Android Beats Windows, Now Officially The World’s Most Popular OS

Android Beats Windows, Now Officially The World’s Most Popular OS
I think this is the biggest news for today.

Windows still has a commanding lead in the PC Market though (the platform which holds the majority of the major productivity apps).

Android may now be on more devices, but the vast majority of them are too weak to run something like Blender at a decent pace.

Samsung Galaxy S8?

In March 2017 Android usage hit 37.93% last month across desktop, laptop, tablet and
smartphone, edging out Microsoft Windows (combined versions, including mobile) on 37.91%.

Probably 98% of those people have both a Windows device and an Android device. One is for work, the other is for tapping things.

GNU/Linux sits on 0.75%.

Sad trombone for the the GNU/Linux users there.

Seriously guys, you need to up your game.

Microsoft hasn’t been using the branding with the butterfly in years, dude.

Didn’t bothered to change my avatar for more than ten years. Wow, the time is passing so fast…

If Adobe products and ZBrush were on Linux I think a lot of artists would change their minds about it…

You should have updated yourself a bit:

Anyway this is old too, windows 7 “beta” :spin:

One clarification, Android uses the same Linux kernel as the desktop operating system better known as GNU/Linux (To better differentiate it with just the Linux kernel). But the GNU/Linux OS for desktop is not the same (neither compatible) as Android. This regards people talking about Linux desktop OS on this Android thread, not regarding Tux which is Linux (kernel) mascot.

Number of users has nothing to do with pro users. Android and iOS are just “toy” OSs. Like gaming consoles. If we are talking about servers, banking, developers, sfx and so on then the number would be quite different.

By pro users are that you are working for someone or you have clients.

Android is in my view a data hoarding system for Google. Users are their clients.

Of course there are exceptions.

This is true over quite a few disciplines other than art as well. End consumer support for products on linux tends to be lackluster at best, yes there are some gems out there, but by and large any paid product will have at best limited linux support. And quite frankly there is quite a few paid products out there in use by professionals, from medical imaging to autocads to any of the hundreds of simulation suites or in some cases, and the one thing that nearly all of them have in common is that there is a huge cost for unexpected downtime. I worked for an outfit for a while were every ten minutes of downtime per line would cost the company 200k, and that was just in paid time for idle workers. If you look at the medical or aerospace professions the bottom line can be measured in lives rather than dollars and when we are talking those types of terms…well supported software suites on well supported operating systems starts to look kind of sexy.

The reason I mention that is this, zbrush and autodesk are not alone in the struggles to keep up with all of the linux forks and distros out there. and quite often it is just not cost economical to have a several team dedicated to keeping a suite of softwares running on what? 10, 20, 50? flavors of linux? Granted when you are working in an environment that is saturated with computer savvy people an OS like linux is a godsend for things such as render farms and server applications. But when it comes to end consumer products I’m on the fence between wanting linux support or one more nifty tool to use or a few other bug fixes.

Hi.

That’s not as hard as people think. No more than 5 base distros cover 98% of GNU/Linux users. In addition you can do what Blender or Krita does, sharing binaries with included libraries that run in most modern distros. But I agree that developing for a new platform costs money. Windows on desktop is a monopoly with an extreme dominant position. So, why invest in support of a new platform that only has a handful of users.

Windows is not a monopoly, All people can install what they want.

Speaking about iOS or Android, i will not even qualify hardware / devices where they run in the same market, league. smartphones/tablet are a complete different form factor. They have replace mobile phone, ( or previous smartphones). you have a potential of nearly 1 phone by peoples living on earth, as it was allready the case in the 90s and early 2000 with mobile phones…

who is not having at least a mobile in their pocket right now ? with Android dominate the mobile OS share, and windows mobile who have nearly disappear, it was a question of time for see this .

Dude, you have no idea. First of all, supporting five base distros is already way too much. There should be one thing that you can target, per architecture. A Windows program built today is likely to work on every Windows version from ten years ago up to ten years in the future.

For Linux, you have to multiply those five base distros with 2-3 versions and several flavors, all of which have incompatibilities one way another. Look at the download page for Mitsuba, where the author really has put some effort into supporting multiple Linux distributions. It’s four different packages just for Ubuntu, plus one for Debian 7. Even though Ubuntu is based on Debian, you can’t expect compatibility, because the dependency packages all have different versions. Since the program hasn’t been updated otherwise, the packages are outdated and likely don’t work on current versions, either.

This “best practice” approach is clearly not feasible unless you have a package maintainer do it for you, for every distribution. Good Luck on them keeping up with your work. Of course, you need to be FOSS too.

Linus Torvalds actually has a side-project called Subsurface and they gave up on this approach. Listen to his rant at DebConf, where he complains about distributions breaking compatibility and refers to shipping binaries for Linux a “major fucking pain in the ass”.

In addition you can do what Blender or Krita does, sharing binaries with included libraries that run in most modern distros.

Krita (and Subsurface) use AppImage, which is a kind of crazy (kind of genius) way to go about the problem, using a virtual file system. Blender builds a large static binary that makes certain assumptions on the base system, which is considered “bad practice”.

Either approaches have at least two major problems:

  • The application does not integrate into the OS properly and may have to be run from a terminal
  • There still is no way to tell which distribution the application will actually run on without testing it

AppImage recommends to build on a “sufficiently old” system (the problems of which I will not go into) to maximize compatibility. The integration problem could be solved by distributions, but seeing that they can’t agree on a package format, they probably couldn’t agree on integrating AppImage (and/or any of its competitors) neatly.

As it is, either solution is “unshippable” to non-technical end-users.

Windows on desktop is a monopoly with an extreme dominant position. So, why invest in support of a new platform that only has a handful of users.

If GNU/Linux was anywhere near a double-digit percentage of users, maybe dealing with all this bullshit would be pay off. At a stable less than than 1% over the years, it’s generally not. If Microsoft didn’t have a near-monopoly on the Desktop, a system with less than 1% of users still wouldn’t be on the radar for most vendors. Even Mac OS with its 5-10% of (likely high-income) users are commonly ignored.

Having said that, these “general population” statistics aren’t necessarily valuable. In some niches, even Desktop Linux users may have a decent market share - creative applications isn’t going to be one any time soon.

SideFX and Allegorithmic offer products for GNU/Linux, right?
I would really like to know if that story about hundreds flavors of Linux is a real problem for them.

But if software companies insist on specific distro packages and not portable versions, that could also be simplified in the following way: Software companies could support the main distros with a company behind them, that results in only “3”: Redhat, Suse, Ubuntu (only LTS on Ubuntu).
Then the Fedora and OpenSUSE people would do the right thing for the software to work there as well. Also Debian and Mint people. So basically companies would be giving direct suport to 3, but indirect supporting a big part of the entire universe of GNU/Linux users.

Professional proprietary software (like 3d, with Maya, Mudbox etc) only really support RHEL and Ubuntu, which works fine since these are the Linux distros which companies Disney, Pixar, Dreamworks etc all use.

They use these distros because they offer stellar enterprise support.

That doesn’t mean you can’t use said programs on other distros, but they are unsupported.

For other Linux software in general, typically you get it from your distro repositories, which if course relies on distro mantainers offering said package there, but in practice it’s not really a problem if your distro is reasonably popular, also there are often user repositories filling in the blanks.

Recently there’s been much development for container style package management, appimage is a earlier simpler type of solution, the new solutions are flatpack and snapy, which offers ‘sandboxed’ applications which allows for distro-agnostic deployment by baking the necessary dependencies into these container formats.

Eh what ? If you package the necessary dependencies it will run just as on Windows where software typically ship the specific .dll’s they developed against to make sure it runs correctly on whatever Windows version the user has .

