On 07.08.21 23:20, Hendrik Boom wrote:

> Inferno is the system I would have preferred Google used instead of
> Android.

Ah, okay, that was a bit confusing ;-)

Actually, I'd like to see a combination of both Linux and Plan9
concepts - using the Linux kernel for all the HW stuff but extend
it to support Plan9'ish approaches (eg. mount-ns).

Much of it is already there, but still need to find a practical way
for having private mount namespaces as unprivileged user (w/o userns).
Already did some experimental implementations of /dev/cap*, srvfs, etc,
so we could design programs pretty much the plan9 way, but that's still
a longer way to go.

> It isn't a candidate for FlyingTux virtualisation.
> But it might be able to alleviate some of the frustrations
> inherent in Android.

Not really, since the most problematic point with those devices is
kernel/driver support. When starting over with a completely new OS,
we'd have to start afresh with all these (usually completely
undocumented) devices. Using Linux as kernel IMHO still was the
best choice - but their userland is a crappy monster

> I get it that there are better ways to have orovided the services Android
> provides. But creating a better system than Android using containers and
> virtual machines is not going to eliminate the horrible constraints that
> Google has foisted on the computing public by their Android system.
> Without copatibility with Android and access to the Android app market,
> it will not replace Android, not will it make it easy for those wishing
> to escape Android still be interoperable with friends wo do use Android.

I don't intend to replace Android anytime soon - I'm trying to create a
Linux distro for mobile devices. Mobile here doesn't just refer to the
device itself, but also the applications (eg. easily moving around app
instances between devices, etc).

For now, FT is just a research project, unless many people join in, so
we can build something practically usable in the field.

> And it does networking. Just about every resource is modelled as a file,
> and the file system naturally extends to other networked systems.

Yes, Plan9 already does this. For typical (existing) applications, I
don't see this a major issue when using containers anyways.
(don't have the resources to rewrite everything, eventhough I'd really
like to enjoy that).

> So putting it on top of Android and then using Infirno apps might well
> be a way to alleviate some of the Android non-interoperability,

But that wouldn't solve the major issues of Android - a huge, pretty
unmaintainable and untrustworthy monster.


--mtx

--
---
Hinweis: unverschlüsselte E-Mails können leicht abgehört und manipuliert
werden ! Für eine vertrauliche Kommunikation senden Sie bitte ihren
GPG/PGP-Schlüssel zu.
---
Enrico Weigelt, metux IT consult
Free software and Linux embedded engineering
info@??? -- +49-151-27565287