Sigh, let me place this here, yet again...
Interesting exchanges I saw with a MeR dev several+ wks ago:
__________________________________________________ ________
xxxx: finishing touches on wayland on top of android drivers, for future non-android systems enjoyments
xxxx: but also at the same time, it's obvious the whole jolla movement thing has started a wave that may make the gadget choice quite interesting
xxxx: non-android is suddenly feasible again thanks to our libhybris
xxxx: as we can suddenly put our stacks everywhere
__________________________________________________ ________
xxxx: i can give you the answer here if that's OK: the SDK only contains X86 binaries
xxxx: so you're only seeing it on stuff like exopc and acer iconia tab
xxxx: did you see the iconia tab video?

yyyy: why no option for arm binaries or virtualisation, is that because vendors simply wont provide them for sailfish, hence the need for libhybris?

xxxx: well, there's virtualization but it's x86 virtualisation – the target of SDK is being an SDK
xxxx: and at this point we're not providing ARM binaries until there's a target that people can run ARM builds on
xxxx: the SDK isn't terribly good as a basis for making device ports with anyway

yyyy: So why no ARM binaries? And why only x86 virtualisation? (the later does not matter so much I imagine)

xxxx: two things really: ARM emulators are slow and doesn't provide very good performance in both CPU and OpenGL

yyyy: so why not pick a bunch of targets yonks ago like canonical has been dong, some high profile android handsets etc.

xxxx: and in 99% cases, building an app / porting it to ARM is a matter of flicking a switch, so, we give a good developer experience using a X86 'emulator' – it's really quite good, if you didn't try it already

yyyy: but canonical's approach seems clever to me, seems to garner way more developer/community interest by making their image available on "actual devices," rather than SDK only

xxxx: it seems to mostly backfire actually
xxxx: ubuntu touch channels are filled up with people who can't get it working on their devices, not app developers
xxxx: i fully agree with you that it's a good idea, but there's a couple of legal issues with providing completely ready images for other devices

yyyy: meaning $$$/time required, which Jolla thinks it cant afford, unlike canonical?

xxxx: ok, there's a very simple reason: the current public stack is qt4/x11 and it can't leverage magic new things like libhybris
xxxx: because you'd have to be on qt5/wayland (or surfaceflinger in ubuntu's case) to really take full advantage of it
xxxx: ubuntu's solution, in practice, is android, with ubuntu in a chroot
xxxx: it doesn't even use upstart, it's a tech demo

yyyy: oh those ubuntu images were all done with a much older stack werent they, before they announced usage of mir/libhybris etc, before they moved mostly to their own entire stack vs a mostly android-based one?

xxxx: mir isn't in use in any of their images yet, and they used libhybris from start
xxxx: what they did, practically, was take libhybris behind doors, make interfacing to android's Surfaceflinger (window manager/compositor) and add a QML ui on top

yyyy: yep

xxxx: and that was their tech demo and they sold the hell out of that

yyyy: so will he 1st major release of Sailfish be based on qt5/wayland, or that's not coming till next year right?

xxxx: we'll see

yyyy: So whenever it does come out, it will be much easier to do the sort of stuff canonical has been doing. i.e. at least 2-3 images for some of the most high profile android devices

xxxx: right, just note http://pastie.org/7736334
xxxx: so it can't ever be a truly pretty experience

yyyy: So as canonical are slowly moving away from what's mostly an android-based solution, they're going to have exactly the same problems as you guys anyway, right?

xxxx: probably, i think it's probably possible to do development builds, but it's really a matter of how you approach it
xxxx: you don't want to have a situation where 1) people wouldn't buy the actual device or 2) that the focus is more on hw hacking than sw development
xxxx: (this is just my personal opinion)
xxxx: with all the building blocks, mer, libhybris, etc, i encourage people to do all this stuff so we get all these building blocks field tested

yyyy: i guess that one of the main reasons why you guys are scared of proving some imge for android devices, as a large part of your revenue model depends on selling your own device, but I really think it could significantly boost sailfish/nemo profile & hacking community

xxxx: i don't disagree

yyyy: hopefully you move quickly towards providing at least 2-3 images for some of the most popular devices, "once" you're much closer to a Sailfish based on Qt5/wayland

xxxx: let's see, it might be a matter that people need to self-assemble images themselves

yyyy: I imagine that'll prolly be long after your 1st device it out anyway… so you'll have less to worry about with regards to potential sales being cannibalised

xxxx: so far we've (as in community) validated libhybris-wayland on intel SGX, allwinner mali, hp touchpad qualcomm adreno..
xxxx: so it's looking good so far

No true, lots has been happening there, even significant chunks of Sailfish are based on it...