Maya and Mudbox for example are creative Linux applications being used in industry top 3d movie/sfx pipelines, that said when we look outside the 3d/sfx market, there not a lot of ‘creative commercial software’ available.

In my opinion it has nothing to do with package compability though, even a small almost single-guy-company like 3d Coat can ship Linux software, there just has to be a market.

I see a trend of Linux sneaking in to the professional desktops via the server side of things. On the server side Linux is king, last I saw about 95% of all servers run some kind of Linux or Unix. Now I don’t think that the change is happening because you can’t integrate Windows and MacOS into a Linux pipeline. Usually it’s because the big organizations want to bring costs down and having to support one platform instead of 2 or 3 is cheaper. Linux gives ultimate control over the environment and the system, something that you will probably never have with Windows or MacOS and they have zero chance of competing on the server side.

Sounds like part of a discussion I had with a Red Hat developer at a conference, we discussed how Linux had been so successful in the enterprise, and his response was basically two things, first off linux is very performant ‘out of the box’ and can be tailored to be even more performant for specific workloads if you so wish, secondly there is no vendor lock in like with Windows and OSX, instead companies offering enterprise support are fighting to serve you, Red Hat, Canonical, SUSE, even Oracle all offer first grade support while competing in pricing, giving you the best value for your buck.

They offer builds for specific versions of specific distributions, not your “98%”.

I would really like to know if that story about hundreds flavors of Linux is a real problem for them.

It isn’t, you can use one of their supported systems or you can fuck off. It’s a problem for the users (or the naive developer that doesn’t know which users need to fuck off).

Of course, these applications are also for reasonably technical people, not your average mass-market consumer.

But if software companies insist on specific distro packages and not portable versions, that could also be simplified in the following way: Software companies could support the main distros with a company behind them, that results in only “3”: Redhat, Suse, Ubuntu (only LTS on Ubuntu).

That’s how it is now, except SuSE users often can also fuck off.

Then the Fedora and OpenSUSE people would do the right thing for the software to work there as well.

They could do that right now. They don’t. They don’t see it as their responsibility.

You may be able to get these programs to work by messing around with your system. That’s not an acceptable solution.

For other Linux software in general, typically you get it from your distro repositories, which if course relies on distro mantainers offering said package there, but in practice it’s not really a problem if your distro is reasonably popular, also there are often user repositories filling in the blanks.

That’s the problem. You don’t see it as a problem. You’ll stay at <1% market penetration forever. That’s fair enough, but then you can also stop complaining about not getting some piece of software ported over.

If you package the necessary dependencies it will run just as on Windows where software typically ship the specific .dll’s they developed against to make sure it runs correctly on whatever Windows version the user has .

The shared libraries are not the only thing that are different across distributions. You also still have to test to see if you got everything right, this is true even on Windows where there are subtle differences between versions as well. Luckily there aren’t literally dozens of different versions of Windows in common use.

Again, you need to put all this into perspective with the tiny market share. Even small hindrances can make the effort not worthwhile anymore.

Recently there’s been much development for container style package management, appimage is a earlier simpler type of solution, the new solutions are flatpack and snapy, which offers…

Gah! I don’t care which it is, pick one and stick with it.

This is why (in principle) I strongly prefer systems like FreeBSD, because those are stable systems made by adults for adults. Linux people are never done fucking around with things and I just don’t have the patience (and neither do “normal” people).

When it comes to the hypothetical takeover of Linux in the PC Market as well, their developers may as well have ample time to iron out the remaining ease of use issues (because nowadays, even those who consider themselves Windows fans are starting to become pessimistic about the future of their OS).

Every slip that Microsoft does is giving Linux another chance to siphon off the userbase (as if they have not had enough already). If they can get a standardized app packaging approach together (so you can download and use apps. the same way as in Windows), Linus and friends may have a real chance.

I think you just don’t understand them and technology in todays time.