "real linux", but, part of the issue is that nothing is "real linux": it's a choose your own adventure story and that's why most people think we're all stuck-up when they try a distro and we tell them that they used the wrong one or to keep sticking it out.
It's contrived; if Android can do the job of a desktop replacement (which, it can, there are desktop modes), then why isn't it a real linux?
>Yeah, Android isn't a "real linux", but, part of the issue is that nothing is "real linux" (...) It's contrived; if Android can do the job of a desktop replacement (which, it can, there are desktop modes), then why isn't it a real linux?
When people say "Linux on the desktop" (assuming they are not being pedants, using it for anything with a Linux kernel that runs on a desktop class machine, e.g. that they'd include Windows if MS replaced the kernel with vmlinux) they don't go for etymological or technical compliance with the term.
They just mean (casually, not as in "real Linux" purity) a family of Linux-derived distros that might even include a company-made desktop environment (e.g. Mint) but do not include Android.
So, phrase-wise this is not as much about having some attributes checked (runs Linux kernel, check, has GNU userland, check, it runs on desktop machines, check), it's more about belonging to a club. Most people saying "Linux on the desktop" don't want to admit Android in that club.
Partly it goes back to the era of the original "Linux on the Desktop" dream 20+ years ago, where the term meant Linux distros taking over from Windows and Mac, and was associated with FOSS, freedom, bazaar style development, community, and so on, winning over proprietary OSes.
The article directly states that it is aware of this, and it is talking about actual distributions of linux in the sense that "desktop" makes sense. It's very clear about it, very early on.
That's also not a super useful argument in the context of desktop readiness.
Go ask the Termux guys (and girls), how much fun they are having with Android's Linux experience.
The Linux kernel is not part of the official Android API surface for app developers, only for OEMs.
Anything that works, is out of sheer luck, might work only on specific devices, or be killed by Zygote and system linker, when the application tries to be naughty and steer away from the official Java/Kotlin/NDK APIs.
Termux is stuck on Android 9 compatibility, until they are no longer allowed to publish such old API level on the store.
No glibc, gcc, glib, no Xorg/Wayland, no most Linux services/daemons.
The only thing that's common between them is the kernel. Well, the kernel alone is bloody useless. Also, Google intends to replace the Linux kernel in Android with Zircon and has already done so for some of its devices.
Secondly, show me the people who use Android as their desktop OS. All the 10 of them in the entire world.
I'm not entirely sure what the purpose of this post is. GNU/Linux is already ready for the desktop; I and others use it. The rest of the article is a mixture of 'X sucks because it's proprietary and doesn't work on Linux', which we already knew about and only the maker of X is to blame, and downright falsehoods such as 'Problems stemming from low Linux popularity and open source nature' -- the libre/oss nature of it is a problem? What?
I agree, most of the criticisms seem to concern overall unimportant technical issues. I'm using Linux since 2008 for all of my work and I haven't had any problems during the last ten years or so. I have no complaints about X, it just works fine.
What is hard in my opinion is migrating to a new machine or changing major hard drives. There is still no good migration assistant and so many things can go wrong with a major change. But that's pretty much my only complaint.
On the contrary, in my experience, unlike with Windows, migrating Linux to a new machine is trivial.
You can just move the SSD or HDD to another computer and there will be no problems because the OS has identified some change in hardware incompatible with your license and refuses to work.
Otherwise, you can just copy the old SSD/HDD to a new SSD/HDD, to clone the Linux installation, and all will work without problems (when copying, there are a few directories that must not be copied, e.g. /dev, /mnt, /proc, /sys, /tmp, but they must be created on the destination SSD/HDD; also the copy must be done with something like "/bin/cp --no-dereference --recursive --one-file-system --preserve=all", as the default cp loses data; I never use the default cp, but only an alias including the options mentioned above, to be sure that the copies are identical to the original).
The copying can also be done through the network, e.g. with rsync, while booting temporarily the destination computer from an USB stick or from a server via Ethernet.
With Windows, you cannot be sure that copying a SSD/HDD actually copies everything, and even if it would copy everything, the copy is unlikely to boot if the hardware is too different.
I have migrated or cloned Linux countless times, in a half of hour or whatever it takes to copy 1 or 2 terabytes of SSD, while with Windows I was forced to always do fresh installations and lose a lot of time to duplicate the old configurations. It is true however that I have not tried any commercial migration tool, as I would not pay for something that I only need infrequently.
>I'm not entirely sure what the purpose of this post is. GNU/Linux is already ready for the desktop; I and others use it.
It's not exactly that you're not sure, it's more you've missed the point (or, rather, multiple points).
First, the article doesn't say that "Linux is not ready for the desktop" - or concern itself with this as an abstract question.
Second, a good chunk of it's literal purpose, it's to be a list of common annoyances in using Linux for the desktop. That's useful in itself.
Third, it doesn't even imply ONLY Linux has grievances. In fact, it explicitly states that you could have a similar list for Windows (and the author even goes on to list a few major and minor such issues).
Fourth, it's serves to let people considering to switch to Linux know that it's not all roses, as some advocates present it.
Also, I find the "GNU/Linux is already ready for the desktop; I and others use it" argument tired. I've used GNU/Linux for the desktop in 1998, but it sure as hell wasn't ready then. Using something on the Desktop (and even being satisfied with it) and it being ready for the Desktop are different things. In some ways it has matured a lot since 1998, but in other ways not so much.
Many use cases aside from "browsing, watching movies, writing some doc, and checking email" are miserable on Linux for the desktop. Case in point: someone is a vlogger (a category including hundreds of thousands today) and wants to edit video. They will immediately find that there are very limited proprietary options for NLEs (basically Resolve iirc), and very crude FOSS ones, which lack many features and crash a lot. They will also find that they might as well forget the plethora of supporting related software available on Windows and Mac.
> First, the article doesn't say that "Linux is not ready for the desktop" - or concern itself with this as an abstract question.
Well, it does, but in a sarcastic manner:
"Yeah, let's consider Linux an OS ready for the desktop :-)."
> Also, I find the "GNU/Linux is already ready for the desktop; I and others use it" argument tired. I've used GNU/Linux for the desktop in 1998, but it sure as hell wasn't ready then.
Conversely, that it doesn't work for certain people does not mean that "it is not ready", which the post does state (sarcastically) as I pointed out above.
> Many use cases aside...
I'm not sure how the browsing, docs and email is miserable, maybe you can expand on that. The video editing is indeed a bit limited from my experience too. However, I don't think "limited proprietary options" is a problem. The community largely and specifically avoids proprietary software. Proprietary incursions into the community are generally seen as a negative thing. And for the lack of codecs, software patents for the most part are to blame.
And then it just comes to my original statement; many things stated in the article are non-issues to most Linux users or just falsehoods:
- Neither Mozilla Firefox nor Google Chrome use video decoding and output acceleration in Linux.
Firefox does.
- NVIDIA Optimus technology is a pain
NVIDIA is a pain.
- You don't play games, do you?
I do.
- Linux still has very few native AAA games.
So "it's not ready" because it doesn't have AAA games? What a pitty.
