[DMcCunney]

Originally Posted by eschwartz

Well, Ubuntu does have Convergence planned with Unity, possibly in 14.10 when they can finally start shipping Ubuntu phones...

I can only wish them luck. Why, precisely, should anyone want to buy a Ubuntu phone?

These days, smartphones are as much fashion accessory as anything else, and bought because they are perceived as cool. Apple's iPhone is cool. Android is cool. MS has the uphill battle of making Windows Phone cool.

There will be a core of users for whom the fact that it runs Ubuntu will be a selling point, but I don't thank that core is anywhere near large enough to make the project successful.

Isn't the important thing having one architecture to build the code for? After that, you can just install one base system and depending on whether you install for a phone/tablet or a PC, the appropriate packages get installed to setup the right environment?

The important thing is to have code that builds on more than one architecture.

One of the things that has helped Linux is the fact that it does. On the desktop and on servers, it runs on the X86 architecture. On tablets and in smartphones, it runs on ARM. My old Linksys WRT54G router had a Linux 2.6 kernel, and ran on a MIPS architecture. Linux has also been ported to SPARC, HP Precision Architecture and other things.

Doing that is not simple. One of the things in the Mozilla Developer's documentation was an amusing set of guidelines for writing portable code, since the expectation was that Mozilla code would be brought up on multiple architectures. Part of the issue was understanding what was portable, and the other was understanding that not all compilers were created equal. Mozilla code is written in C++, and the code needed to compile under any of the various C++ compilers that might be used (with the HP-UX C++ compiler a particular problem child.) There were an assortment of admonitions of the form "Just because it works in Microsoft Visual C++ on Intel, don't assume it will work elsewhere."

For that matter, the chief architect of Windows NT, David Cutler, came over from the late Digital Equipment Corporation where he led development of DEC's VMS OS for their VAX line. Dave was insistent that NT be portable code, capable of being brought up on architecture other than X86. That foresight made it possible for MS to bring up Windows 8 on ARM in the RT flavor for tablets. (And ARM is beginning to gain traction in the server market,, because of lower power consumption. Power and cooling are major data center concerns as servers proliferate, and anything that reduces power consumption and heat generation will be looked upon with favor. The fact that NT can be brought up on ARM might be quite important to MS down the road.)

Of course, those comments apply to the OS kernel. Applications are another matter. But even there there are large possibilities. Development is increasingly done in cross-platform scripting languages like Java and Python, where the binary code is the same regardless of architecture, and current hardware is fast enough that you often don't need to compile to native code to get adequate performance.

So far, all attempts to provide one desktop that scales for both have been a miserable failure.

I think the big problem with Metro was the absense of the Start Menu. Successive versions of Windows have changed the UI in various ways, but there were still common things like the Start Menu to allow users to find their way through the changes.

What I've seen of the Metro interface doesn't put me off that badly, but having no tie to the prior UI that users can use to find things and get stuff done while they learn the new UI is deadly. MS seems to have figured that out, if rumors that the Start Menu will return in Win 9 are accurate.

I don't expect that Android on the desktop will so much as voice an opinion on UI, though it will probably come together with Yet Another Failed One-Size-Fits-All Desktop.

Looking at the variety of launchers available for my Android tablet, I concur. It's as much fun as window managers for Linux. But something needs to establish itself as the baseline and default offering, even if you can switch it out for something you like better. We'll see whether Google's UI designers can do a better job that other folks have so far been able to.
______
Dennis