User:John Beals

From ArchWiki

Hello-

What I have learned: Separation Kernel, is it right for Arch?

My thoughts on Arch is that it is an excellent 'anarcho' effort (which I mean in an only good way), but in the way Linux is expanding, complication is not benefitting the effort (from this view). My view is that Linux has become Systemd and Dbus, and that doing Linux implies comprehending these two, and thus becoming a linux server expert.

Resistance to Systemd is apparent (but futile ;) ), but I see truly-free systems going in a wholly other direction (if they choose), that being the tiny hypervisors that are being created for appliances. I long campaigned for a free L4, and L4 is free now, but along with L4 are many free micro kernels, and even smaller, the Separation Kernel (SK). The benefit of the SK is that it can host applications in ways that Android cannot, as OSs - and safely. Further, it is the boot system, and it seems to me it can be flashed into motherboards. With RiscV (5) coming, there is an excuse to write increasing amounts of code in assembler following the precise competitive but unified development model that Linux evolved. But that for me, is on the user side such as a fully-functioning computer on a wrist watch, but not the server side. Of course servers will implement Sks, but probably not too soon; they are meant for appliances.

Given that RedHat seemingly invented Dbus and Systemd, that seems a good place to go (especially since I am looking to re-enter the indu$try). I want to reinstall Arch to explore some shortcuts I thought of, but Arch is making me want to go Linux from Scratch - after all these years! This is mainly to get an intuitive feel for SystemD and Dbus, which I don't expect from the new CentOS I just loaded (on an old disk). What I expect from RedHat is what they have saturated my FB stream with... I assume most of you get the same if you are still on FB. I built my world in Perl, but Python seems to rule the roost, and I can give many examples of why I see that.

The OS group on reddit simply will not discuss Rust, but given the RiscV reality, I am not certain Rust is necessary. What would be nice is to add a shell to manage an SK, perhaps from system flash. In my view, this shows a radical path for Arch or a similar 'anarcho' system that would put it well into the future. I remember when Linux was just a ripple growing into a 20-yr surfable wave that keeps on growing. But given what I have learned, there is a new wave developing, equally surfable that may carry us to where we really want (or should want) to be: centenarians.

Just getting started 18 February 2021‎

Strong possibility that I will be back in tech after almost 20 years. Reason I feel confident is that nothing has changed! I mean that in a nice way. Exceptions, of course, are containers, and, somewhat related, systemd.

Systemd is what is most relevant, and I think that what would be a good exercise to find a non-systemd distro, and convert it to systemd. Then, knowing how systemd works, contribute to the full conversion, by which I mean eliminating all those symlinks.

As far as containers go, that's all kind-of weird to me. I can fully grok Docker, but the K8s world is really confusing. If it's free and not owned or even made by Redhat (with OpenShift being the only Redhat component-which is also free), then why did IBM buy Redhat for obscene billions?