- To be fair you can now run thousands of Windows games through DirectX to Vulkan/OpenGL translation (Wine, Proton, Steam for Linux) but this incurs translation costs and decreases performance sometimes significantly.
No, not 'significantly' for dxvk.
- Also, anti-cheat protection usually doesn't work in Linux.
For good reason. Blame the dev, and don't make it work on Linux.
- Microsoft Office is not available for Linux
Thankfully.
- LibreOffice often has major troubles properly opening, rendering or saving documents created in Microsoft Office.
And whose fault is this? Use ODT.
- Several crucial Windows applications are not available under Linux.
Thankfully. Also, 'crucial' is subjective.
- In 2022 there's still no alternative to Windows Network File Sharing.
They are sarcastic and rebuttals of the idea that 'Linux is not ready for the desktop' because it doesn't have things the author of the post considers necessary.
As a GNU/Linux user for many years, I strongly disagree. This is a single thing which keeps millions of users away from Linux and hurts adoption dramatically. Of course it's intentionally done by the Micro$oft.
It's the same on Windows and it's what people use now, they're not gonna stay on office 2007 forever (although to be fair I have yet to work for/with someone who uses Office instead of Google docs or libreoffice, the only people I know who use "traditional" office are old people who got it with their computer)
The issue was never that Linux wasn't ready for the desktop. It's that the average user was never ready for the desktop. Perhaps evidenced by the fact that as soon as they were given the alternative of handheld computing appliances, they stopped even owning desktops in droves.
Don't worry though; those appliances are all powered by Linux or BSD at the end of the day anyway.
If there's somewhere Linux isn't quite ready for, it's the enterprise office environment...and it probably never will be, unless Microsoft really screws the pooch and Active Directory falls apart. With AS400's and DB2 still widely used across a variety of industries, I think it's plain to see that corporate America doesn't change much once it has a computing system established without having a REALLY good reason to do so.
> It's that the average user was never ready for the desktop.
Yeah. And we do little to educate in general, while MS and Google are busy selling Chromecrooks and Office licenses to kids to ensnare them in the corporate trench and profit from their data.
> the pooch and Active Directory falls apart
I suffer from this shitty 90s legacy code every day. MS has a good grip on corporate America.
> All native Linux filesystems (except ext4) are case sensitive about filenames which utterly confuses most users. This wonderful peculiarity doesn't have any sensible rationale. Less than 0.01% of users in the Linux world depend on this feature.
Lots of useful tidbits in this article, but I would disagree with the above.
1) The Unicode rules for converting lowercase to uppercase, then back to lowercase, are so convoluted, that normal developers or end-users cannot fully understand them. If I recall, the round-trip is impossible in some cases because information is lost in the conversion.
2) I have seen at least 2 occasions where one of my coworkers changed the case of a directory name on a MacOS (case-INsensitive as I discovered), pushed the changes to the remote server (GitLab or GitHub), which then proceeded to corrupt their git project badly. They had to blow away their local git repo and re-clone the project to recover. I never quite understood what happened, but I recall having to fix some stuff on my Linux box, which didn't have the case-folding problem.
I think if a file name looks different (because of case), it should be treated as being different. But I'm probably in the minority.
Strongly agree. The way the roman alphabet is represented on digital computers is to treat upper- and lowercase letters as completely separate sets of characters. People are pretty used to treating these sets of letters as functionally equivalent across "cases", so there's an argument to be made that this is a weird and unnatural choice. Maybe it would have been better to treat the "case" of a letter like we treat fonts, as information separate from the plain text that's only ever used in rich text display. But since this representational convention is settled across every platform, it is a much weirder and more confusing choice to then treat these characters as equivalent when naming files, and causes more confusion than it clears up
What I especially love is the internationalization problem examples you can provide that relate to this. One simple example is what happens in case sensitivity if I have a file named "Große" and then try and open "GROSSE".
On macOS they will be merged, similarly text find will match ss or ß. That said lossy case conversion is indeed a general problem, which is why the ease of case conversion in older text APIs cause problems - and indeed the general assumptions that there are cases, or whether or not the case is actually important.
But then apps also continue to believe that one key press == one character, in the face of reality - not helped by the early browser english centric APIs: onkeypress, down, etc being the primary mode of handling text entry when not deferring to the browser. There are better web APIs nowadays, and hopefully things like the standard IMEs that English people use (emoji input on macOS, for example), mean that more devs actually think about these things, though in general everytime I encounter some site taking over text entry themselves I assume that they're broken for non-english or at least non-alphabet based languages.
Tangentially related, there are issues when making .iso's, depending on the ISO level used (this is independent from the OS), it is a capitalization issue that affects - besides german or other language characters - also more common ones in any languages, most notably the - (dash/minus) is capitalized as _ (underscore).
It’s a dumb argument anyway because it’s misunderstanding who the “users” are that would be impacted by this type of thing.
I cannot think of a workflow outside of programming where I would need to think about case sensitive filenames. If I’m an end user, I’m not programming, and I’m being forced to type accurate filenames into a box, either I’m doing it wrong or the software I’m using is an abomination.
If I’m programming, on the other hand, I’d better bloody know how files work.
> I think if a file name looks different (because of case), it should be treated as being different. But I'm probably in the minority.
Completely agree. Also ext4 _is_ case sensitive on filenames.
I think that _new_ users of a system should adapt to that system and not the other way around. Linux file systems have been case sensitive from the beginning. Changing that now would break things for users like me who are using linux for over 25 years.
Classic desktop Linux distributions like Mint totally work on the desktop. You only need to use it on regular HW that isn't too brand new that the kernel doesn't support it. For most folks it probably covers about 80% of what they need to do with minimum of fuss. I don't think there's a difficulty quotient for using Mint compared to the leap from Win10 to Win11. But personal bias will prevent a lot of folks giving Mint fair attempts to win them over. Not everybody wants to leave their comfort blanket.
Chasing a moving target is no fun, exhausting even, but it would be a lot easier to just upstream the driver. The article fails to mention that AMD did just that and have vastly better UX than Nvidia.
- No GUI for multi user audio device sharing:
It sucks to have to use a terminal and read a manual so you can listen to music while your brother is on YouTube (somehow), but at least you may learn a useful skill.
- lack of support for some printers/scanners/peripherals:
Obviously things take time if the company offers no official support; volunteers only implement what they need to move on with their life. Generally if a printer is supported on macOS it should work on GNU/Linux due to sharing the underlying framework (CUPS).
- Power management:
While Linux have had its issues, things are vastly improved nowadays, to the point I suspect it's better than Windows on most hardware (but I have no personal experience).
- X.org/Wayland:
No excuses here. X.org is a hot mess of legacy garbage, and Wayland is not all it was hyped to be, with some equally brain-damaged design choices. I think once Wayland becomes the new normal transitioning to other display servers get easier due to simpler APIs than Xorg (I'm sorry for everyone who had to endure that in their lives).
- Fonts:
Still a mess, but improving rapidly.
- kernel difficult to debug:
Is Mach or NT really better in this area? At least Linux is completely open and has a gazillion knobs that you can tweak.
- Linux kernel code is low quality:
It's got some ugly interfaces and warts, like most hugely complex organically grown software projects, but that doesn't mean the code is poor quality.
- Low-memory conditions:
The kernel will still become unresponsive when running low on memory, but there are user-space workarounds for that (as mentioned in the article, I agree that earlyoom should be default).
- anti-cheat protections fail to work:
I personally do not want software making changes on my machine without my explicit consent. This is a feature IMO.
- Money, enthusiasm, motivation and responsibility:
I adore the diversity and creativity in FOSS. Many, perhaps most developers are doing it from passion instead of greedy interests. If you care about a particular project or feature, they will likely be happy to accept donations.
Hope this did not sound too condescending. I am grateful posts like these exists giving everyone a chance to straighten their priorities.
This complaint has nothing to do with Linux. If you copy the Windows fonts or the MacOS font to a Linux computer, they will work as well as on the original systems.
I have stopped using my Apple laptop many years ago, but I am still using on Linux a typeface kept from the Apple laptop, i.e. Hiragino Mincho Pro, because I have not found yet a better Japanese typeface.
The problem with fonts is that the free fonts used to be very ugly, so the default Linux installations were also ugly.
Now there are a few high-quality free fonts and everyone who wants better fonts can choose to buy good fonts from the many online stores.
For me, the quality of the typefaces is an extremely important part of a computer experience. I would not pay money for Windows, because it does not offer any feature that is useful for me, but I have bought a large number of beautiful typefaces and I use them on my Linux desktops and laptops, instead of the default fonts provided by most Linux distributions.
So my Linux documents look even better than any documents using the default Windows or MacOS fonts.
The quality of the fonts has nothing to do with the operating system but only with what you are willing to spend. When spending money on Windows or MacOS a part goes to the included fonts, so it is expected that those must be better than free fonts.
This article is kinda a mixed bag of complaints that ranges from very reasonable and important to completely nonsense, which is to say it's very representative of discussions of the relative merits of operating systems
For example, a lot of driver issues and architecture flaws (e.g. on wayland/X) it points out are really useful to consider for both maintainers of those subsystems and users of a desktop computer. But then you also get stuff where things are considered "impossible" for an average user because they involve... editing a text file? Using a shell at all? Not every difference in how you use something is a problem with it. I think having the exact same user interface for an entirely different operating system is a pretty insane expectation
The examples in this article about gaming feel very outdated now - the sheer fact that the Steam Deck exists at all is good evidence of how different the current state of Linux gaming is to how the article suggests.
A lot of big titles run okay. Anti-cheat is getting more and more support for Linux. Proton + DXVK in particular provides excellent performance for a wide range of titles (and it really is within a few % of Windows). Wine and Proton quite often provides superior compatibility on older titles to running them on modern Windows - a reasonable chunk of my selection of early 2000s titles are broken on Windows 10/11 but work fine on Linux.
With regard to the suitability of Linux in general, I've said in previous comment threads on HN about how I try Linux on the desktop every 6 to 12 months to see how I get on with it, and this time (since about April 2022) it has actually stuck pretty well - the combination of hitting more issues with newer Windows, and the fact that Linux (for me Arch + KDE on AMD) has genuinely improved means I now spend more time on Linux than on Windows for basically the first time ever.
Linux is far from flawless, but it's making excellent progress.
Yes just browsing quickly I see almost nothing that applies to my computers, and I have used Linux on all my laptops and desktops for about 20 years.
Of course, it is pretty certain that there are some combinations of hardware where Linux may have problems.
For example:
"Under Linux, setting multi-monitor configurations especially using multiple GPUs running binary NVIDIA drivers can be a major PITA."
In this sentence, the only point that could be true is "multiple GPUs", because I have not tested simultaneously "multiple GPUs" with "multi-monitor".
I use "multiple GPUs" for computation with CUDA, and there are no problems there.
I have been using 3 or 4 monitors with an NVIDIA GPU and with their Linux driver for very many years and nothing could be simpler. Even when the monitors have different sizes and resolutions, it is trivial to arrange them in any layout you want with the dialog box of the NVIDIA settings program, in a few seconds, which will be remembered later. There exists no PITA.
Another supposed issue "30bit displays are unusable under Linux".
This is certainly not true. Since 2015, I have used only 30-bit monitors.
The only problem that I have encountered with 30-bit monitors was not related to Linux, but to Java. There are certain commercial programs that have installers written in Java and those installers crash on 30-bit displays, both on Windows and on Linux.
This is an old Java bug, which has not been solved for many years. So much for the myth of writing in Java for running the program anywhere.
Maybe some of the other complaints have been applicable somewhere, but I do not recognize any of them. For example Firefox and Chrome have always worked for me on Linux at least as well as on Windows. Both have ugly behaviors or bugs that I dislike, but those misfeatures are the same on Windows as on Linux.
The only complaint that I recognize is with NVIDIA Optimus on laptops, because some years ago I had to spend a couple of days with a laptop until succeeding to make Optimus work under Linux.
I have two NVIDIA GPUs in my main machine, and last year i set up a box with three GPUs, all NVIDIA, for a multiseat config. The box worked just fine both in multihead as in multiseat.
NVIDIA can be a pain, but most of the wrinkles are ironed out, unless you have very old (or very new) cards
Have you read the rest of the sentence? It lists current unresolved issues.
Who the hell cares you can enable 30bit mode _only_ under X.org when the most crucial applications fall apart under it?
> I see almost nothing that applies to my computers
Yeah, it's addressed in the article:
A lot of people who are new to Linux or those who use a very tiny subset of applications are quick to disregard the entire list saying things like, "Audio in Linux works just fine for me." or "I've never had any troubles with video in Linux."
I won't reply to any further comments down in this thread because as always people read a few lines, disagree vehemently and close the page. If you don't bloody care to read everything then just don't post comments, OK? I understand most people nowadays have a severe case of ADHD and only capable of watching 30 second TikToks, but guess what this webpage is not a TikTok.
"(Samba is not native, it's a reverse engineered SMB implementation, it's very difficult for the average Joe to manage and set up.)".
How is Samba not native ? SMB is a Microsoft proprietary protocol, but now fully documented. So if Microsoft doesn't release an SMB server for Linux then others must. As an aside, Microsoft now employs Samba Team member Steve French and actively participates in Linux kernel SMB client development.
The "difficult for the average Joe to manage and set up" I'll give them :-).
From the article: "No CIFS/AD level replacement/ equivalent (SAMBA doesn't count for many reasons)".
But then doesn't list any reasons. Hmmm. Agenda much ?
A lot of these examples are just the opinions of the author rather than actual problems.
For example:
>3. General graphics APIs issues:
> No high level, stable, sane (truly forward and backward compatible) and standardized API for developing GUI applications (like core Win32 API - most Windows 95 applications still run fine in Windows 10 - that's 24 years of binary compatibility). Both GTK and Qt (incompatible GTK versions 1, 2, 3, 4 and incompatible Qt versions 4, 5, 6 just for the last decade) don't strive to be backwards compatible. The Qt company also changed the licensing model for their toolkit which makes using the library under Linux problematic to say the least.
Firstly, I've seen a LOT of Windows 95 apps that do not "still run fine in Windows 10" -- in fact I'm not even certain how you could quantify "most" here because I made a living for several years doing updates of old apps for companies to get them to run on modern supported systems. In many cases using Wine things just worked but many older API calls broke older apps in Windows 10.
> There's no universal unified graphical toolkit/API (and e.g. Wayland developers will not implement it) akin to Win32 which means applications using different toolkits (e.g. Qt, GTK, EFL) may look and behave differently and there's no way to configure all of them once and for all.
This isn't true on Windows or MacOS either and I'd argue isn't a necessity anymore. Don't get me wrong, I get the appeal. Look and behaviour stability of a whole slew of first-party apps went out the Window with Windows 8 (arguably earlier at the introduction of the ribbon and auto-hiding the menu bar).
For the past ten years it's been "stick the web on the desktop" so you get apps like Teams and Slack that look nothing like more traditional applications and have no consistency from app to app. Not to mention there's no behavioural consistency across platforms (let alone within each platform) in those kinds of apps.
The author throughout suggests that if there's no purpose-built configuration GUI, it's a problem. Here's an example:
> Both PipeWire and PulseAudio are not configured out of the box to support multi-user mode (e.g. multiple users sharing the same device simultaneously) and configuring them to enable this feature is impossible for the average user (no gui whatsoever, you need to edit text files).
Then they go on to say:
> ! It should be possible to configure pretty much everything via GUI (in the end Windows and Mac OS allow this) which is still not a case for some situations and operations.
If you never do anything advanced on either of those operating systems then I agree you can get everywhere with just a GUI. However, this is just factually false -- Windows and MacOS do not provide configuration GUIs for every possible setting.
If you have to open regedit to do anything, this is functionally not superior to "editing text files", and MacOS simply doesn't offer you configuration where they don't think you need it, forcing people to edit .plists or use the defaults command-line instead.
"It should be possible to configure pretty much everything via GUI (in the end Windows and Mac OS allow this)"
Yes, this sentence only shows that the author has extremely limited Windows and MacOS experience, because there are plenty of things that cannot be configured through the GUI in those.
I no longer use Apple computers, so I do not know if this is true any more for the recent MacOS versions.
However, when I had an Apple laptop, I was extremely surprised to discover that there were plenty of things that could not be configured through the GUI, but they could be configured with some undocumented command options added to command-line utilities whose origin was in FreeBSD, but which had been extended with additional options for OS X.
I no longer remember all the cases where I was forced to use Apple CLI tools, but at least one of them was needed to be able to associate to some WiFi access points, others than those overpriced made by Apple, because not all the WiFi configuration options could be modified through the GUI.
Also in Windows, a couple of years ago I had to install Windows Enterprise on several kinds of embedded computers and in each case I had to spend several days until convincing Windows to work acceptably, e.g. in order to work fine with a very slow SSD.
In every case, the Windows Knowledge Base was not good enough to solve all the problems, so I had to do many Internet searches through forums of Windows users who complained about the same issues, in most cases without receiving any useful reply.
Fortunately, in every case I have found some reply with the right solution, and in all cases the solution was not in the Control Panel or in any other GUI tool, but in invoking some obscure Windows command-line system-management tool, using some magic command-line options.
The only difference vs. Linux was that the Apple and Microsoft CLI tools did not have adequate documentation and only some gurus were aware of what can be done with them.
> Firstly, I've seen a LOT of Windows 95 apps that do not "still run fine in Windows 10" -
Absolute most Windows 95 era 32bit applications run just _fine_ under Windows 10, period. Sans games. Games were often coded for then current GPUs and modern GPUs have different API.
> However, this is just factually false -- Windows and MacOS do not provide configuration GUIs for every possible setting.
I've not needed to use regedit in Windows or edit text config files since forever. You can fine tune stuff but absolute most people never need it.
It is pretty frustrating that Linux failed to take advantage of Windows spending a decade with the tiles two-screen desktop and all the associated disasters of Windows 8.
Ultimately, linux on the desktop needed a big sponsor. IMO it should have been Intel who had some spare billions to put the heat on Intel and keep abreast of hardware adaptation since Linux is shoehorned into so many things, but Intel was ruled by drooling MBAs and that's why it's in the predicament it's in.
Sun would have been a good one if it didn't blow up, and so tied to Solaris.
Maybe the sponsor should have been the EU. The US had OS marketplace supremacy with Microsoft and to a lesser extent Apple. The EU should have poured a couple billion into it for economic independence, and fostering their own dynamic software industry. AMD was sort of EU's hedge on Intel, Airbus was their hedge on Boeing.
Heck the US military should have done it, and still can, for effective cyberdefense. Microsoft and Apple won't open their codebases for hardening or deep dives into hardware+software security, but Linux can. Alas no way the lobbyists for the private OSs would allow it.
Linux is a lot like democracy, the worst OS, except for all the others tried.
The history of Linux Desktop, since Windows 2000, since Microsoft does something that will affect users, we get this prophecy that the time of Linux Desktop will come and regular users will switch in droves to GNU/Linux.
Windows Me, Windows XP toy interface, DX 10 being Windows Vista only, Windows 8 mess up, Windows 10 telemetry, ....
Meanwhile they mess up the userspace so much that everyone is shipping Electron apps for Linux users.
>"! Distros' repositories do not contain all available open source software (libraries' conflicts don't even allow that luxury). The user should never be bothered with using ./configure && make && make install (besides, it's insecure, can break things in a major way, and it sometimes simply doesn't work because the user cannot install/configure dependencies properly). It should be possible to install any software by downloading a package and double clicking it (yes, like in Windows, but probably prompting for a user/administrator password).Linux distros.
! Applications development is a major PITA. Different distros can use a) different library versions, b) different compiler flags, c) different compilers. This leads to a number of problems raised to the third power."
I don't know that I can help with the other problems enumerated; but...
NixOS -- goes a long way toward solving the exact problems quoted above!
Now, NixOS does come with a bit of a learning curve however -- so it's not recommended for all types of users!
If a user has put in the prerequisite time and energy to learn Unix/Linux -- then with the additional time/effort to learn NixOS -- they can largely be free of the above problems -- which are well-known to the Linux software development community at this point in time...
Incredible to see how many people in these comments are mad that someone dare challenge that their dear GNU/Linux is a great desktop OS. This is part of why it will never get there. The Linux community just doesn’t Get It. Too much autistic nerd energy, not enough people with a functioning sense of empathy.
I am getting a bit tired of those articles on "linux on the desktop". I started using Linux in 1997 and it has been my main work os (linux on the desktop) since 2000. I have used Gnome 1, Gnome 2, Xfce 3, Xfce 4, E16, E17, KDE Plasma 5. I have used Linux on x86 (laptop+desktop+server), amd64 (laptop+desktop+server), ppc (laptop), arm (home server). I use X.Org.
For me, it totally works on the desktop. And since I really want to use Linux as a desktop os, I take into account the limitations and live with them. That means I stopped using nvidia gpus and just use an amd gpu or the built-in intel. That also means that to this day I still check linux hardware support before buying hardware. I live with the fact that not all games are gonna run on Linux. I bought games from Loki, I was very happy with the work from Ryan Gordon (icculus). I am happy with smaller indie games that run on Linux, since I no longer have the time to invest in big AAA titles anyways. I am a software engineer myself and do sometimes contribute to open source projects in my free time, when something is bothering me enough.
Linux does not have to gain a big market share as a desktop os to become a success, it has proven itself and it is a huge success already. Even Microsoft is investing to make the use of Linux painless from within Windows. I assume that they are aware that certain development tools just work better on Linux and that deployment in the cloud most of the time means deployment on Linux. So if Linux on the desktop does not satisfy your personal requirements, you are probably better off using another os.
A lot of the issues that are mentioned in this list (and others) are often personal preference, just wrong, or well known. However, the thing that bothers me the most is new users insisting that Linux should behave differently than it has done for a very long time, because that would be more convenient for users coming from another os... and it 'has' to change so that Linux can finally become the major desktop os! I believe a lot of Linux desktop users use the os as a work(station) os and I think that there should always be some respect for existing users. And I say that both for new users coming to Linux and for existing Linux projects that want to draw in new users or just want to change things to change things. When you use a tool to get work done and get it done as efficiently as possible, there is nothing more annoying and frustrating than having to deal with unnecessary changes in your tools.
Having just installed Debian 11xfce on a Dell Intel i5 laptop, and fighting with the power applet crashing b/c power manager fails to start, I'm having flashbacks to 20 years ago with the same problems. Still having to add nomodeset to grub for the screen to work has been stopping people from using linux for 20 years.
Seriously, if a fresh install won't boot, how's a newbie not going to feel overwhelmed. For those same 20 years, people have been saying to me ( even IT people ) that they'd like to try linux but don't have the time to troubleshoot everything. Even if they sink a few hours into getting the screen to work, laptops still burn through the battery and they go back to Windows since they use it on battery.
I use both Windows and linux, and with Windows quality crashing, people keep proclaiming the year of linux. But there's still a long way to improve linux before it gets as good as poor quality Windows.
Web Archive has most of the revisions. There are changes but far and between. Overall issues get mostly added (for instance a large chunk of Wayland ones), sometimes I remove stuff when it gets resolved but it's quite a rare occurrence.
67 comments
[ 3.3 ms ] story [ 129 ms ] threadYes, but that's neither here, not there.
When people say Linux in this context they don't mean Android with the Linux kernel, they mean traditional distros of Linux + GNU userland.
FWIW you can run Android on a desktop; you can run ChromeOS too.. those are both linux.
But you also have; GNOME+Linux, KDE+Linux, i3wm/sway+Linux.
Each of those are so wildly different as to be effectively different operating systems; people are not interacting with the kernel directly and normal users are not interacting with the Command-line in an effective way either. (and even if they were the the differences between distros and window managers would make it different enough that it could only be tangially related really).
Yeah, Android isn't a "real linux", but, part of the issue is that nothing is "real linux": it's a choose your own adventure story and that's why most people think we're all stuck-up when they try a distro and we tell them that they used the wrong one or to keep sticking it out.
It's contrived; if Android can do the job of a desktop replacement (which, it can, there are desktop modes), then why isn't it a real linux?
When people say "Linux on the desktop" (assuming they are not being pedants, using it for anything with a Linux kernel that runs on a desktop class machine, e.g. that they'd include Windows if MS replaced the kernel with vmlinux) they don't go for etymological or technical compliance with the term.
They just mean (casually, not as in "real Linux" purity) a family of Linux-derived distros that might even include a company-made desktop environment (e.g. Mint) but do not include Android.
So, phrase-wise this is not as much about having some attributes checked (runs Linux kernel, check, has GNU userland, check, it runs on desktop machines, check), it's more about belonging to a club. Most people saying "Linux on the desktop" don't want to admit Android in that club.
Partly it goes back to the era of the original "Linux on the Desktop" dream 20+ years ago, where the term meant Linux distros taking over from Windows and Mac, and was associated with FOSS, freedom, bazaar style development, community, and so on, winning over proprietary OSes.
That's also not a super useful argument in the context of desktop readiness.
The Linux kernel is not part of the official Android API surface for app developers, only for OEMs.
Anything that works, is out of sheer luck, might work only on specific devices, or be killed by Zygote and system linker, when the application tries to be naughty and steer away from the official Java/Kotlin/NDK APIs.
Termux is stuck on Android 9 compatibility, until they are no longer allowed to publish such old API level on the store.
I think they've already discontinued support for Google play and ask users to install Termux from F-Droid.
Android has 0 to do with Linux.
No glibc, gcc, glib, no Xorg/Wayland, no most Linux services/daemons.
The only thing that's common between them is the kernel. Well, the kernel alone is bloody useless. Also, Google intends to replace the Linux kernel in Android with Zircon and has already done so for some of its devices.
Secondly, show me the people who use Android as their desktop OS. All the 10 of them in the entire world.
What is hard in my opinion is migrating to a new machine or changing major hard drives. There is still no good migration assistant and so many things can go wrong with a major change. But that's pretty much my only complaint.
You can just move the SSD or HDD to another computer and there will be no problems because the OS has identified some change in hardware incompatible with your license and refuses to work.
Otherwise, you can just copy the old SSD/HDD to a new SSD/HDD, to clone the Linux installation, and all will work without problems (when copying, there are a few directories that must not be copied, e.g. /dev, /mnt, /proc, /sys, /tmp, but they must be created on the destination SSD/HDD; also the copy must be done with something like "/bin/cp --no-dereference --recursive --one-file-system --preserve=all", as the default cp loses data; I never use the default cp, but only an alias including the options mentioned above, to be sure that the copies are identical to the original).
The copying can also be done through the network, e.g. with rsync, while booting temporarily the destination computer from an USB stick or from a server via Ethernet.
With Windows, you cannot be sure that copying a SSD/HDD actually copies everything, and even if it would copy everything, the copy is unlikely to boot if the hardware is too different.
I have migrated or cloned Linux countless times, in a half of hour or whatever it takes to copy 1 or 2 terabytes of SSD, while with Windows I was forced to always do fresh installations and lose a lot of time to duplicate the old configurations. It is true however that I have not tried any commercial migration tool, as I would not pay for something that I only need infrequently.
It's not exactly that you're not sure, it's more you've missed the point (or, rather, multiple points).
First, the article doesn't say that "Linux is not ready for the desktop" - or concern itself with this as an abstract question.
Second, a good chunk of it's literal purpose, it's to be a list of common annoyances in using Linux for the desktop. That's useful in itself.
Third, it doesn't even imply ONLY Linux has grievances. In fact, it explicitly states that you could have a similar list for Windows (and the author even goes on to list a few major and minor such issues).
Fourth, it's serves to let people considering to switch to Linux know that it's not all roses, as some advocates present it.
Also, I find the "GNU/Linux is already ready for the desktop; I and others use it" argument tired. I've used GNU/Linux for the desktop in 1998, but it sure as hell wasn't ready then. Using something on the Desktop (and even being satisfied with it) and it being ready for the Desktop are different things. In some ways it has matured a lot since 1998, but in other ways not so much.
Many use cases aside from "browsing, watching movies, writing some doc, and checking email" are miserable on Linux for the desktop. Case in point: someone is a vlogger (a category including hundreds of thousands today) and wants to edit video. They will immediately find that there are very limited proprietary options for NLEs (basically Resolve iirc), and very crude FOSS ones, which lack many features and crash a lot. They will also find that they might as well forget the plethora of supporting related software available on Windows and Mac.
Well, it does, but in a sarcastic manner:
"Yeah, let's consider Linux an OS ready for the desktop :-)."
> Also, I find the "GNU/Linux is already ready for the desktop; I and others use it" argument tired. I've used GNU/Linux for the desktop in 1998, but it sure as hell wasn't ready then.
Conversely, that it doesn't work for certain people does not mean that "it is not ready", which the post does state (sarcastically) as I pointed out above.
> Many use cases aside...
I'm not sure how the browsing, docs and email is miserable, maybe you can expand on that. The video editing is indeed a bit limited from my experience too. However, I don't think "limited proprietary options" is a problem. The community largely and specifically avoids proprietary software. Proprietary incursions into the community are generally seen as a negative thing. And for the lack of codecs, software patents for the most part are to blame.
And then it just comes to my original statement; many things stated in the article are non-issues to most Linux users or just falsehoods:
- Neither Mozilla Firefox nor Google Chrome use video decoding and output acceleration in Linux.
Firefox does.
- NVIDIA Optimus technology is a pain
NVIDIA is a pain.
- You don't play games, do you?
I do.
- Linux still has very few native AAA games.
So "it's not ready" because it doesn't have AAA games? What a pitty.
- To be fair you can now run thousands of Windows games through DirectX to Vulkan/OpenGL translation (Wine, Proton, Steam for Linux) but this incurs translation costs and decreases performance sometimes significantly.
No, not 'significantly' for dxvk.
- Also, anti-cheat protection usually doesn't work in Linux.
For good reason. Blame the dev, and don't make it work on Linux.
- Microsoft Office is not available for Linux
Thankfully.
- LibreOffice often has major troubles properly opening, rendering or saving documents created in Microsoft Office.
And whose fault is this? Use ODT.
- Several crucial Windows applications are not available under Linux.
Thankfully. Also, 'crucial' is subjective.
- In 2022 there's still no alternative to Windows Network File Sharing.
It's available since 1992: https://www.samba.org/
- Linux doesn't have a reliably working hassle-free fast native (directly mountable via the kernel; FUSE doesn't cut it) MTP implementation.
I can transfer files to my phone just fine.
- Too many things in Linux require manual configuration using text files.
No.
etc.
So are all the "thankfully"s in your reply that you seem to consider rebuttals of anything.
I said the exact opposite:
"Many use cases ASIDE from "browsing, watching movies, writing some doc, and checking email" are miserable on Linux for the desktop" [empahsis mine]
> Thankfully.
As a GNU/Linux user for many years, I strongly disagree. This is a single thing which keeps millions of users away from Linux and hurts adoption dramatically. Of course it's intentionally done by the Micro$oft.
Don't worry though; those appliances are all powered by Linux or BSD at the end of the day anyway.
If there's somewhere Linux isn't quite ready for, it's the enterprise office environment...and it probably never will be, unless Microsoft really screws the pooch and Active Directory falls apart. With AS400's and DB2 still widely used across a variety of industries, I think it's plain to see that corporate America doesn't change much once it has a computing system established without having a REALLY good reason to do so.
Yeah. And we do little to educate in general, while MS and Google are busy selling Chromecrooks and Office licenses to kids to ensnare them in the corporate trench and profit from their data.
> the pooch and Active Directory falls apart
I suffer from this shitty 90s legacy code every day. MS has a good grip on corporate America.
You're on Hacker News, I presume you're a professional programmer, that says a lot about Linux readiness.
Lots of useful tidbits in this article, but I would disagree with the above.
1) The Unicode rules for converting lowercase to uppercase, then back to lowercase, are so convoluted, that normal developers or end-users cannot fully understand them. If I recall, the round-trip is impossible in some cases because information is lost in the conversion.
2) I have seen at least 2 occasions where one of my coworkers changed the case of a directory name on a MacOS (case-INsensitive as I discovered), pushed the changes to the remote server (GitLab or GitHub), which then proceeded to corrupt their git project badly. They had to blow away their local git repo and re-clone the project to recover. I never quite understood what happened, but I recall having to fix some stuff on my Linux box, which didn't have the case-folding problem.
I think if a file name looks different (because of case), it should be treated as being different. But I'm probably in the minority.
But then apps also continue to believe that one key press == one character, in the face of reality - not helped by the early browser english centric APIs: onkeypress, down, etc being the primary mode of handling text entry when not deferring to the browser. There are better web APIs nowadays, and hopefully things like the standard IMEs that English people use (emoji input on macOS, for example), mean that more devs actually think about these things, though in general everytime I encounter some site taking over text entry themselves I assume that they're broken for non-english or at least non-alphabet based languages.
JFYI:
https://msfn.org/board/topic/182116-winsetupfromusb-problem-...
I cannot think of a workflow outside of programming where I would need to think about case sensitive filenames. If I’m an end user, I’m not programming, and I’m being forced to type accurate filenames into a box, either I’m doing it wrong or the software I’m using is an abomination.
If I’m programming, on the other hand, I’d better bloody know how files work.
Completely agree. Also ext4 _is_ case sensitive on filenames.
I think that _new_ users of a system should adapt to that system and not the other way around. Linux file systems have been case sensitive from the beginning. Changing that now would break things for users like me who are using linux for over 25 years.
- Lack of API/ABI stability in the kernel:
Chasing a moving target is no fun, exhausting even, but it would be a lot easier to just upstream the driver. The article fails to mention that AMD did just that and have vastly better UX than Nvidia.
- No GUI for multi user audio device sharing:
It sucks to have to use a terminal and read a manual so you can listen to music while your brother is on YouTube (somehow), but at least you may learn a useful skill.
- lack of support for some printers/scanners/peripherals:
Obviously things take time if the company offers no official support; volunteers only implement what they need to move on with their life. Generally if a printer is supported on macOS it should work on GNU/Linux due to sharing the underlying framework (CUPS).
- Power management:
While Linux have had its issues, things are vastly improved nowadays, to the point I suspect it's better than Windows on most hardware (but I have no personal experience).
- X.org/Wayland:
No excuses here. X.org is a hot mess of legacy garbage, and Wayland is not all it was hyped to be, with some equally brain-damaged design choices. I think once Wayland becomes the new normal transitioning to other display servers get easier due to simpler APIs than Xorg (I'm sorry for everyone who had to endure that in their lives).
- Fonts:
Still a mess, but improving rapidly.
- kernel difficult to debug:
Is Mach or NT really better in this area? At least Linux is completely open and has a gazillion knobs that you can tweak.
- Linux kernel code is low quality:
It's got some ugly interfaces and warts, like most hugely complex organically grown software projects, but that doesn't mean the code is poor quality.
- Low-memory conditions:
The kernel will still become unresponsive when running low on memory, but there are user-space workarounds for that (as mentioned in the article, I agree that earlyoom should be default).
- anti-cheat protections fail to work:
I personally do not want software making changes on my machine without my explicit consent. This is a feature IMO.
- Money, enthusiasm, motivation and responsibility:
I adore the diversity and creativity in FOSS. Many, perhaps most developers are doing it from passion instead of greedy interests. If you care about a particular project or feature, they will likely be happy to accept donations.
Hope this did not sound too condescending. I am grateful posts like these exists giving everyone a chance to straighten their priorities.
This complaint has nothing to do with Linux. If you copy the Windows fonts or the MacOS font to a Linux computer, they will work as well as on the original systems.
I have stopped using my Apple laptop many years ago, but I am still using on Linux a typeface kept from the Apple laptop, i.e. Hiragino Mincho Pro, because I have not found yet a better Japanese typeface.
The problem with fonts is that the free fonts used to be very ugly, so the default Linux installations were also ugly.
Now there are a few high-quality free fonts and everyone who wants better fonts can choose to buy good fonts from the many online stores.
For me, the quality of the typefaces is an extremely important part of a computer experience. I would not pay money for Windows, because it does not offer any feature that is useful for me, but I have bought a large number of beautiful typefaces and I use them on my Linux desktops and laptops, instead of the default fonts provided by most Linux distributions.
So my Linux documents look even better than any documents using the default Windows or MacOS fonts.
The quality of the fonts has nothing to do with the operating system but only with what you are willing to spend. When spending money on Windows or MacOS a part goes to the included fonts, so it is expected that those must be better than free fonts.
For example, a lot of driver issues and architecture flaws (e.g. on wayland/X) it points out are really useful to consider for both maintainers of those subsystems and users of a desktop computer. But then you also get stuff where things are considered "impossible" for an average user because they involve... editing a text file? Using a shell at all? Not every difference in how you use something is a problem with it. I think having the exact same user interface for an entirely different operating system is a pretty insane expectation
A lot of big titles run okay. Anti-cheat is getting more and more support for Linux. Proton + DXVK in particular provides excellent performance for a wide range of titles (and it really is within a few % of Windows). Wine and Proton quite often provides superior compatibility on older titles to running them on modern Windows - a reasonable chunk of my selection of early 2000s titles are broken on Windows 10/11 but work fine on Linux.
With regard to the suitability of Linux in general, I've said in previous comment threads on HN about how I try Linux on the desktop every 6 to 12 months to see how I get on with it, and this time (since about April 2022) it has actually stuck pretty well - the combination of hitting more issues with newer Windows, and the fact that Linux (for me Arch + KDE on AMD) has genuinely improved means I now spend more time on Linux than on Windows for basically the first time ever.
Linux is far from flawless, but it's making excellent progress.
Of course, it is pretty certain that there are some combinations of hardware where Linux may have problems.
For example:
"Under Linux, setting multi-monitor configurations especially using multiple GPUs running binary NVIDIA drivers can be a major PITA."
In this sentence, the only point that could be true is "multiple GPUs", because I have not tested simultaneously "multiple GPUs" with "multi-monitor".
I use "multiple GPUs" for computation with CUDA, and there are no problems there.
I have been using 3 or 4 monitors with an NVIDIA GPU and with their Linux driver for very many years and nothing could be simpler. Even when the monitors have different sizes and resolutions, it is trivial to arrange them in any layout you want with the dialog box of the NVIDIA settings program, in a few seconds, which will be remembered later. There exists no PITA.
Another supposed issue "30bit displays are unusable under Linux".
This is certainly not true. Since 2015, I have used only 30-bit monitors.
The only problem that I have encountered with 30-bit monitors was not related to Linux, but to Java. There are certain commercial programs that have installers written in Java and those installers crash on 30-bit displays, both on Windows and on Linux.
This is an old Java bug, which has not been solved for many years. So much for the myth of writing in Java for running the program anywhere.
Maybe some of the other complaints have been applicable somewhere, but I do not recognize any of them. For example Firefox and Chrome have always worked for me on Linux at least as well as on Windows. Both have ugly behaviors or bugs that I dislike, but those misfeatures are the same on Windows as on Linux.
The only complaint that I recognize is with NVIDIA Optimus on laptops, because some years ago I had to spend a couple of days with a laptop until succeeding to make Optimus work under Linux.
NVIDIA can be a pain, but most of the wrinkles are ironed out, unless you have very old (or very new) cards
Have you read the rest of the sentence? It lists current unresolved issues.
Who the hell cares you can enable 30bit mode _only_ under X.org when the most crucial applications fall apart under it?
> I see almost nothing that applies to my computers
Yeah, it's addressed in the article:
A lot of people who are new to Linux or those who use a very tiny subset of applications are quick to disregard the entire list saying things like, "Audio in Linux works just fine for me." or "I've never had any troubles with video in Linux."
I won't reply to any further comments down in this thread because as always people read a few lines, disagree vehemently and close the page. If you don't bloody care to read everything then just don't post comments, OK? I understand most people nowadays have a severe case of ADHD and only capable of watching 30 second TikToks, but guess what this webpage is not a TikTok.
How is Samba not native ? SMB is a Microsoft proprietary protocol, but now fully documented. So if Microsoft doesn't release an SMB server for Linux then others must. As an aside, Microsoft now employs Samba Team member Steve French and actively participates in Linux kernel SMB client development.
The "difficult for the average Joe to manage and set up" I'll give them :-).
From the article: "No CIFS/AD level replacement/ equivalent (SAMBA doesn't count for many reasons)".
But then doesn't list any reasons. Hmmm. Agenda much ?
It's a completely alien to Linux daemon.
For example:
>3. General graphics APIs issues:
> No high level, stable, sane (truly forward and backward compatible) and standardized API for developing GUI applications (like core Win32 API - most Windows 95 applications still run fine in Windows 10 - that's 24 years of binary compatibility). Both GTK and Qt (incompatible GTK versions 1, 2, 3, 4 and incompatible Qt versions 4, 5, 6 just for the last decade) don't strive to be backwards compatible. The Qt company also changed the licensing model for their toolkit which makes using the library under Linux problematic to say the least.
Firstly, I've seen a LOT of Windows 95 apps that do not "still run fine in Windows 10" -- in fact I'm not even certain how you could quantify "most" here because I made a living for several years doing updates of old apps for companies to get them to run on modern supported systems. In many cases using Wine things just worked but many older API calls broke older apps in Windows 10.
> There's no universal unified graphical toolkit/API (and e.g. Wayland developers will not implement it) akin to Win32 which means applications using different toolkits (e.g. Qt, GTK, EFL) may look and behave differently and there's no way to configure all of them once and for all.
