Slashdot Mirror
Ford Ditches Microsoft Partnership On Sync, Goes With QNX
Freshly Exhumed writes: Ford's in-car infotainment system known as Sync will soon evolve to add a capacitive touch screen, better integration with smartphone apps and, eventually, support for Android Auto and Apple CarPlay in version 3, thanks to a switch of operating systems. After years of teaming with Microsoft, the automobile giant has switched to BlackBerry's QNX, a real time operating system renowned for stability.
It's a car, not a fucking tablet.
Fusion owner here with the "my ford touch" sync system + touch sensitive climate/media controls on the console. Having to take your eyes off the road to make sure you're touching the right 1x1 inch area on the screen, or small indentation seems silly.
Every car I've had prior had physical buttons for these things that after about a week of owning the car could be operated completely by touch alone.
Only the most deluded Microsofties crave Microsoft's dog food ecosystem. Most of the world doesn't care, and only tolerates Microsoft products as a necessary evil in certain situations.
Ditto. I don't want a car media interface. I want dedicated physical buttons & knobs for the climate, radio volume & input selection.
And a place on the dash to mount my own phone/tablet with a nearby USB plug.
I have never seen an electronic car interface that was any good at all, and that includes every navigation system I've ever seen. My phone has better navigation (Waze rocks), better audio, and a better interface than anything a can manufacturer could ever try to copy.
I only upgrade my car every 10+ years - an even then it might not be a new car. Hey Detroit - stop trying. Give up. Let Apple/Android/[new startup] give me the tech I want. If you want to get fancy, give the phone a read-only API to the car's status.
You're an idiot, aren't you?
I remember back in 1995/1996 or so ... a 1.44MB floppy with a bootable image of QNX. It booted onto pretty much any machine we could find, identified all of the devices, found the ethernet, and had a web browser.
It was faster and more robust than Windows 95 was by a bloody long shot.
Blackberry bought QNX because it has had a reputation as being pretty bomb proof for a long time.
Lost at C:>. Found at C.
As a real Sync user (from 2012), my experience has been that its problems have more to do with user interface than "stability". Even if QNX improves on the latter, it does nothing for the former.
Well, it might help indirectly. Every hour the developers don't spend trying to debug the OS is an hour they can instead spend on making the user interface work better. I suspect that a lot of mediocre products appear simply because there were so many showstopping bugs to chase down that there was never any time to smooth out the rough edges.
I don't care if it's 90,000 hectares. That lake was not my doing.
Microsoft actually had very little to do with the MyTouch system which was the second generation Sync system. Microsoft helped make the original Sync system which for 2007 was actually quite reasonable. For the second generation released in 2010 that was the one that Ford essentially had to rewrite internally to fix, that was done by BSquare -- admittedly that company was made of former Microsoft people and ran on Microsoft's Auto Platform.
The point is however blaming Microsoft for the interface they didn't have anything to do with is like blaming kernel maintainers for Gnome doing something stupid in their interface.
QNX is a great choice of an OS for a fairly fixed ecosystem of hardware where reliability is paramount, but just because the OS is good doesn't mean the interface will be.
Oh come on, Windows 95? The OS that couldn't even sit idle without eventually crashing? That's a real low bar.
I've heard from people who work with QNX that it has plenty of bugs. It may be secure, but it's actually not that stable.
It makes sense that QNX is overhyped and not near as good as some claim. Being proprietary and small, they simply do not have the resources to polish it and keep it polished. Linux has many huge companies paying for hundreds of talented developers to work on every part. In many cases, the best algorithms for many of the problems an OS faces, such as task scheduling, storage management, and networking, are complicated and difficult to implement well. It's no accident that there are more than a dozen good file systems for Linux, each with their points. Windows is still plodding along with NTFS and FAT. And QNX? They simply cannot keep up, even if they rip good code straight from Linux. They're going to skimp on features and choices, and what they must have will be the most dead simple method that delivers adequate performance, and spin that as a virtue because the code is smaller and therefore easier to audit and prove correct. If they discover that their design imposes a fundamental limitation, they live with it, while the Linux world can think of going for a redesign, because the resources are there. QNX could never think of doing a massive reworking of the system like the replacement of X with Wayland or Mir, or the development of btrfs.
Intellectual Property is a monopolistic, selfish, and defective concept. It is "tyranny over the mind of man"
Yes, QNX has been around for a long time.
\What most people don't get is what a realtime OS is, and why it matters. Other multitasking OSes are generally "best-effort" OSes, but in a realtime OS, the whole scheduling system is based on giving guarantees, making sure that things happen within a certain time frame or a certain order.
The overhead is huge, which is why you don't se RT on any normal desktops or servers, but in something like a car, airplane or hospital device, you would rather know that 100% of the requests get served in 100 ms, than having an average time of 10 ms, but a worst case time of 1000+ ms.
If you know the worst case, you can program your systems to operate within them.
Linux does have a RT version, in part supported by Ingo Molnar and Theodore Ts'o, but it does not see heavy use. In part, this has been because for a realtime OS to be successful, all the parts have to play ball, not just some. And in part it is because a realtime OS is quite a bit slower on average, and most regular users would rather have improved average speeds than improved worst-case.
But for a car? Give me a realtime OS any day. I don't want traction control to cut in a tenth of a second too late because the kernel was busy doing garbage collection, time synchronization, and handling an urgent warning that the oil temperature was too high.
AC, for obvious reasons.
I'm a former contractor for the last iteration of Sync. The reason why it felt like it was always in beta is because it always WAS in beta.
That's why when you go in for a tune up or an oil change they take a USB drive and flash new firmware into your car. It's a Microsoft product. OF COURSE it's going to be patches forever. Hell, it's worse than that even. It's Microsoft CE 6 Automotive pack running Adobe flash for the UI. Betcha you didn't know that! The whole design is - to my way of thinking - completely bonkers. I worked support for a bunch of web guys designing the UI. They couldn't read C/C++ and had to have me read it for them. "I need a function that returns the current time adjusted by time zone." And I'd have to go plowing through the source and nearly nonexistent documentation to find it for someone.
You'll note that if you take your Toyota in for an oil change they just change the oil, not give you the latest patches to their unfinished infotainment device in your dashboard every time. You can buy map updates if you like because the DVD doesn't know about new construction (obviously) but that's about it for Toyota. Know why? Because the thing is actually finished. That's how it's done.
Good lord, you should have seen the results from the focus team when we showed them a pre-delivery beta. It had about a 10% approval rating. People HATED it. Nothing worked.
Sync wasn't finished on the delivery date, it's not finished now, it's never gonna be finished.