The libhybris situation is a pretty miserable compromise. You land on an old, busted Android kernel, with a userland that mostly works. It exists only because the major SoC vendors have no motivation to support a chip after they've shipped a BSP that successfully builds and makes the hardware occasionally provide the desired features.

(Apologies for being a bit harsh; I've wasted a ton of time recently hacking on Allwinner and Amlogic chips)

The libhybris situation for the time being is "good enough".

Sony last year announced their intention to "mainline" their Xperia devices (their main phone brand)[1]. This could eventually mean running Android on the latest mainline kernel, and perhaps even GNU/Linux[2].

[1] https://developer.sonymobile.com/2016/02/17/work-on-the-main...

[2] Jolla (developers of SailfishOS, a real-Linux-on-mobile OS) announced in February that they are partnering with Sony to bring SailfishOS to the Xperia X (https://jolla.com/wp-content/uploads/2017/02/Sony_Jolla_pres...), which should hopefully be out this month. It's taken a long time, but real-Linux-on-the-mobile is getting closer.

Credit to Sony for this work. They seem to have almost exactly the right emphasis for me - good cameras, expandable storage, and open devices.

what do you mean by real-Linux-on-mobile OS?

isn't​ Android a mobile os running linux?

I hope you meant real-GNU /Linux-on-mobileOS

note: "Android is running Linux" is certified by Linus Torvalds

I think you're nitpicking, and your comment does nothing to add to the discussion, but I'll say this:

If I'd said "Linux-on-the-mobile", you could just as easily have said "Linux (kernel) is already on the mobile!".

If I'd said "GNU/Linux-on-the-mobile", you could have argued that Termux[1] adds GNU tools to your Android phone.

Half the things you'd expect on a "Linux" system (e.g. systemd, a Wayland compositor, X, DBUS, PulseAudio) are not associated with FSF/GNU at all.

Without nitpicking, when you say "Linux" most people think "Red Hat" (or Debian, Ubuntu, Arch, etc), not "Android".

[1] https://termux.com/

Linus has a vested interest in taking credit for the most prolific mobile platform. I don't think there is any common sense in the 'Android is Linux' argument beyond that though.

As a software platform, it isn't even remotely similar.

Android is an operating system built on top of the Linux kernel project. "Android is not Linux but ${DISTRIBUTION} is" is just stupid, operating systems simply cannot be compared to a kernel.

Not quite right, NDK applications are constrained in what they can touch and many syscalls have been removed.

What does Linus have to do with it?

just to mention what I am saying is not just my opinion.

That's not how it works though.

Linux is a kernel, and Linus has all the right in the world to make that claim. It doesn't in any way strengthen your argument though because of the different userland. You are comparing apples (operating systems) with oranges (kernels).

sorry that I created confusion.

I tried to tell that his "real-Linux" (kernel) already exists (I don't​ know if he meant (in another comment, he said that he meant userland) userland e.g gnu, freedesktop.org etc by "real-Linux")

Jolla/Sailfish had (and quite frankly, continues to have) many problems, making it unsuitable as a phone, or for mobile development.

* Jolla has always used (and continues to use) outdated hardware, Ubuntu Phone OS shipped with or ran on reasonably performant hardware. Only after several years is Jolla now discussing the possibility of porting the OS to a somewhat modern device (Sony Xperia)

* The Sailfish SDK relies on VirtualBox for the emulator and for native compilation, a drain on resources and memory, and an indication that the development team was not well versed in creating cross compilation ecosystems or tools. Ubuntu Phone SDK was limited to Ubuntu Linux, but ran natively (Sailfish SDK required VirtualBox even on Linux)

* For years, Jolla withheld from or misled the public on the state of internal affairs (Canceled tablet, no refunds for crowdfunders, CEO and key employees quitting, lack of funding, relying on volunteers for key areas of development)

* Sailfish OS has always lacked basic security features for applications (QT Quick embedded in plain text) or for the phone itself (encryption, permissions). Sailfish's security problems may be just as problematic as Tizen's

* Sailfish OS was never true open source, even less so than Android, although it has always been advertised as such.

* Very poor/non-existent developer relations - A few developers attempted to make games for Jolla, but encountered serious problems that the developers weren't interested in fixing (i.e. SDL2 on Jolla not supporting landscape mode, preventing games from being submitted), even worse, the developer for the SDL2 Wayland port flat-out refused to fix the problem or provide the source code for other people to fix the issue. Such problems or resistance from the developers were never encountered on Ubuntu Phone OS.

> Jolla has always used (and continues to use) outdated hardware

But Jolla has shipped. Jolla 1, Intex Aquafish, Turing Phone, Inoi R7[1]. Limited runs of Jolla C, Tablet. Sony Xperia X should be coming soon. This is a startup of (these days) less than 50 people, cut them some slack, they're competing with giants.

> The Sailfish SDK relies on VirtualBox for the emulator and for native compilation

It's infinitely more efficient to build and ship 1 VirtualBox OS Bundle, v.s. building and maintaining at least 3 different cross-compilation toolchains (one for each of: Windows, MacOSX, Linux). With limited developer resources, I'd choose the latter route too.

I also prefer this setup to things like scratchbox2 and Android's (or iPhone's) emulator. It's portable between base OS's and it makes use of existing and well-known tools (VirtualBox).

> For years, Jolla withheld from or misled the public on the state of internal affairs

The Tablet fiasco could have been communicated better. At the same time, that's the risk you take with Kickstarter/IndieGoGo. It's a startup.

I don't agree with any of your other points, they're a startup, it's not all smooth sailing.

> Sailfish OS always lacked basic security features for applications (QT Quick embedded in plain text) or for the phone itself (encryption).

In regards to packaging, or app isolation, Jolla's wisely chosen to work on "mobile problems" and leave package management to the heavyweights (e.g. RH with Flatpak).

VPN support was added in the last release. The public git repos show ongoing work to filesystem encryption.

> Sailfish OS was never true open source, even less so than Android

Significant pieces of SailfishOS are proprietary. However, I think it's a more "inclusive" OS than Google's Android. Google develops Android. SailfishOS is GNU/Linux, it's developed by Red Hat (e.g. Linux kernel, systemd), Intel (Connman, ofono), Qualcomm (BlueZ), Qt (Qt Company), Mozilla (Firefox), GNU (CLI tools), KDE (Calligra), Collabora (gstreamer), etc.

I'm hopeful that one day SailfishOS will be fully open source.

> Very poor/non-existent developer relations

Disagree. They hold fortnightly meetings (#mer-meeting on Freenode). Most of the code is public (https://git.merproject.org/). There's a public Bugzilla (https://bugs.merproject.org/). There's discussion boards (https://together.jolla.com/). Community translation portal (http://translate.sailfishos.org/).

[1] https://together.jolla.com/question/136143/wiki-available-de...

> But Jolla has shipped. Jolla 1, Intex Aquafish, Turing Phone, Inoi R7[1]. Limited runs of Jolla C, Tablet. Sony Xperia X should be coming soon

Proving my point - all of these devices (except for Experia, and possibly the Tablet) ran on hardware outdated at release time. Ubuntu Phones ran on reasonably performant hardware (Edge, Meizu MX5 and Google Nexus 5) at release time.

Your personal preference for VirtualBox, or VirtualBox being "well known", is irrelevant. Virtualization is unnecessary and a waste of resources compared to cross compilation and emulation via QEMU.

> At the same time, that's the risk you take with Kickstarter/IndieGoGo. It's a startup.

The tablet problem was more than a matter of communication. It demonstrated that Jolla wasn't ready for prime-time the way they advertised. Backers will empathize with a product not ready for prime-time if the people involved are transparent. Whether you like it or not, unaccounted product delays, CEO's quitting without explanation, staff being laid off, and other internal matters do erode consumer confidence and burns bridges.

> VPN support was added in the last release. The public git repos show ongoing work to filesystem encryption.

Perhaps Russian government involvement is finally starting to pay off? They certainly couldn't have been bothered to add these things towards the beginning.

> I'm hopeful that one day SailfishOS will be fully open source.

Sure, with less than 0.1% of the market space, what do they have to lose. They've certainly lost a lot of other things though.

> Disagree. They hold fortnightly meetings

That wasn't always the case, certainly not the case in 2014/2015. Sailfish's lead SDK developer actually spent more time promoting his personal CMake alternative than his work on the Sailfish SDK.

Ubuntu Phone OS developers were always responsive to SDK issues, even when they didn't have the answer.

Error by me in the last sentence, I meant to say "former", not "latter".

> It's infinitely more efficient to build and ship 1 VirtualBox OS Bundle, v.s. building and maintaining at least 3 different cross-compilation toolchains (one for each of: Windows, MacOSX, Linux). With limited developer resources, I'd choose the latter route too.

Developer driven development (DDD). Often leads to bad market fit if the target market is not developers.

Would have dropped Android for something like Tails or OpenBSD for phones.

Hmm, I'd heard that the point of Mir was mostly that it would make using existing Android drivers easier than Wayland would. I guess that wasn't true.

I don't think any consumer outside of the linux crowd and a few neckbeards cares whether their phone is using Mir or other technological choices

Most likely not. However, I think Canonical should have used Wayland instead[1], and invested their time/effort/money elsewhere.

[1] granted Mir may have kicked Wayland development into gear