UNIX

That said, the only reason I do not switch my headless systems over

to BSD is that I simply lack any experience on that field and I do
not have the motivation to acquire the needed skills atm.

I've been a Linux user, sometime developer, and very-occasional kernel contributor for a pretty long time. And I can say without sense of embarassment that the only reason I don't switch my headed systems from Windows to Linux (on poorly-supported chipsets) or from Linux to *BSD (on supported chipsets) is that I lack the experience in those fields.

There was a point in time that I had to switch from Windows to Linux, even on a desktop development workstation, out of sheer necessity: the Windows solution of the time was performing an order of magnitude slower than what it should have been. That point in time has come and gone.


Haven't run Linux on a desktop or laptop, other than in a Virtualbox, in a hell of a long time.

The growth of mobile has brought us to a new level of proprietary hell, where Linux is less attractive.

Newest revision of Intel Atom uses a PowerVR GPU with no Linux support. Much of the ARM world uses the same PowerVR chipset, but at least closed source drivers are available, of questionable quality on some distros.

It should be emphasized that on most low-power or embedded SoCs, including without limitation the PowerVR IP core, the driver support is limited to OpenGL ES. X servers exist but full GLX is never likely to appear. So basically, good luck running Unity 3D if that's your bag, baby. It would have been appealing to run the new low-power hardware as a desktop machine. But it seems like this is not meant to be.

Unless the partial birth abortion called Windows 8 is highly successful, it's unlikely that Atom will have much to do with mobile and will be relegated to the niche of low end laptops. ARM may be one of the *very* few companies that actually manages to fairly license its technology to multiple manufacturers without beating anyone up, and the embedded market is thriving for it.

So if you run with the assumption that the conventional desktop is dying (after all -- Netcraft confirms it) we've basically got three contenders: Linux (in the form of Android), iOS (look for convergence from Apple real soon), and 'doze-8 (which in this market is an also-ran).

Android is probably the best thing that ever happened to end-user Linux.
Google managed to strike just the right licensing balance to keep the platform open source but also prevent it from fragmenting. It can still fragment if they make life painful enough for the other players, but they know this so they won't.

Best of all, the amount of FSF code in Android is so minimal that even the most die-hard free software communist wouldn't be able to call it "GNU/Android" with a straight face. And the fact that Linus Torvalds isn't being an ass like RMS and insisting that they call it "Android/Linux" punctuates that even further. (Torvalds is an ass for other reasons but not this one.)

I for one would like to see a full Android runtime made available inside traditional Linux, propelling both forward at an even greater speed. For that matter, I'd like to see it made available for Mac and Windoze as well.
Boy would that twist the knife.

Anyway, the idea of traditional Linux/Unix becoming non-viable because of complex interdependencies is overstated. Miguel de Idiot was saying these same things TWELVE YEARS AGO (see "Let's Make Unix Not Suck") and in those twelve years the system has gotten better and adoption has gotten wider. The fact that the obstacles to majority adoption which have been put in its way by Micro$oft have not yet been surmountable is not Linux's fault, nor will it stop Linux from continuing to improve.

When a system becomes sufficiently complex that the complexity becomes a liability, that liability becomes a problem to solve. After 20+ years of problems of this class being solved instead of accepted as failure, it's quite a bold risk to assert that the failure route will be taken *this* time.

(ARM itself is a fascinating story -- its remarkably low power consumption was actually an *accident*. Read http://tinyurl.com/83hhe29 for more on that.)

Nov 25 2012 10:20pm from IGnatius T Foobar @uncnsrd
Unless the partial birth abortion called Windows 8 is highly
successful, it's unlikely that Atom will have much to do with mobile
and will be relegated to the niche of low end laptops. ARM may be one

Atom is an important part of Intel's ecosystem going forward. What else do they have that's sub-10-watt and runs fanless? The newest generation Atom would have been a great choice for a Linux-based embedded processor running digital signage if they had put an open GPU on it. As it stands there is no Linux graphics support on that chip, you just have a frame buffer driver. If you want something open, you go back to the Atom D525, possibly with an Nvidia chip tacked on if can work with their closed drivers.

I do think Atom will continue showing up in phones here and there, not in most of them though. It's important for the industry to have a competitor to ARM, of some kind.

(ARM itself is a fascinating story -- its remarkably low power
consumption was actually an *accident*. Read

Kind of amazing what you can do if you have an attitude that you could build something simple with a small design team. While IBM wasted time and money running endless preliminary simulations of their architecture, Wilson just rolled up her sleeves and built something "good enough." They had a low-power design target and luckily exceeded it by a factor of 10.

Might have had something to do with their 6502 experience. 6502 was RISC with variable length instructions.

Running an x86 android device is an interesting experience. Most apps are built for Davlik and run just fine, but when you get into performance-intensive apps (read: Angry Birds) you find that they're built as native code and either are "not available for your device" when you try to download them from the repo, or they simply don't work when you go to run them.

Anyway, the re-convergence of the Android code into the mainline Linux kernel is a great bit of progress. I hope that someone figures out a way to get an Android userland and a traditional Linux userland running on the same screen at the same time.

way to get an Android userland and a traditional Linux userland running

on the same screen at the same time.

Won't even happen on the typical Android hardware, unless OpenGL goes beyond ES

It's really amazing that they not only created a low power chip by accident, but also managed to keep it that way. Intel is going to be a non-starter in the market for low end access devices, similar to the way players other than Intel have been non-starters in other places.

Eventually they're going to have to buy an ARM license.

The more time I spend with my Transformer the more I'm convinced that the landscape is finally starting to change.