[lnanek2]

Working at a company that ships an embedded Android product, I know Android sometimes costs us far more than it helps us. It took us months to figure out how to route some audio from incoming A2DP Sink Bluetooth profile to the speakers, for example. Fighting with patches, and Android app Java code, and Bluetooth app Java code, and JNI glue, and CPP Android services, and C Bluedroid stack. Meanwhile anyone on the team could have gotten it working in a day on Linux running BlueZ.

Then we constantly have the same issues working with the display. We output ARGB for an augmented reality product, but Android is not designed to output anything but RGB to the final LCD. So similarly, if we were just dealing with a Linux frame buffer, we'd have had a much easier time doing the customization than fighting all of Android's various SurfaceFlinger and other layers between the app and the display.

[pjmlp]

You experience is a good example why I say that Android isn't Linux, regardless how the FOSS community sees it.

Maybe for your type of work it looks different, but from app developer perspective, they could replace the kernel with something totally different, keep the constrained set of official NDK libraries and no one would notice the kernel wasn't Linux any longer.

[fla]

Even the NDK API isn't reliable. For example, some Samsung devices don't implement the complex number functions from libm.

And there is no way to know this in advance (beside checking if symbols are defined on a specific device)

[pavlov]

Try Tizen? It's a real Linux, and much more than its popular reputation as Samsung's also-ran Android clone.

[biktor_gj]

Well Tizen's openness is a lie too, at least in wearables. Both Gear S & Gear 2 are already depreciated by Samsung, and the wearable image you can build yourself from source is missing most key components you would use on a watch (for example, you won't have S Health is closed source, Nokia maps -included in Gear S- is closed source too), and even the sensor framework is unusable in an opensource form, since half of the apis are closed and undocumented (you can check all this by taking a look at the srpms from build.tizen.org)

Even if you could overcome that, you would still need to make your own companion app for the phone and framework, since all that is Samsung proprietary too.

That's exactly why I'm the sole mantainer of Gear 2 & Gear S Android Wear ports...

IMHO it would be easier to clean up AOSP and make an open variant of Android Wear, rather than reimplement it all from scratch, but who doesn't like a challenge! At least they can use some parts of Nemo/Sailfish OS

[gue5t]

Tizen's openness is also a lie for phones. Just like AOSP, community involvement is not welcome in the Tizen development process and doesn't have any role in deciding what ships on phones.

[darklajid]

Are you advocating an OS from the guys that are worse than Apple, closed as much as possible? The "CM cannot support us", "If you flash anything, we blow a fuse", "Crapware is good for you" Samsung?

Granted, I haven't tried Tizen. But I AM using a Samsung device at the moment and let me state it as clear as possible: Samsung is the worst, the absolute worst, phone manufacturer in my frequently changing usage. I wouldn't trust Samsung with anything.

Incidentally I've been at a Samsung HQ (not phone related, printers) a while back and the manager in charge stated something like "We're great at hardware, we suck at support". I had a hard time keeping a straight face.

Honestly: There's no phone company that is worse than Samsung. I am not convinced that Tizen can fix that.

(I have a S6E right now and I constantly want to throw it against various obstacles)

[pavlov]

The poster I was replying to was working on an embedded device, and they would like a system that's similar to Android but more like regular Linux. Tizen is that system.

Samsung's policies on their devices have nothing to do with how another company might deploy Tizen.

[darklajid]

I didn't mean to attack you or your position.

I'd argue that Maemo/Meego were that system, but I already admitted to know really not a lot about Tizen.

You might be perfectly correct. I wouldn't want to touch anything (Software. Hardware is a different story) made by Samsung with a 10 foot pole and cannot imagine that Samsung would be great as upstream, however you plan to deploy the software.

Again, apologies. My beef is with every piece of Software Samsung published around me, not with you, your opinion or Tizen. I'm extrapolating and generalizing and obviously have a gripe here..

[pavlov]

No offence taken, I can certainly understand your position.

[pjmlp]

Popular by who? Not for app developers for sure.

Are you aware that since Samsung got their hands on Meego, Tizen already went through three SDKs and userspace re-rewrites?

- First it was just Meego Qt SDK renamed for Tizen

- Samsung then replced it with the SDK from their BADA OS, a ugly Symbian influenced variant, including the same Symbian C++ patterns.

- Now it is based on the Enlightenment Foundation Libraries and C, with a C++ wrapper library

But this is only for the hardware where developers are allowed to write native apps, as what Samsung is usually pushing for is their Web app approach.

For example for TVs you need a special contract to deploy native apps.

[pavlov]

"Popular reputation" means "how it's commonly perceived".

I'm not saying Tizen is popular; on the contrary, I was trying to say that it has a reputation as a pointless also-ran Android clone. But past that prejudice, it's actually a competent Linux platform for embedded systems.

[ashitlerferad]

Why did you go with Android when it seems like a GNU/Linux distro would have been better?