Or, if you aren't insistent in scraping all of Android off of the device, it's certainly possible to make use of the HW OpenGL ES support and hardware composition provided by the SurfaceFlinger.
Yeah, running your own compositor and window manager / X server against the bare metal is challenging, but one could bolt this stuff on top of the facilities provided by Android, expose a suitable shared memory interface, and then run x/gtk/gl/etc apps with a connection to your bridge (either in a chroot'd container or just compiled standalone)
A little convoluted, sure, but something that could actually enable existing Linux desktop *applications* to run effectively as first class citizens in a stock Android environment (with 500M+ devices out there, that seems like an interesting target).
I haven't missed Linux desktop apps on my phone or tablet enough to dig into something like this, but if people believe that Android would be massively better if you could run a bunch of those apps, there are ways to do it (and get reasonable performance) without users having to "root" their phones, install custom builds, etc.
MeeGo to return next month with Jolla phone launch (The H)
Posted Oct 10, 2012 21:45 UTC (Wed) by raven667 (subscriber, #5198)
[Link]
I'm no expert but it seems that there maybe an opportunity for some standardization between fundamental technologies like SurfaceFlinger/AudioFlinger and Wayland/Pulseaudio. If they could co-exist or if apps written against one API could be run on the other with a thin shim then you could more easily move apps back and forth between traditional desktop and mobile systems. One of the benefits of the Apple iOS/OSX ecosystem is that they share a lot of the same underlying infrastructure. Even better would be to standardize on one stack for A/V although it might still be sufficient if both stacks use the same underlying kernel support infrastructure.
If we end up with competing stacks with competing kernel drivers then that weakens traditional desktop linux much more than Android. If the systems can cooperate or standardize then they can add to each others strength in new ways, as demonstrated by Valve recently. If all the drivers were upstream in the kernel then the talent for optimizing them can be pooled and multiplied. If the A/V systems are standardized than that also provides a nice target for other distributions like Jolla, webOS, etc.
MeeGo to return next month with Jolla phone launch (The H)
Posted Oct 10, 2012 21:56 UTC (Wed) by shmerl (guest, #65921)
[Link]
It already happened (Android made a competing stack). Whether something can be bridged - is a good thing to explore, but the way things will progress is that regular Linux (including mobile distros) and Android will go separate paths in the foreseeable future. That was the point of my dislike to Android design decisions expressed above.
MeeGo to return next month with Jolla phone launch (The H)
Posted Oct 11, 2012 2:43 UTC (Thu) by daniels (subscriber, #16193)
[Link]
We had a good and really productive conversation about Wayland and how it relates to SurfaceFlinger, HWCompositor and WindowManager in the Android sessions at Plumbers. It's definitely possible for a Wayland compositor to reuse some of the smarts from there, but it's a fair bit of work.
MeeGo to return next month with Jolla phone launch (The H)
Posted Oct 10, 2012 21:46 UTC (Wed) by shmerl (guest, #65921)
[Link]
That probably wouldn't be efficient enough given the mobile hardware performance restrictions to run let's say KWin with PlasmaActive on top of it, but I'm not sure if anyone tried that anyway.
Using the lowest layer without any further Android stack to run other Linux DEs might work with help of something like libhybris: