30 years of Linux: OS was successful because of how it was licensed, says Red Hat On the 30th anniversary of the announcement of Linux by Linus Torvalds, Red Hat has said that it all worked out because of the way the OS was licensed. In a post today celebrating the anniversary, Red Hat said: "The reason that Linux has been arguably the most successful operating system of all time is due to the fact that its …
This article has made me realise I've only been a Linux user for about 20 years.... I can remember a friend at University had installed it, when it can only have been about 2 years old though. At the time, another friend's Hackintosh seemed much more exciting, and Java sounded like it was going to be amazing... although in my defence, I was never expecting to program anything, I'm a mechanical engineer who was learning C as part of his degree course at the time...
My first experience of Linux was around 1997, which was *counts on fingers* about 24 years ago - we managed to get funding for the university computer society to buy a relatively high-spec machine (might even have been some dual-CPU monstrosity IIRC), and then managed to persuade the IT team to let us stick it in the corner of one of the computer labs and plug it into the network!
Admittedly, at the time, it was just this weird, almost retro text-based system which looked like something out of the movies, and which we accessed by firing up X-windows and twm on one of the "normal" computers in the lab (which generally ran Windows). I didn't really start tinkering with it more until I left uni, acquired an ancient 486 all of my very own, and got a job with internet access, which meant I could download Debian (or maybe even slackware?) onto a huge stack of floppies...
My first experience with Linux was 1997 also, on a Motorola Macintosh clone with a PowerPC 603 CPU : it had a 2.1.125 kernel (with loadable modules !) and KDE: 24 years later and I'm still using Linux and KDE. But instead of the crappy RedHat RPM hell I use Debian-based distributions.
That's the problem. Good "enough", but never actually, objectively better than the proprietary counterparts.
That statement has some problems. First, it is begging the question that people claim Linux IS more secure than the Microsoft or Darwin kernel. Not sure if many people claim that outright, there are patches, configurations, and SELinux that improve your generic kernel, which is oft for ease-of-use, not security.
My biggest complaint about your article is the cherry picking of examples. Sure you can find examples where proprietary kernels have potentially better security models. That is why climate deniers, anti-vaxxers can always trot out some idiot to argue the against the consensus that an university was foolish enough to give a PhD. Doesn't mean that those examples are not genuine security issues that need addressing. Proprietary kernels are black boxes; it is easy point out all of Linux's flaws but not see all the potential architecture problems or hacks that keep proprietary kernels operating.
Sidenote: does the Linux kernel even touch fonts? Microsoft pushing something to userland that should have never been in the kernel isn't a great flex.
So objectively? Incomplete information, I'll just stick best practices over relying on my kernel.
I've seen many examples on various forums (here included) of people saying that Linux is inherently more secure than Microsoft. I've even been downvoted (and called a Microsoft fanboi) for suggesting that while Linux is secure, it's not invulnerable.
I am not judging Linux when I say that. I don't believe any software is invulnerable to those who are talented and motivated enough. That's just a fact of life. Nothing humans build is perfect.
well i can say that "Linux is definitely more secure than windows" because i can prove linux's strength and weaknesses by looking at the code and without causing a world incident by disclosing a vulnerability that MS minions won't patch anyway (think about PrintNightmare).
The point here is that if you find a weakness in linux, and if you have sufficient skills, you can patch it yourself and push the fix to the rest of the world, whereas with proprietary kernels, even if you let the companies know that they have a problem, it is NOT guaranteeed that they will do anything to correct it, until it's maybe too late.
> Proprietary kernels are black boxes; it is easy point out all of Linux's flaws but not see all the potential architecture problems or hacks that keep proprietary kernels operating
It's a trade-off: open-source is theoretically better, since anyone can see the source code and fix it. The problem is that the number of people with the holy trinity of resources needed to do this (aka: the skills and experience to find and fix things, the time (or funding) needed to make and test the changes, and the motivation to work through the processes to get said patches rolled) is highly limited.
Meanwhile, black-hat types are far more motivated to go crawling through source code looking for flaws, since there's a potential monetary reward waiting for them.
So, yeah. You pays your money (or not, if it's open source), and you takes yer chances.
Anecdotally, I've recently had a major issue with an Ubuntu kernel upgrade on my Lenovo Thinkpad laptop; processes were spinning out of control, hammering the CPU and then turning into zombies when killed, which were both linked to the "id = 1" parent process of the entire server *and* still chewing up major amounts of CPU resources.
The only solution was a reboot, which I had to do every 2-3 hours, until I spotted that the kernel had been updated and reverted.
And that brings me back to the point above: this is a pretty major bug to be sneaking into a kernel, and one that maybe (at a stretch, admittedly - maybe a denial-of-service could be triggered) could have hacking implications.
But, I don't know if I have the skills or experience to go delving into the kernel to figure out what's going on, nor do I have the time to do said delving, and since it's a work laptop, I definitely don't have either the time nor motivation required to go and figure out how to submit a bug report or submit some sort of patch.
Instead, I reverted back to the previous kernel, and I'll leave things as-is for a while to see if a later kernel update fixes the problem.
I would say proprietary systems are a bit more like climate change evangelists claiming that co2 can drive climate without releasing the data in the same way that MS vulnerabilities are patched without MS revealing what really caused them in the first place.
Of course if climate change could be driven by humans then it would need a gas with a much higher global warming potential than 1, but I guess you know that don't you?
Yes, what success Linux has had is down to GNU and the GPL, no question, but it is **NOT** an OS. It is a kernel. It is critically important to much of the world, but so are many other common components that form part of an actual, functioning OS.
OpenSSL, SSH, BASH and on and on and on.
Those get woven together to form working OS which allows people to get things done.
El Reg is meant to be a tech new site and it can't even get the most simple thing about Linus correct.
Yeesh.
"To have a complete OS you need at least a kernel, an init system, and the toolchain to compile it. Therefore, Linux alone is NOT an OS"
Then nor is MacOS, Android, and others.
All Linux distributions include this. Are we really going to argue the semantics about this?
You are correct. Linux is a kernel.
However all of those other parts making up the OS are available on Hurd (Kernel) and *BSD (operating system) too. However they still don't quite have the same popularity (though my myself am an avid BSD user). So arguably these extra components are less important to the popularity of GNU/Linux as a platform compared to a) kernel b) license.
So really the article talking about Linux specifically rather than all the userland "stuff" is relevant and makes sense in this context.
What would be the reason of GNU/Linux being so popular ?
Probably marketing and hardware support. The BSDs really didn';t have RH and other commercial support/distributions. And so Linux grew in popularity because you could buy it at the local computer outlet, and it (mostly) supported YOUR hardware. That kind of thing.
Sometimes that's all it takes.
BSD got held back by a stupid lawsuit by AT&T, which led to people believing they (as users) could be sued for using it, meanwhile, GNU was free of lawsuit FUD and just needed a functional kernel to be made whole; a role which Linux happily fulfilled. It really couldn't compete once the GNU toolchain became a defacto standard (even among proprietary UNIXes as a required supplement). To this day, the GNU userland is still far more fully featured than actual UNIX tools, meaning it isn't just the kernel which suffered as a result.
That said... This same stupid legal minefield is the one which holds Linux back in the modern day and most of it caused by the "land of the free" which clearly favours non-free software and restricting what people can/can't do with the products they buy. Apparently, it's fine that Windows and macOS include code which isn't legal to use in a commercial setting without separate licences paid (by the end user, read the notes in their respective EULAs) to various entities but it's clearly the end of the world for commercial Linux vendors to do the same; meaning many distributions will always be crippled-by-design (missing codecs, hardware audio/graphics acceleration support etc.) in their vendor-supported configuration.
TL;DR; Laws are there to be broken. If the people in charge had bigger balls, we'd all have better computers right now as a result.
Cobblers!
My pet peeve is people talking about an OS when they don't actually mean the OS, they mean the applications running on the OS. Bash - for example - has nothing to with the OS. It's an application. I can run Bash on my Windows PC, but that doesn't make it Linux.
The OS is - or should be - the kernel, the internal nuts and bolts, with next-to-no (or even just 'no') "things that a user can run". Because things that a user can run can - practically by definition - be replaced with "other things a user can run" and if you replace one such thing with another, that would obviously *not* mean you're now running a different OS.
An OS controls higher level programs' access to the hardware. Linux does that. Bash is not an OS, it is a user program. SSH is not an OS, it is a user program.
Anything that has to ask Linux for resources or access is not part of the OS or it would do these things itself.
This is not a complicated concept.
Oooohh, look, a NIT to pick! pick, pick, pick, ...
That's kinda how I read that. [most nits aren't worth picking]
As for the article, I didn't realize Linux was already 30! I guess I've been using it since it was 12...
and a "well done" for the use of [L]GPLv2.* (I think many of us would be unhappy if it ever went to 3.*)
The polemic untertone of the TS and the downvotes he's received in turn nonwithstanding, his statement is factually true: Linux is not an OS, it's a component of an OS, a kernel. And that is not just a nitpick: it is the real reason why "Linux" was never widely adopted on the desktop.
For (leaving Android aside for the moment) "Linux" in the usual sense of the word, namely: the Linux kernel plus the GNU userland tools (and a few others) is in reality a /family/ of OSes, all of which are closely related enough to look "mostly the same" from the (Unix/Linux nerdy) end-user PoV but which in reality present a bewildering variety of ABIs that makes packaging /commercial/ applications for "Linux" a chore, particularly where end user apps with a GUI are concerned. And one which doesn't yield an attractive remuneration either, as, most software that is being used on GNU-Linux being FOSS, the average Linux nerd expects commercial applications on "Linux" to be cheaper rather than more expensive as compared with Windows and OSX. For a commercial developer "Linux" thus means: more effort for less ROI, or in other words: "Nah, not worth it"...
However, no OS can ever thrive on FOSS end user applications alone. That is a hard fact of life. And this is why "Desktop Linux" has remained a niche product. Sure, lately we've seen a plethora of novel packaging schemes where each application has to provide its very own userland runtime environment, either directly or by pulling in dependencies, the only viable option for portable applications in the Linux environment, apart from linking the whole damn thing statically on an ancient kernel version. But it's all too little, too late and still too balcanised to stand a real chance of becoming a game changer.
And there's another problem: There are use cases that require specific hardware drivers the implementation of which must remain a trade secret, such as operating high-end audio hardware, which is all but impossible in a GPL-licensed monolithic kernel. For high-performance GPUs at least, solutions exist, but I suspect those are only being tolerated because even the most radical GPL-Taliban want a good FPS-game from time to time...
Thus it turns out that the licensing (which created the distro balcanisation as well as the driver problem) is indeed responsible for both the success of "Linux" (in the server room) and its dismal failure (on the Desktop).
And Android? Well, Android does not have problem #1, as it is factually a proprietary OS (with a FOSS core component, just like OSX) and problem #2 is not really relevant on a smartphone, where you cannot swap out the GPU or attach an external high end audio interface. Which proves my point.
Feel free to vent your frustration by downvoting me, but know that it ain't gonna change a thing...
Define OS in your world without introducing arbitrary and subjective opinions.
Your thoughts on what is an OS are simply wrong.
Moving on to GUIs, you're on firmer ground but still fairly soft. In reality the game has been changed and continues to change, but you're using the scorecard from the old game. I don't care if hardware that needs to use secret sauce to work has a problem with Linux; that's the old world and I'm happy for it to die a death.
The GPL has delivered me the computing I want, complete with a desktop I've used for 20 years while Windows and OSX failed to provide what I consider basic functionality. If you think Linux has failed because it hasn't reproduced all the same stupidity that paid-for-but-magically-not-owned software has, then there's not much I can say to you except that I hope you continue to enjoy living in your nice, expensive prison.
"even the most radical GPL-Taliban want a good FPS-game from time to time"
Good to know I'm not a GPL-Taliban then because I couldn't give a stuff for games.
And it comes as a surprise to discover that an OS can't thrive on FOSS end-user applications alone. The one set of non-FOSS applications I do use - occasionally - is as much for nostalgia and familiarity from Unix days than anything else as most Linux users do actually use FOSS equivalents.
To those who voted this comment down:
An OS consists of a kernel (or executive), a shell, and a collection of system utilities.
This is the original definition of the term, which hasn't lost validity just because marketroids are unable to grasp the details of technology.
Linux-based OSs consist of a Linux kernel, one (or more) of a variety of shells, and a bunch of utilities mostly from GNU.
I'd refine that slightly. An OS consists of a kernel and a standardised operating environment that application programs can assume will exist and make use of. So this may include a shell and will almost certainly include a set of libraries and utility programs. Some of the environment may be optional, for instance the X window system and associated libraries is an optional component of the Linux OS. An application which needs a GUI will use X but not all applications may need a GUI.
The utilities/libraries in an OS do not have to be exclusive to that OS, they may be used on a number of different operating systems.
No. An OS may consist of nothing more than a kernel.
The earliest computers most certainly had an operating system. But they didn't have - by any contemporary meaning - a shell or a collection of system utilities.
The "operating" word in an OS refers to the operating *of the hardware". It does not mean the operating *by the user*. An OS is the software that operates the hardware.
After all, an embedded OS may well have no shell or "system utilities".
An OS may typically contain a kernel, a shell, and other applications/utilities. But to suggest that a kernel on it's own is not an OS is just factually - historically and currently - false.
I have been using Linux on the desktop (and laptop) for a number of years and I love it. However, in order to do that with CentOS 7, I had to "hack" things in order to use MATE desktop (Gnome 3 sucks!) and kernel-lt that supported certain hardware. Whilst M$ Windows off the DVD still needs drivers, consumers rarely see that, as shop-bought systems are usually ready to go. There has never been an easy path for consumers to migrate to something that was familiar. Another commenter on another thread is complaining about the case-sensitivity of *nix compared to Windows - something I have seen catching people out (web "developers" with spaces in their file names...).
There there is the lack of application support. iTunes for Linux, anyone? The software houses claim there is no demand, so they do not create a Linux version, which means you cannot use their software on Linux, which means there is no demand. Flatpak may be able to fill that gap. It has certainly helped me run the latest Skype, Signal, KiCAD, and OpenShot Video editor on CentOS 7. For other applications, I have had to use mock to rebuild upstream Fedora packages so I could use certain Amateur Radio applications in CentOS. I suspect Red Hat assume those who want to use those apps will just use Fedora, not an EL version on the desktop?!
And we have fragmentation. I have already stated my dislike for Gnome 3, and thankfully, others agreed and made MATE. With so much choice, it is really hard to create documentation, or tell a friend/relative over the phone to click on X, then Y, to sort something. This of course is also true for Windows with the UI constantly being screwed with!
And of course, we have to have the distro argument. Which do you use? Debian and its clones, or Red Hat and its clones. Different package management, different paths, different release and support cycles. I like to use the EL versions so it "just works". Unfortunately, the toys I want to play with are only available in the bleeding-edge versions, such as Fedora (speaking from a Red Hat perspective). So I have to go without, rebuild, or run it virtualised. A recent issue has stopped me dead in my tracks from trying to migrate to Rocky Linux 8. The rather handy 'mail-notification' utility is not available in EL8, as it needs libgnome-devel, which is deprecated (also an issue with Ubuntu, according to posts I have read). Attempting to rebuild it on a Rocky8 build VM, and I hit a wall where libgnome-devel requires compat-openssl10, and that refuses to build the -devel package, as it clashes with openssl-1.1.1. So I am stuck. No EL8 desktop for me at the moment. Do I go back to running Fedora to save on rebuilding the toys? The fedup upgrade utility is rather handy, and you can stay on the last supported version to try and avoid all the things that are borked, but you still have to use bleeding-edge Firefox, and they like to break add-ons! I am afraid it is of little use to suggest: "just switch to Debian". I have spent 20-odd years playing in the RH field, and trying to get my head around where things are in the Debian clones gives me a headache. Apologies to those who love and maintain Debian, but the path and packaging differences are another reason why M$ Windows remains the dominant "thing" on people's desktops and laptops.
Feel free to downvote. As a SysAdmin of 20+ years, I have seen lots of changes, not all for the better. I now simply want things to "just work" and not have to spend hours trawling through changelogs and forum posts to find someone decided to change how one package worked, and now my entire IT estate is on the fritz! I am looking at you systemd.
Upvoted because you correctly point out that Gnome 3 sucks. And also for systemd. That is just more shit to learn. It does not make things simpler. Bring back SysV init and Gnome 2. :-)
El Reg also points out there is no place like GNOME: 41 is in beta (aka Gnome 4). Oh fuck.
That is just more shit to learn
I'm sure there are many 'SMUG's (Small Minded Useless Gits) who would twist this into a "you REFUSE to learn":comment of some kind, and then go on praising the glories of Gnome 3/4 and (the cult of) SystemD. (or Win-10-nic for that matter)
(Even though the rest of us understand that Gnome 3/4 and SystemD are NOT Linux)
From the article: "While I want very much to be using one of these operating systems in the future, I could not recommend one of these devices for a 'regular person' at this point, and for myself, they're not capable of serving as my daily everyday device yet."
I now simply want things to "just work"
As an IT Manager of 25+ years and IT professional for 35+ years I agree. Despite all the complaints for the most part Windows does "just work" and, more critically, users are used to it.
I've been saying this for years here on El Reg, with very mixed opinion results.
80%+ of desktop users just want this thing to work. They not only have no idea how it works in the first place, they have no interest in finding out why it does. They just expect this machine to do what is expected of it.
Like a toaster. Or a microwave. Or their refrigerator. Or, for a lot of people, their car.
Devices are expected to work. The age of knowing why, and being able to repair it yourself when they don't, pretty much peaked decades ago.
I switched a female relative's computer from Windows Vista to Devuan Linux a couple of years ago. Not only did it fix the thing so it worked well again, it did nearly everything it had done before (but better) and did not need replacing with a "New, Shiny" Win-10-nic machine. She knew very little about anything computer-related except using mail, web browsers, managing music files on an iPod nano, playing videos, and playing games. Well, some of the games she played as a kid were no longer available (but I offered to set up the games in a VM if she needed it, so far no).
Her successful OS paradigm transition took less than a day. Mate desktop helped. Everything just worked. It was an older Sony Vaio laptop, and I set it all up on a new hard drive. No problems since.
I would say that, for her, Linux is kinda like using a Mac. Switching from PC to Mac is generally painless (except for the bank account). Switching from Windows to Devuan+Mate, also pretty painless (and the hard drive needed replacing anyway, so no harm to bank account).
More people need to understand this.
It depends upon your computer usage requirements. For many home users, who like the relative you described do rather simple everyday things, absolutely, the switch from one OS to another, any OS to another, really isn't much of an issue. The fundamental programs that they interface with, web browser, mail, etc, are somewhat the same across many platforms.
For some of us, power users or professionals doing a job using specialized, high function or industry-specific tools, no, switching is pretty much out of the question. I use a specialized industry-specific integrated inventory control / billing system at work, plus InDesign and Photoshop, plus a legacy photo catalog application to track our 1,000+ designs that also interfaces to a USB microscope camera through to a Ethernet-to-USB gateway.
My co-worker uses a very specific, industry-oriented application CAD overlay on top of Rhino3D to get all of our CAD modeling work done, plus the very specific, Windows-only applications to analyze, package and transmit the designs to the 3D printers and milling machines.
At home I use Photoshop and InDesign, a Windows-only API interface to remote control my cameras during product shoots, plus DaVinci Resolve.
In other words, no way-no how can / am I switching away from Windows without suffering an almost complete collapse of my productive workflow, both at work and at home. My productivity is directly interwoven with Windows software compatibility.
So YMMV with the idea of Linux being a personal solution.
Yes, moving OSes will always be dependent on the critical software being available on the new OS.
In a previous post designing silicon hardware the critical software was linux based (presumably a descendent of unix software) and wasn't going to be portable. To help out it was decided that everyone should have a windows PC (presumably for word and email) and the business critical software used by the bulk of the employees (but not the management) was put on a remote linux session. *sigh*
"So YMMV with the idea of Linux being a personal solution."
Whilst that is undoubtedly true, you have already acknowledged that, for the vast majority of users, Linux is a satisfactory personal solution. Of course, for you, switching is out of the question. Although, in the unlikely event that enough of the developers who supplied your software tools switched, then not switching with them sounds like it would also be out of the question.
The Windows world has the SAME issue with respect to OS Version upgrades ... suffered huge losses of productivity with Windows-Based APPLICATION SOFTWARE incompatibilities when forced to update from W2K to XP then XP to W7 and recently W7 to WX ... W11 is next on the radar.
MS has never provided migration paths for even its own Applications either forcing people to purchase 3rd party transfer software or having to purchase new compatible Hardware and Software ... Printers are a prime example ... Office Products are another Pet PEEV MS won that game by making a suite of office tools ... then turned around and completely changed their UI more than once
VM's can mitigate some of the incompatibilities but HW dependant apps still suffer ... have a VM running Xenix (an MS UNIX variant) with an application circa 1986.
Today's shinny computers (WITH CHICKLET KEYBOARDS) seem to only work well as long as you use them in the way the vendors designed them ... at the start of the PC revolution, even though there was a huge amount of upstarts, IMHO, at least there were plenty of SOLUTIONS available ... today the big players just continue adding unnecessary FLUFF and or completely redesign UIs offering less and less user consideration or support ... Linux or any other UNIX flavour (Ubuntu etc) provide a basic Computer Platform that can satisfy the majority of non-commercial users.
There is another revolution in which UNIX flavours are dominating and that is in the field of IOT
<<Windows-Based APPLICATION SOFTWARE incompatibilities>>
I'm still running, on a W10 PC software written for W98. The software I've written in Delphi happily runs on anything from XP onwards.
I don't have any older boxen to try on but I may see if I can install W2K on something just to see. I suspect it will be the hardware that will be the problem.
Much as I like to diss Microsoft their backwards compatibility is generally pretty good.
I have to agree. My circumstances are different, retired but still playing with computers. I have 3 x W7 machines, 1 x W10 machines and 1 Linux machine (Mint 20.2 Cinnamon) which is an 11.6" notebook for casual use whilst blobbing out in front of the TV.
The Linux machine works well (better than it did with W10), mainly runs a browser (Slimjet) and a couple of apps I've written. Those use WINE.
I could duplicate my development environment from my W7 machine but the amount of time to either find alternatives to all the software acquired over the years or re-write my homebrew stuff would be massive.
If only I'd started using Linux when I was developing on PICK based minis or writing code in Delphi.
Couldn't agree more. A 78 year old friend of mine had also a Sony Vaio which wasn't working too well under Windows. I did what I could to make it a bit better and also installed, with his agreement, Linux Mint. I said 'just try it, if you don't like it, I'll remove it'. A few months later, he came round, to have Linux removed, I assumed, but no, it was Windows that went. He is almost computer illiterate, but loves the fact it's so reliable and quick.
A few months ago I replaced a previous laptop with a new one as my daily driver. The old one have been bought with W10 but replaced with Linux. New I've shunted back in its original hard drive and reinstalled W10 from the reinstall medium.
My what a mess.
It was restored to - I think - autumn 2017. It repeatedly failed to run updates to what was effectively a virgin W10 install. And it persistently complains about a bad exe for one of the many cloud services it keeps trying to flog to me. Searching tells me that these are common afflictions and none of the advice given works for me nor, it seems, for many if not most of those who enquired in the past.
The UI is tamed a bit by unpinning every single live tile so I'm left with a menu that looks fairly sane except that it's strictly alphabetically rather than functionally organised. Even previous Windows versions were functional alhough memory is hazy as to the extent that they could be entirely customised in the way that KDE menus can be.
Updating by the expedient of downloading and installing what amounts to a rolled up service pack got over the stalled updates and left a few duplicated menu items and a few others I'd like to get rid of as well. Some can be uninstalled from the menu but others take me to the general purpose install/uninstall app of yesteryear that used to list all the installed programs so that I can use that to uninstall them - except that none of the non-unistallable programs are listed.
My what a mess. And people are prepared to pay for this.
Having just complimented MS on their backwards compatibility I now find myself agreeing with this. My story:
Recently, I was given a laptop that had been bought with Vista. This had been changed for Linux at some point and I decided to install W10 (can't remember why now) - no problems. I then decided to bung more RAM in (from 2GB to 4GB) and an SSD. Cloned the HDD to SSD using Macrium Reflex - all good.
Now comes the fun part. A mate has a similar spec laptop and I suggested installing an SSD. To give him an idea of real world improvements I decided to do some speed tests. I still had the original HDD so I thought - get it up to date with W10 and with the various apps I'd installed. Off I went to clone SSD -> HDD with Macrium - seemed OK but when installed in the PC no way would it boot - it seemed to tell me that winlogon.exe was missing (it wasn't).
Over the next few days I tried repairing the HDD, cloning the SSD to the HDD using a second machine. No go. I tried installing the latest W10 iso. No go. Eventually I installed W10 from an old DVD burnt when it came out and was a free upgrade. That worked. Just another day or so to update and install some software and I was there.
However, bad as that was for a real sob story I'd have to tell you about my attempts to install Linux onto an ASUS X205TA.
The UI is tamed a bit by unpinning every single live tile so I'm left with a menu that looks fairly sane except that it's strictly alphabetically rather than functionally organised.
Congratulations on completely and utterly failing to understand what the tiles area is for.
Forget about the "live tiles" bullshit. That space is there for you to organise shortcuts for the applications you use regularly in a way that makes sense to you.
The alphabetical list is just a much cleaned up All Programs menu for everything else.
The software houses claim there is no demand, so they do not create a Linux version
True, and yet there are still many Linux applications that are commercial.
Ohe of my observations is the way runtme and package linkage are too connected to the distro. You mention Flatpak which would attempt to rectify this. My preference: static linking. Then make updates (to THAT version) free if there are any important fixes later.
Probably the biggest obstacle to cross-platform is the use of "Micros~1 Shiny' as a framework. Old MFC applications could be adapted (using something similar to wxWidgets which is very MFC-like) as long as they're not using the latest 'Shiny". The 'Shiny' always has a lock-in effect to ONLY supporting Windows. And nobody who is SANE would want to use ''.Not" to start a new project (or to support an OLD one for that matter). LOTS of attempts to port games, etc. but Micros~1 still provides their "Shiny" to keep as many devs SOLELY on THEIR platform as possible...
(Seriously this should make Java even MORE popular, I would think)
Don't bother with the clones: come and talk to the Debian developers. Step forward six or so years from CentOS 7 to Debian 11. The reason there are 200 Debian derivatives rather than 15 or so for Red Hat?
Community - which Red Hat have rather squandered and fragmented in the last year or so. If Fedora is too fast pace - you've nothing. (Though the Fedora developers and users are a great community in themselves).
MATE - no problem - you don't even have to find a different install medium. www.debian.org will give you a download link for the netinstall medium.
CentOS 7 runs out of support in 2024, so lots of people will be looking for an EL alternative. For servers, I suspect lots of people will migrate to Rocky. Desktops/laptop are going to be a little trickier, especially for those of us with special usage cases. There are days of testing and debugging involved in switching versions, let alone distributions!
Around 2008, I worked in a company who employed one of the Debian maintainers. He had the honour of leading the whole project for a period. The IT manager asked him where we could procure the same sort of official training as you can find with the Red Hat Certified Engineering course. The question has never been answered. The things you get to play with at Uni do not always translate into the commercial world. As I told a college lecturer on my BTEC HNC Electronics Engineering course, those of us on day-release had full-time jobs, and we were paid to bash things into a calculator and get on with the work, not spend half the day working things out from first-principles on a bit of A4 paper.
In a more recent company, I was forced to rip out a Debian based business-critical customer-facing https WebDAV server and replace it with CentOS 7. Whilst I applaud and support the community-driven nature of Debian (and other projects), it was too slow to back-port essential security features that Red Hat had already taken from Apache 2.4 and stuffed in 2.2. My CentOS 7 server at home was scoring better on the Qualys SSL tests than the biz-critical system at work, so it had to be replaced. The security of customer data was more important than the platform it ran on.
There is also an issue of commercial support. In the same company as previously mentioned, I was in charge of a large-ish IT estate comprising both Dell and HPE kit. The EL packages they provide are essential for SNMP monitoring, iLO/iDRAC integration, and RAID controller management. I have been able to re-configure and re-build huge RAID arrays whilst the system was still running, and swap failed power supplies within minutes of receiving the email alert. Debian, and to a certain extent Ubuntu, are largely ignored and left to the community. HPE/Dell/Oracle are only going to spend time and effort supporting those versions of Linux that make them money, especially when they are spinning their own!
Tell me: Where can you get commercial training for CentOS? Commercial support? LPI will train for all Linux. To be honest, there's almost no difference between Red Hat and Debian now: same systemd, same apps, same desktop environment. [I've CentOS 8/ 8Stream / Rocky / Almalinux and Debian 11 on one set of VMs - put a user in front of any of them and you'd only notice that the Debian is more up to date and slightly more jazzy than Rocky]. RHCE is a course which demonstrates how to pass an exam.
Dell and HPE will support Ubuntu and Debian now: Lenovo are preinstalling other Linuxes - and second hand Thinkpads are the stock machines for savvy users.
Re "The reason there are 200 Debian derivatives rather than 15 or so for Red Hat?"
I actually think that is the problem with Linux. Choice.
Don't get me wrong, choice is usually good, but when you hundreds of derivatives ,with little or nothing to distinguish them, it's hard to decide which to use. Certainly if your sole experience of computer is basic, such as maybe only having used Windows, and wanting to explore other options.
I think that's where the likes of Ubuntu, and even Raspian are good. They are both Debian derivatives, but are building brand recognition. They are both heading for the point where the average punter in the street who has a mild interest in computers (ie is a little interested in how they operate, rather than just using them as a device to do stuff on) will recognise the name. Ubuntu because Canonical have done an excellent job of marketing it, and Raspian because the Pi itself seems to generate a fair amount of publicity.
+1
Been using Debian GNU/Linux (+ KDE) as my main desktop since 2002, so almost 20 years.
KDE 3.5: Brilliant. KDE 4: Too much 'alpha' for most people - too unstable for basic users and too many features missing from 3.5 for the power users - and it did KDE a lot of damage - I held out with KDE 3.5 for ages, even tried trinity for a while.
These days KDE 5 is where 3.5 used to be: It's stable, configurable, and awesome. Windows-beating in every way IMO, and it isn't sending buckets of 'telemetry' to microsoft.
I came to Linux one night in 2016. I pushed the 'reset' button on my PC halfway through a bare metal Win98 install (3.5hrs in and counting)
Linux cames a cover CD from a UK PC magazine and I had it fully up and internet connected in 45mins
It's been "good enough" since then to be my main PC OS, (Windows been relegated to a VM box for at least 15 yrs)
All power to Linux for the future
It doesn't really add up at all, since Win98 was released 23 years ago. .
Maybe he meant to type 15+ years and counting and also meant to type 2006?
+1 anyway because Win98 really does belong in a VM only. At least there it has some kind of memory segmentation..
I came to Linux one night in 2016. I pushed the 'reset' button on my PC halfway through a bare metal Win98 install (3.5hrs in and counting)
I'm guessing that should have been 2006? I can't imagine why on earth anyone would be installing Windows 98 in 2016.
I hope it was at least 98 Second Edition...
Doing some clearing out of the spare room during lockdown I found my original first edition Yggdrasil distro booklet, CD and floppy disk media from 1992!
Yggdrasil was the first commercial live distro ever made released, a 30 page booklet with bootable floppies and the CD, if you had one of the handful of CD drives that were compatible back then. I remember a mate gave it to me after a visit to a tech show where it he'd bought it discount for £30, he didn't have to time to play with it so he lent it to me and I never gave it back.
A little piece of Linux history I'm proud to own!
I got a red fedora when I bought RH 2.0 (?) on CDROM.
It has the old all-lowercase RH shadowman logo inside, and was apparently made by a company named "Lite Felt"
Apparently they're still a thing: https://www.millerhats.com/store/Lite-Felt-Hats/Lite-Felt-Fedoras
By chance I was once in someone's house and noted a red fedora on a hat stand. Apparently they had been in marketing a Red Hat in the UK previously. They knew less about computers and IT than my mother-in-law's dog, but they'd kept the hat!
They were please that I recognised the branding - which I suppose was the point of it!
For the most part I run the installer (and users of certain other systems get to practice that quite a lot) and things do Just Work. What is really great is it then Just Continues To Work. Unlike some.
Although Bluetooth and WiFi are the worst faff I'd had the misfortune to endure in a very long time. What do most of the mundanes want? Right, most of them have never seen a network cable. Anyone working on the UI for those, please take a look around at the other OSs. Even if it's just Android. Biggest blocker as far as I can see.</rant>
Now I've said that, my personal desktops have been exclusively GNU/Linux since about 1997. It's almost always been capable of what I want from it, and it just keeps getting better. Anyone who has contributed to that, have my thanks and a few of the icon.
I do worry about what will happen with Linus does eventually step aside, which is inevitable at some point.
It will leave quite the power vacuum and let's not kid ourselves: quite a few large organisations would love to get control over such an important piece of software.
They would never dare with Linus in charge, but I do worry about there being a proper succession.
Microsoft didn't just vanished after Bill Gates left the company. Same thing for Apple, HP and many others. Neither will do Linux.
