Greetings.

Didier Kryn via Dng - 19.06.26, 16:08:21 CEST:

>      Asternias uses two main arguments in its marketting: the
> compatibility with Linux from the application's side, and the kernel
> written in Rust.

>

>      About mimicking the Linux User-space API: Well, I'm a reader of the
> MUSL mailing list and can read pretty often remarks on badly designed
> Linux system-calls. And reports to the Kernel group are just vain, so
> that the libc maintainers (of all libraries, not only MUSL) need
> work-arounds.

Interesting. If it is low enough volume, I may just subscribe to that
mailing list.

> So, maybe some improvements would be welcome, and
> attention to userspace developpers. Going more radical: why choose
> compatibility with Linux and not AmigOS instead? You, who know it,
> would certainly consider it a better option.

Nice idea. But then what is it about application compatibility?

Like about these application based operating systems called web browsers.
They are challenging to port to AmigaOS API to say the least.

>      About the kernel written in Rust: I'm suspicious on the Rust buzz.
> No doubt it is better than C, but, when it comes to switching to another
> language, the minimum a serious programmer would do is a review of all
> languages; it would lead to something else than Rust.

Ada?

I wanted to look more closely into V programming language. I like their
approach of self-hosting the compiler with minimal dependencies.

Not sure whether it matches or outperforms Rust regarding safe
programming. But it offers some in that area.

Its compiler is blazing fast, they claim: "V compiles 1.2 million lines of
code per cpu per second."

https://vlang.io/compare

>      So, maybe the best option would be a micro-kernel written in Ada
> with AmigaOS system-calls.

Maybe. Did not really research much about Ada. Any pointers where I can
start? I would like to have an overview. Of course, I think there has been
an Ada thread here recently, so maybe I can just look there for pointers.

>      Why compromise, when you start a brand new project? This is
> something I can't stand with. It is like if the authors were Linux fans
> and Rust fans, fully ignorant of alternatives. It always disapoints me
> to see how much software engineers are unable to make rational choices.
> It is particularly disapointing because these people are most certainly
> excellent progammers with plenty of energy and dedication.

Good questions. Maybe if you add applications to drivers, it is just quite
much to develop something new from scratch? If you invent everything new,
then everything existing will be incompatible.

I'd love that. But will such an approach ever gain enough traction?

It would be great to look at: What works and what doesn't? And then take
the best out of any excellent operating system approaches.

But it would be incompatible.

Best,
--
Martin