Be warned this post is not so much a “feature request” as a flight of fancy to provoke discussion. And I am unapologetically a fan of Genode which is a fascinating piece of technology, and hopefully this thread might inspire others to try it even if the topic itself is not taken forward.
Google for a while was working on an operating system, Fuchsia, that would replace the underpinnings of Android. Fuchsia development has been seemingly put to bed after being used in a handful of homehub devices. Whilst Fuchsia might be an option for /e/OS, I wanted to discuss its European cousin, Genode, for this role.
Like Fuchsia, both are microkernel and use “capabilities” as the basis of system security. Genode is however still being developed, unhindered by the whims of silicon valley big business. It is packaged as a stand-alone operating system called Sculpt that you can play with on your PC that is very cool as it has a unique graphical means of managing the system. Sculpt is also available for the Pinephone, which not only demonstrates its viability as a mobile system, but is a serious statement of intent they see mobile devices as a future application.
Now I understand that as far as Android is concerned the operating system is only there to provide some sort of Java virtual machine as that is what the apps use, and Google were able to substitute Fuchsia for Linux in this role. Genode is described as an OS framework that can be assembled in a modular fashion and can use different microkernals depending on application Might a minimalist version be crafted to provide the basis of Android?
What would be the benefits if Genode were to replace the Linux underpinnings within /e/OS? It would get the existent apps working whilst providing a modern underpinning for future evolution of /e/OS. Genode would in fact provide /e/OS with a more sophisticated basis than Google Android (unless they more to Fuchsia) and Apple’s iOS, an advantage that would help /e/OS, with comparatively few development resources keep up with these bigger firms. Personal security would be improved, both due to the fundamental architecture and the “leitzcentrale” graphical management screen you can experience in Sculpt, which should also free up developer time in the long term. Genode, originating in Dresden, would help achieve EU technological sovereignty. Huawei, for example, went beyond a simple fork of Android to make their “Harmony OS” and I understand this is now incompatible with the main branch. Perhaps the EU needs to the same?
I hope this has been an interesting pitch of Genode. There is lots of documentation available on the website for further reading, and also a friendly community keen to help new users get acquainted.
I applaud your enthusiasm, but no genode is cute (hw demo time at ~33min mark) - but if you want to be avantgarde, there’s mobile linux with sxmo already
Thank you for that link, that talk covers a lot of useful ground. However the rest of your post did make me wonder if you even read the summary, never mind watched the podcast to which you linked :
Your post invites comparison with Sxmo - and, by implication, mobile (GNU/)Linux more broadly - and I note these mobile Linux efforts have never really gained traction. Genode - even in its Sculpt form - is not just another “mobile Linux” distro - or any other sort of (GNU/)Linux for that matter*. Genode would in fact “de-Linux” /e/OS which uses that kernel at present. To get round the lack of apps for Sculpt, my suggestion is to use the Genode “OS Framework” to replace Linux kernel under /e/OS and thus leave the actual user experience and apps the same (at least to start with).
Genode is better thought of as a kit of parts - hence their description of an “OS framework” - from which task specific OS an be built. The task in this instance being to replace the core of /e/OS with a more modern underpinning. In fact, Genode is so good at virtualising guest OS, it might be possible to achieve the same by runing /e/OS apps in their own virtual machines as required within Sculpt (presumably with security benefits).
(*) Actually it can be, there is the option to build it on a linux kernel rather than the microkernel for developmental purposes - but that is not the normal intended application.
Genode can run Android with a linux kernel in a VM, but I’m not sure what is its ultimate benefit, other than emulation/VMs helping when transitioning. Running a full Android seems orthogonal to Genodes overall design goal (“trust because you grasp”?). I’d rather see them doing their own thing. In that spirit I linked the niche UI of sxmo, projects following their vision.
This response of its project lead is I think a good answer to your personal quest.
Thank you once again for more excellent links, which I found helpful. even though I consider myself acquainted with Sculpt (if not yet a daily user). The last link in your previous post gives a good example of why I like this project. I have been privileged with similar wordy and considered response from developers in Genode (and Haiku) forums to my own noob questions. I doubt you get that with Linux
Speaking of which, I opened a sister thread to this one over on the Genode forum. One poster there brought it to my attention that Genode does maintain a to-do list and that includes porting Android ART (the virtual machine that runs the Android binary apps). So both methods of either virtualising Android in its entirety or running its apps will hopefully be available.
Other than that I agree with you that Mobile Sculpt would be super-cool. What seems to matter nowdays - just as much as the traditional barrier of having a library of software for your novel OS - is the cloud services. If Murena was able to provide cloud for both /e/ and Sculpt then that would solve the issue, and users could in fact run both OS on different devices as they transition over. And whilst you can’t really charge for software these days, the cloud makes things a commercial proposition.
in general the lego bricks don’t stack that easily. Genode is at 1 device after porting the pinephone drivers from linux. Not to discourage you. Small things do add up - I’d work on those, especially something that is a personal itch you want to scratch.
I wonder if something like the CFM1 phone resold by Murena is open and well documented enough to port Genode (as well as an obligatory mobile GNU/Linux distro)? It is powered by a Mediatek processor rather than the the legal firm of Qualcomm and Partners, who when not doing court cases specialising in intellectual property also make phone SoCs, or so I’m told.
The latter has, joking aside, proved rubbish at alternative operating system support with their PC ARM chips.
Murena may in fact gain a few - still useful - extra sales for their phone by making it easy to port alternative operating systems.
I wish I had the skills, but in their absence my own small things are more to do with helping get the word out about Sculpt, updating the wiki page, etc.
Among its competitors, qcom does most foss contributions. Murena is not a hardware company. If you’re lucky a Genode dev will pickup a FP5, there are drivers that can be ported. Maybe the Android Runtime thing will happen, that’d be interesting.
Best investment imo from a Genode perspective is into native apps that talk open protocols. How about a Sculpt UI based contacts app, using an existing carddav library? could be fun
What you say seems about right, because the SoC in the Fairphone 5 is more of a general one, and I speculate, would be better documented and supported for people and OEM who want to go more “off piste” than the mainstream phone SoCs.
Android Runtime in Sculpt vs native apps, I don’t know if one way is “better”. Genode certainly hope to add contacts into basic phone functions available through Leitzcentrale. Longer term there might be a mix of stuff ported over (e.g. with Qt) but getting all the Android stuff running would be a win. And the architecture of Sculpt is arguably a better foundation for running it safely than traditional Linux based Android.
Google is reducing Androids trusted footprint via AVF, launching micro vms (fuchsia too apparently). It’s an incremental approach that can yield similar benefits.
genode 
: