Take your pick: Linux on Windows 10 hardware, or Windows 10 on Linux hardware Enthusiasts with time and hardware on their hands have a few extra options for the weekend. One committed group of Linux fans has got Ubuntu working on a Windows Arm laptop, while Pi fans have made it easier to bring full Windows 10 to the diminutive computer. Pi fans get the full Windows 10 experience Windows 10 IoT Core has …
Been running Kubuntu & Xubuntu on ARM for 5 years. Well as croutons on ChomeOS (which neatly avoids the driver issue) - but hey they work brilliantly. Live swopping between the two on a tiny 2/16Gb laptop never ceases to amaze me.
But bring it on - the Linux apps i386/x64 only app developers need a bit of incentive to make their goodies architecture agnostic.
One of the beauties of Linux is that it pretty much doesn't matter what it's running on. So long as you have drivers, you recompile your kernel and software and off you go.
I often forget whether I'm on a Raspberry Pi, an Ubuntu-based desktop in a VM, or a remote dedicated server running Apache - from SSH they all look and work the same and have the same software available to them, and the only things that differ are distro-specific.
I've actually sat there trying to work out why a pre-compiled binary I uploaded over WinSCP wouldn't run... and then realised the machine was an ARM-based Raspberry Pi and the binary Intel x86. But it didn't matter, the software had source... so I just downloaded the source package instead, recompiled (using the same CMake/gcc etc. process as I would anywhere else) and installed and it worked first time.
I have to say... if you said to me "start IT all over again from nothing", I'd be hard pushed to see a reason to include either a closed-source OS or closed-source binaries at all. And can see no reason that any OS shouldn't be available on all the major architectures.
I even struggle to understand why programs for things like Linux aren't distributed in a single file that includes source and binary for each platform (a bit like the old Mac "universal binary" but with source as the first ELF section, and then each platform - if there isn't already an ELF section for the architecture, just compiles the source on first run, and puts the resulting binary into a new ELF section for that architecture). Then literally your program will "run" on any platform that's compatible. It'll automatically be optimised for the platform in question. People can still distribute programs that will work with zero-compilation necessary for the major platforms, and porting to a new (compatible) platform is as simple as running the program and waiting a little while as it compiles for the first time, and platforms that don't need to load a particular architecture might have a larger executable, but it can be safely pruned if necessary (embedded versions) and wouldn't load the other architecture's ELF sections into memory at all.
(Whenever I mention this someone pipes up something about some 1960's UNIX thing that did this, which I'm sure is true, but now more than ever I can't understand why it still doesn't work like that - for any user-space, open-source, C/C++-based, Linux-API programs at the very least... which covers about 90% of any Linux distribution. Do it right, and you could have a MinGW layer on Windows that automatically does the same for you there for the same binaries when they are run on that platform).
It's really time - especially with the web abstracting out actual executables, the OS's abstracting out the underlying architecture, and things like cross-platform libraries abstracting out hardware - that we started doing this.
Once we can get away from device specific images (i.e like the IBM PS/2 Compatible solved many many years ago already), then I might look at an ARM device.
Until then, I would just be buying something that will be unusable in ~2 years once no updated images are released.
If I wanted to buy something immediately obsolete, I would just buy Apple products!
"Until then, I would just be buying something that will be unusable in ~2 years once no updated images are released.
If I wanted to buy something immediately obsolete, I would just buy Apple products!"
I'd buy android...
After all the fruity device my son uses for audiobooks etc is now more than 4 years old, and is asking for the latest software version to be installed.
Whereas many android devices are 'fire and forget' from their manufacturers, never seeing any updates.
You'll always need a device specific kernel, or maybe you should steer clear of x86 specific images as well?
"You'll always need a device specific kernel, or maybe you should steer clear of x86 specific images as well?"
Not so. An x86 image will be capable of running on any x86 processor, assuming you can get the firmware to find it and boot to it and the drivers are working. Sure, some x86 systems hide the required things to do that, but that's not many of them and you could simply not buy them. Also, some chips that are old enough won't run a x86-64 image either at all or without a 32-bit bootloader.
That's quite different from the ARM landscape, where nothing will boot at all unless someone has tailored a bootloader for the specific device running it and for the specific image being run. This is why it's hard to flash a custom ROM to an android device* and why companies seeking a system that can't be reimaged are moving to it. For single-purpose devices, that's fine, but for computing, it has many limitations. I would want a device like this only if they had a universal firmware system that could boot anything that was written for the architecture. I'm not asking for universal drivers. I'm not asking for source available on every component. I'm asking only for the right to boot to any image I want; it's my responsibility to make sure it can boot.
*Installing a custom ROM on android devices: You can't have an image for the version of ARM used by the chip, nor by the chip manufacturer and architecture, nor even for the specific chip. Your ROM must be recompiled for each device you want to install it on, with a lot of fiddling with it and it never working. That's why these alternate androids only ever run on a few flagship phones.
For me, this IS the year of the Linux desktop, just as it has been for the last eight years or so. It does all I need and more with Mint 18.3 and FOSS software. CAD, graphics, documents, spreadsheets, photo editing etc..
Yes, I enjoy a tinker as much as anyone, and I could easily run Windows inside my Linux system.
But why on earth would I ever want to???
"I don't I miss how young I felt in 1995."
That depends on how young you were in 1995. It's a relative term.
Possibly, but from my experience '95 to now, if I were now age in 95, I think I'd still miss the energy and the lack of an increasingly crusted and jaded mindset from watching the world and politicians repeat obvious mistakes of the past and call it progress.
I agree that Mint is stable enough for true desktop use. I run a business using Mint 18.3, which also covers the household computing. No Windows, no problems, for me.
Mint could provide a desktop for the masses, and doesn't need any particulay "year" in which to do so. What's required is an end to the business agreements that force every PC buyer to purchase a Windows license. All Windows' domination flows from that: The games, the corporate hegemony, all of it. However, the tech giants do not want to rock the boat, for now, and you can hardly blame them.
Personally, I quite like the current situation. We Linux gurus can enjoy being contrarians while also having full control of our tech experiece. More seriously, it is good to have two systems in competition. The losers are those Windows users who have a poor computing experience. When family and friends ask for PC help, they often talk about problems which simply wouldn't arise in the Linux world.
Don't flame an old penguin. As I say, I'm happy with Windows, I just don't use it personally. And I like Bill Gates and Microsoft generally, for all that they have done.
What's required is an end to the business agreements that force every PC buyer to purchase a Windows license. All Windows' domination flows from that:
Actually, no, it doesn't.
What it flows from is that practically every bit of serious commercial software that businesses need to be productive is written for Windows. While alternatives do exist, they are considerably less capable and productive.
The alternatives being less productive means that you'd require more people to do the same job, which is why people are willing to pay out significant sums of money for software. (ie, a bit of software costing £5k a year is still cheaper than hiring somebody for even £15k a year) If that software happens to run on Windows, then that OS is the one that's required.
For most businesses, the killer app is (still) Exchange+Outlook. Personally, I think that the web version of office 365 may well end up enabling killing windows on the desktop.
"What it flows from is that practically every bit of serious commercial software that businesses need to be productive is written for Windows"
Indeed. And why are so many business apps written for Windows? Because the market for that software is huge. Because there are so many PCs running Windows. Because every PC comes with a forced Windows purchase.
(notwithstanding the rise of SAAS - As stated I run a business on Mint 18.3, the accountancy software is SAAS)
I agree, if you can set up your needs and workflow properly, nothing wrong with going full linux. Set up a call center a few months back with minimal cash. All the pc's were cheap second hand jobs (some unlicensed and all riddled with malware). Mint on them all,voip software and odoo as the back end sales/crm software (on an ubutu linode).
Cheap as chips and runs like a charm, job done.
"While alternatives do exist, they are considerably less capable and productive."
I'm reminded about the story of the bees and aerodynamics. According to aerodynamic theory bees can't fly. Nobody told the bees that so they continue flying quite happily.
Can we have a bee icon please?
"Ansible is way more powerful than group policies and can be applied to any Linux instance. How do you think administrators manage tens of thousands of Linux servers?"
Sure Ansible is powerful. But its also way way more complex to setup, use and administer. Group Policy is built in, and you can do pretty much everything from a single GUI with already there user and group based filtering.
By the way the Windows built in Desired State Configuration and Powershell already give you much of what Ansible does - so much so that Ansible on Windows actually uses those as the underlying toolset.
> How do you think administrators manage tens of thousands of Linux servers?
That is one problem, however, for Linux to be ready for prime time it needs the tools needed to manage tens of thousands of Users and client systems/devices, something MS Group Policy does out-of-the-box with lots of pre-defined (and tested) 'recipes'.
Not saying Windows good - Linux bad, just pointing out that Ansible which uses DSC and Powershell, isn't a full replacement for GP.
"Personally, I think that the web version of office 365 may well end up enabling killing windows on the desktop"
Almost no one uses that. It's like Google Craps - pretty lame. Pretty much every company using O365 installs the full fat version. Also seeing as O365 has had a good few hiccups lately would you want to rely on a web only client?
"Also seeing as O365 has had a good few hiccups lately would you want to rely on a web only client?"
Depends.
Depends on e.g. who runs the underlying hardware and software, and what facilities the end users organisation actually needs, and how important they are.
If the underlying hardware and software is run by a bunch of off-premises el cheapo cowboys, whose business objectives are inherently incompatible with those of their victims/customers, what could possibly go wrong?
If the underlying hardware and software is run by a bunch of on-premises people who understand what provides actual value to the organisation which employs them....
>> What's required is an end to the business agreements that force every PC buyer to purchase a Windows license. All Windows' domination flows from that:
Penguin/Mac user here, but that's just not true. It's prefectly possible to buy a PC without a copy of Windows, I bought 6 for a test rig only last month.
What keeps Windows as the main player in corporate environments is the ability to manage hundreds or thousands of machines centrally, using Acive Directory, Single Sign-On, Group Policy and other tools specifically designed by Microsoft for that purpose. There's no equivalent for MacOS or Linux and that's why IT departments keep using MS products. Yes, they're shite, but they're the only shite that does that stuff.
You know, I quite like Mint. I've even rolled it out to all the machines of one of our customers, but something about the sanctimonious cries of "I've been using linux since nineteen eighty six and it does everything I need much better than windows", that makes me want to switch them all over to Win10.
Oh, Linux on an ARM laptop would be nice. It's the old Asus EEE laptop come full circle, useful for sysadmins, developers, and as a kid's computer. And you can have more colours than bright green.
>Could be "a great way of teaching the kids" about clusters..
Don't know about the Kids, but being able to run up a Parallel Sysplex to play around with, without the expense of having a roomDC full of big iron would at times be useful...
However, I suspect this might be rather slow on a Pi...
I wouldn't blame anyone in particular. Windows grew up on x86 architecture, built up a wealth of software compiled for that architecture and despite efforts by Microsoft to diversify to MIPS, ARM, PPC, Itanium etc., they all flopped.
The floppage is simple - companies have finite resources to develop software and they're not going to waste time and money cross-compiling, testing and supporting an architecture nobody uses.
If Microsoft had any sense they would have built UWP around LLVM to alleviate devs from needing to make this choice. Devs compile to LLVM and the OS does the final step during installation.
Off topic, but one thing I like about Microsoft is the effective way they have changed as conditions demanded it. Change is almost impossible for large organisations, and failure to change quick enough is what kills them off, or at least shrinks them to a nubbin. Facebook can't change. Neither can Google, probably. Not fundementally, not culturally. EDS couldn't, nor could IBM, and even Nokia, once the kings of change, almost died of intransigence. Apple needs a change now, as the smartphone market it founded saturates and cools off. They have done it before, in the 80s, but that was a long time ago.
Microsoft has widened its interests effectively and with great forethought. If its PC dominence collapsed tomorrow, it would still survive as a tech giant. That would have been unthinkable in the 90s. But they have entered new markets and flourished there in a very impressive way. If the Zuck hasn't studied the history of Microsoft, he probably should.
Microsofts model has always been copy the current leader. From CP/M, Wordstar, lotus 123, dbaseII, Lisa..., today its AWS (Azure). If they fail, they try again ( Windows wasn't successful until V3.1).
Sometimes they have to wait for the incumbent to stumble first, sometimes they trip them up.
I'm sure they are currently trying to do Social Media, AI, even blockchain,
>I'm sure they are currently trying to do Social Media, AI, even blockchain,
Well...
Social media: Ms ran So.cl for 5 years and closed it down 2 years back. It seems current efforts are more focused on tools to help with social media aggregation, monitoring and analytics.
AI: Not sure what they actually doing, but there is currently a tv campaign (UK) that is promoting Microsoft AI.
Blockchain: They launched a blockchain development kit late last year.
So looks like someone has learnt not to blindly copy the leader but to provide tools that enterprises will be needing.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
