Fuchsia overview

What this is really about is Qualcomm. They have inadequate competition and that allows them to abuse everybody else.

In a competitive market, open source drivers win. Look at the desktop market -- nearly everything is open source, and the biggest holdout is nVidia, because for a while there nobody was challenging them on performance. Now that AMD has competitive GPUs again, not only does that provide a competitive option with open source drivers, nVidia is now losing market share to them.

OEMs don't really like proprietary drivers. They're a pain. Whenever new driver versions come out, it becomes the OEM's problem to test and distribute them instead of having the Linux kernel team deal with that hassle. The open source drivers also generally have fewer bugs (because anybody can fix them), which leads to fewer support calls and more satisfied customers who make repeat purchases.

And customers naturally prefer hardware with longer hardware support, when it's available.

What we need is more competition for Qualcomm. Google could do that if they wanted to -- throw some money at making competing phone chips. Apple does it, why can't they? And then publish open source drivers and sell the chips to anybody.

We also have AMD as a dark horse now. The Ryzen 4000 series goes down to 10 watts -- it's so power efficient you could squeeze it into an Android tablet as-is (and then be a lot faster than most tablets), and it wouldn't take much on AMD's part to do something that could work in a phone. Android is an open source system with apps that run on the JVM, so the architectural difference shouldn't matter much.

Or we could get the antitrust authorities to stop letting Qualcomm buy every competitor that springs up, which they've been doing for years.

But you solve the competition problem and you get open source drivers. Along with all the other benefits of real competition.

I can't help but think that if microkernel architectures were such a good idea, that would explain why every popular OS is a microkernel architecture (oh wait...). Snark aside, this is a serious debate since the inception of Linux [1].

It's important to look at Fuchsia through the lens of what problems it solves in Android/Linux because that tells you a lot.

As we know, receiving updates in Android is, well, a clusterfark. There are two key problems:

1. Phone manufacturers need to write or update drivers essentially with each Android release; and

2. Those drivers are by nature of Linux being a monolithic kernel, written in kernel space. Kernel space code by third-parties is going to be inherently more unstable to your device than if they were written in user space.

So when Google talks about Fuchsia having a stable binary ABI for drivers, they're talking about solving both of these problems (at least as they see them).

So the question I've always had about the justifications for pouring billions of dollars into Fuchsia (I mean that quite literally) is: could this really not have been solved in Linux, even if it's just the limited context of Android?

I'm not a driver or Linux expert by any stretch of the imagination. I'm sure there are people here who are so this is my question to you: could you have created a system for ideally user space drivers with a stable ABI in Linux and avoided all the costs of completely reinventing that wheel?

My other question about Fuchsia is: does Google foresee themselves adopting the same vertically integrated solution that, say, Apple does? It's worth noting that Apple by not providing iOS to third parties is somewhat paradoxically immune to antitrust investigations (I mean really... how does that work?).

Evidence against Google adopting the Apple model is the fact that Fuchsia is open source, which is a curious decision. This seems to imply that Google either wants to maintain an Android-like ecosystem or they simply see it as their only viable option.

I don't know how anyone can expect that Samsung is going to move to Fuchsia. Samsung already chafes under the yoke of their Google dependence with Android. They've tried to get out from under it (ie Tizen) but luckily for Google, Samsung is terrible at software (at anyone with Samsung crapware on their Galaxy can attest; Bixby anyone?).

Maybe Google thinks the insurance of Fuchsia being open source is necessary for third-party adoption to have any chance but still, the only way I see Samsung going along with this is that there's absolutely no other alternative.

[1]: https://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds_deb...

You're talking out your ass, Google.

Everyone knows how much Google cares about open source and open platforms...

> The open source community is not, for example, going to make their own high performance gpus.

Of course it will.

3 year old hardware is still very new by Linux standards. Especially wrt. hardware classes for which a unified support framework has yet to be provided, including RGB LED's and fingerprint scanners. The sleep performance thing is an interesting example, since the whole issue was caused by hardware OEM's deprecating the old ACPI "sleep" state and deciding that OS's should manage standby states entirely on their own - this too will require a lot of really hard work to be supported properly across zillions of devices.