This isn't true on Windows or MacOS either and I'd argue isn't a necessity anymore. Don't get me wrong, I get the appeal. Look and behaviour stability of a whole slew of first-party apps went out the Window with Windows 8 (arguably earlier at the introduction of the ribbon and auto-hiding the menu bar).
For the past ten years it's been "stick the web on the desktop" so you get apps like Teams and Slack that look nothing like more traditional applications and have no consistency from app to app. Not to mention there's no behavioural consistency across platforms (let alone within each platform) in those kinds of apps.
The author throughout suggests that if there's no purpose-built configuration GUI, it's a problem. Here's an example:
> Both PipeWire and PulseAudio are not configured out of the box to support multi-user mode (e.g. multiple users sharing the same device simultaneously) and configuring them to enable this feature is impossible for the average user (no gui whatsoever, you need to edit text files).
Then they go on to say:
> ! It should be possible to configure pretty much everything via GUI (in the end Windows and Mac OS allow this) which is still not a case for some situations and operations.
If you never do anything advanced on either of those operating systems then I agree you can get everywhere with just a GUI. However, this is just factually false -- Windows and MacOS do not provide configuration GUIs for every possible setting.
If you have to open regedit to do anything, this is functionally not superior to "editing text files", and MacOS simply doesn't offer you configuration where they don't think you need it, forcing people to edit .plists or use the defaults command-line instead.
Yes, this sentence only shows that the author has extremely limited Windows and MacOS experience, because there are plenty of things that cannot be configured through the GUI in those.
I no longer use Apple computers, so I do not know if this is true any more for the recent MacOS versions.
However, when I had an Apple laptop, I was extremely surprised to discover that there were plenty of things that could not be configured through the GUI, but they could be configured with some undocumented command options added to command-line utilities whose origin was in FreeBSD, but which had been extended with additional options for OS X.
I no longer remember all the cases where I was forced to use Apple CLI tools, but at least one of them was needed to be able to associate to some WiFi access points, others than those overpriced made by Apple, because not all the WiFi configuration options could be modified through the GUI.
Also in Windows, a couple of years ago I had to install Windows Enterprise on several kinds of embedded computers and in each case I had to spend several days until convincing Windows to work acceptably, e.g. in order to work fine with a very slow SSD.
In every case, the Windows Knowledge Base was not good enough to solve all the problems, so I had to do many Internet searches through forums of Windows users who complained about the same issues, in most cases without receiving any useful reply.
Fortunately, in every case I have found some reply with the right solution, and in all cases the solution was not in the Control Panel or in any other GUI tool, but in invoking some obscure Windows command-line system-management tool, using some magic command-line options.
The only difference vs. Linux was that the Apple and Microsoft CLI tools did not have adequate documentation and only some gurus were aware of what can be done with them.
Absolute most Windows 95 era 32bit applications run just _fine_ under Windows 10, period. Sans games. Games were often coded for then current GPUs and modern GPUs have different API.
> However, this is just factually false -- Windows and MacOS do not provide configuration GUIs for every possible setting.
I've not needed to use regedit in Windows or edit text config files since forever. You can fine tune stuff but absolute most people never need it.
Ultimately, linux on the desktop needed a big sponsor. IMO it should have been Intel who had some spare billions to put the heat on Intel and keep abreast of hardware adaptation since Linux is shoehorned into so many things, but Intel was ruled by drooling MBAs and that's why it's in the predicament it's in.
Sun would have been a good one if it didn't blow up, and so tied to Solaris.
Maybe the sponsor should have been the EU. The US had OS marketplace supremacy with Microsoft and to a lesser extent Apple. The EU should have poured a couple billion into it for economic independence, and fostering their own dynamic software industry. AMD was sort of EU's hedge on Intel, Airbus was their hedge on Boeing.
Heck the US military should have done it, and still can, for effective cyberdefense. Microsoft and Apple won't open their codebases for hardening or deep dives into hardware+software security, but Linux can. Alas no way the lobbyists for the private OSs would allow it.
Linux is a lot like democracy, the worst OS, except for all the others tried.
Windows Me, Windows XP toy interface, DX 10 being Windows Vista only, Windows 8 mess up, Windows 10 telemetry, ....
Meanwhile they mess up the userspace so much that everyone is shipping Electron apps for Linux users.
! Applications development is a major PITA. Different distros can use a) different library versions, b) different compiler flags, c) different compilers. This leads to a number of problems raised to the third power."
I don't know that I can help with the other problems enumerated; but...
NixOS -- goes a long way toward solving the exact problems quoted above!
Now, NixOS does come with a bit of a learning curve however -- so it's not recommended for all types of users!
If a user has put in the prerequisite time and energy to learn Unix/Linux -- then with the additional time/effort to learn NixOS -- they can largely be free of the above problems -- which are well-known to the Linux software development community at this point in time...
Perhaps it's worth a look?
https://nixos.org/
It's a single person.
If you give me $1-2 billions (I'm not joking) I'll fix pretty much everything on the list and make Linux a better OS than Windows and MacOS combined.
I'm not a bloody programmer, secondly the list is so vast no one person can possibly fix everything.
For me, it totally works on the desktop. And since I really want to use Linux as a desktop os, I take into account the limitations and live with them. That means I stopped using nvidia gpus and just use an amd gpu or the built-in intel. That also means that to this day I still check linux hardware support before buying hardware. I live with the fact that not all games are gonna run on Linux. I bought games from Loki, I was very happy with the work from Ryan Gordon (icculus). I am happy with smaller indie games that run on Linux, since I no longer have the time to invest in big AAA titles anyways. I am a software engineer myself and do sometimes contribute to open source projects in my free time, when something is bothering me enough.
Linux does not have to gain a big market share as a desktop os to become a success, it has proven itself and it is a huge success already. Even Microsoft is investing to make the use of Linux painless from within Windows. I assume that they are aware that certain development tools just work better on Linux and that deployment in the cloud most of the time means deployment on Linux. So if Linux on the desktop does not satisfy your personal requirements, you are probably better off using another os.
A lot of the issues that are mentioned in this list (and others) are often personal preference, just wrong, or well known. However, the thing that bothers me the most is new users insisting that Linux should behave differently than it has done for a very long time, because that would be more convenient for users coming from another os... and it 'has' to change so that Linux can finally become the major desktop os! I believe a lot of Linux desktop users use the os as a work(station) os and I think that there should always be some respect for existing users. And I say that both for new users coming to Linux and for existing Linux projects that want to draw in new users or just want to change things to change things. When you use a tool to get work done and get it done as efficiently as possible, there is nothing more annoying and frustrating than having to deal with unnecessary changes in your tools.
Seriously, if a fresh install won't boot, how's a newbie not going to feel overwhelmed. For those same 20 years, people have been saying to me ( even IT people ) that they'd like to try linux but don't have the time to troubleshoot everything. Even if they sink a few hours into getting the screen to work, laptops still burn through the battery and they go back to Windows since they use it on battery.
I use both Windows and linux, and with Windows quality crashing, people keep proclaiming the year of linux. But there's still a long way to improve linux before it gets as good as poor quality Windows.