Due to GPL, large organizations will not get control unless they fork it and change license to a proprietary one but they can already do that. If they still decide to do it, they will have to maintain it which would turn into a too big a headache. So big that not even Microsoft itself would consider it.
Anyway, Linux blueprints are all over the web so any interested party can continue working on it. Just look what happened when Red Hat / IBM tried to kill CentOS. Worst case scenario, we still have *BSD.
since Fedora 6.
Got fed up with m$ and windows..... 'Please re-install every 6 months for the best results' sorry but us normal people dont have a day to waste installing win 98/update pack then hunting down the GFX and sound drivers... and various other drivers.
So fedora 6 (as it was a redhat product) .... partion........ install ..... login and got myself a nice dual boot machine oh and the codecs? installed via the software installer. Only thing I had to hunt down was the DSS module for DVDs.
And glad to say I've been a dual booter ever since .. windows for games, linux for work.
(and if enough of my games worked native on Linux.... Linux for games too)
Nobody ever put to serious use a computer running only the OS. I did not make much use of Linux until Star Office came along. Since then I have used SO v5, SO v7, Open Office v1.o to v3, and Libre Office v3 to v6.
I look forward to a feature in El Reg about these office products.
I remember first downloading Linux when I was at University (around 1995). I think it was Slackware. It was split over 20 floppy disks. I remember that downloading the images of the floppy disks was fast (for the time) as the mirror site was on the Janet network. However burning the image to floppy disk took ages
Linux will never make it as a successful desktop replacement due to a simple fact - it's licensing model - i.e. it is free!
All desktop overlays I have tried have been just ok but not great, they are clunky, don't look great but I have to admit most do get the job done. Unfortunately in the world of the consumer "User Experience" is key (dare I say it, Apple is probably best at this) and this takes a very different skill set - not individuals who like programming at the lowest levels but teams that include creative designers, as well as programmers who can take the design, understand the vision of the designer and translate that into code in order to bring the initial vision to life. This takes a lot of collaboration.
Government of Bavaria moved to Linux and Open Office to save costs and breakaway from Microsoft. 10 years later they confirmed the experiment failed and moved back to Windows - I believe the cost of the experiment was astronomical in the end (could be wrong but final figure was into 10 figures!)
It will take a brave company that attempts that experiment again...
They themselves are "trying that experiment again"
I'm sure the move to MS had nothing to do with MS relocating a regional HQ there.
Still, they are moving back to Linux: https://www.zdnet.com/article/linux-not-windows-why-munich-is-shifting-back-from-microsoft-to-open-source-again/
The conditions that caused Android to arise mostly boil down to Google's desire to turn mobile users and their data into product as they had done with desktop search.
That was always going to happen and Linux was both an excellent and convenient place to start, but not the only place: Android did not come into being because someone at Google thought, "hey - cool kernel what shall we do with it?" so for Red Hat's Mike McGrath to say "no Linux […] no Android" simply isn't true.
There's a bit more to my comment that Tim left out (to be clear, I don' think he misrepresented me). I had added something like "Android may have looked like something else, but it wouldn't be what it is today" something like that. I agree with you but we're not in an alternate universe where Linux didn't exist, we're in the one it does and really the only distinction today with people is whether they define Android as Linux, based on Linux, or something else entirely. The distinction, to me at least, is academic.
That would have been around 97. It came with Mosaic. Mosaic wouldn’t work with the Netscape website - I had to ftp my first usable browser. Other than that, I had in my own hands the first proper OS on my own hardware since the Amiga. It was an experience and then I forgot it for a few years before one of the nerds at the office converted me to the healthy goodness of Ubuntu.
Linux on a desktop? Honestly, if I were starting a business I think I would have Microsoft for specific and limited reasons. Mainly checking the office templates produce documents that interchange well between us (Libre Office) and customers (inevitably Office). I reckon a business could save a fortune taking that approach. Overall, I’d prefer a Linux desktop plus Windows VMs (or a few copies of Windows accessible via Remote Desktop) for those who need them.
Just my opinion. I don’t blame anyone for going with what they know and the Microsoft tools are more than adequate for their purpose. What I don’t get is why people use Windows for development but that’s probably just me being a bit blinkered.
The GPL definitely helped the kernel but let's remind ourselves that GNU Hurd was GPL. And it it should have had a natural advantage when it came to attracting developers because the project kicked off first and was backed by the FSF.
But Hurd had such a pathetic pace of development, mired in politics and correctness that it actually spurred Linus into making his own kernel. His kernel wasn't pretty or "correct" but it did something their kernels didn't - it ran, did the basics and became a bootstrap to build out a userland of software. Given how expensive commercial x86 Unices were at the time, it is no wonder it was greeted with enthusiasm and took off.
In other words by taking the pragmatic approach of producing something, no matter how imperfect and iterating on it, the Linux kernel succeeded and Hurd died on its arse. I'm sure eventually Hurd will reach a version 1.0, but honestly the time for that was decades ago.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
