Google's not-Linux OS documentation cracks box open at last

Re: Call me a cynic....

If you look at how successful the Chinese versions of Android are without Google Play Services are, you can see that it can be done.

Re: Call me a cynic....

"Why should anyone contribute to something that is ultimately to make a multi-billion dollar advertising company even richer?"

I'd suspect there is more money in the Google Play Store than Android itself, therefore it would be a pretty terrible business model to make it Open Source, why would they give something away for free that their competitors could re-use to challenge Google? That would be a terrible business model

Re: Call me a cynic....

Why use the google play store? I've been using Aptoide, and there are other alternatives for the play store.

Even better: get LineageOS on your android device (if it is supported), the LG L90 I've tried it on runs better than ever.

Fuchia not such a dream future

An OS I can't contribute to that just gets tossed over the wall every so often. Also combined with the suggestion this will only install over network (with local device caching). Yeah a great win that!

I'm not sure anyone can realistically replace the Linux kernel (even with Google's resources) and provide the same level of functionality. For all Android's sins (and there are many), I can still get access to standard Linux things that are unlikely to be present in Fuchsia (iptables, monitoring tools, proc filesystem etc) some maybe, but certainly not all and not with the depth of functionality.

Also Linux will have much greater driver/filesystem support than Fuchia ever realistically could. This allows third parties to take Android AOSP sources and tune the kernel build for new hardware/filesystems.

As a recent Google ex-employee said, Google are now more interested in competing than brining new technology. This is likely the main motivation (to lock out Amazon and other vendors that use AOSP) from simply taking their OS (even though this was encouraged before they had market share).

This is unlikely to benefit end users...

Blackberry's microkernal OS could run Android apps - something about shimming in API calls as required - and Google have more sway with Android app developers than Blackberry did. Maybe Google will just encourage them to develop their apps with tools that easily allow Fuschia as well as Android (and iOS) apps to be built?

"Surely anything that replaces Android would have to be able to run the millions of Android Apps or else it will become another Windows mobile, Firefox OS, Blackberry etc."

That's an instantiation of a generic class of comment. You could replace the names with CP/M, MS-DOS and any other formerly popular OS you care to think of.

Right now we need a mobile OS that doesn't allow apps to poach each others' data. Will Google deliver that?

The article mentions Fushsia presents a POSIX system call interface and that libc will be available. Between those two items, you're close as damnit to drop-in user land tools, e.g. Dalvik/ART/Play Services. Android currently uses a Googly variant of libc, but libhybris has made it possible to run the rest of the Android stack on stock libc. I suspect Dalvik and ART (more so because they're pre-compiled) apps won't notice an OS change. As long as the same POSIX/Linux, OpenGL, OpenCL, etc APIs are available, native apps may also not notice.

Oreo's separation of the OEM blobs from the main OS also helps. Fushsia, like SailfishOS, LineageOS, etc, could just piggyback off the existing Android HAL blobs.

This is probably the road Samsung had been aiming for with Tizen, i.e. being able to just swap it out under the waterline, but Google's progressive moving of functionality out of AOSP into Play Services scuppered that idea. The locking down of the Play Store against "uncertified" devices may well be one more ratchet of that thumb screw.

"it will become another Windows mobile"

Apple switched OS and architectures letting old app work for a while. Nothing hinders Google to make most actual Android apps work somehow while they are replaced, ending support as soon as it feels it is needed.

Moreover, when you customers have little other places to go, it's far easier - and the only competitor of Android is iOS - which only runs on quite expensive devices.

Surely anything that replaces Android would have to be able to run the millions of Android Apps

If you have something that runs POSIX stuff it's not that difficult. FreeBSD will happily run Linux binaries, with no overhead, with the relevant libraries installed.

On top of a microkernel Google should be able to recompile most of Android fairly easily, though it's not entirely clear whether they're planning to do this with Fuchsia.

anything that replaces Android would have to be able to run the millions of Android Apps

Get to work on Sailfish then.

Re: impress me

If it closes the way to help get end-users root access

If anyone thinks an open source (to allow passers-by to drop in 'of a lunchtime' and do some work the staff haven't) mobile phone O.S by a corp. that really just wants to peddle adverts and collect marketing info is going to enable Software user freedom the likes of which the FSF dream of...

Not to mention that the first phone carrier/phone manufacturer isn't going to totally ruin it with some floppy ill-fitting skin-dance like demented cannibals.

....They're deluding themselves.

Re: Operating sytems introducing new looping techique.

32-bit handles, not pointers. So a limitation of 2^32 objects, not bytes.

Re: Why C?

Good point, though that's probably still just because world + dog can link against C libraries. Rust probably has not-invented-here problems for Google, there's Go... though that requires its own runtime.