PowerNex: A Kernel Written In The D Programming Language

[RamblingMadMan]

The GC is optional, i prefer D over the old and ugly C/C++ languages

Well, if you look at C++ >= 11 or C11 you can hardly say "old and ugly" anymore. Especially C++17 + Concepts.

[schmidtbag]

Apparently, I'm the only one here who sits in the middle of this whole situation. Sure, there's nothing wrong with creating a new kernel or even OS using entirely D, but, what good does it serve? This isn't going to be the new Linux - it will never take off. D isn't a very popular language, and there are plenty of other kernels/OSes out there that are better than Linux and are still less popular (and likely always will be). How will this be any different?

If a new OS is to be made, it needs an edge. A distinct advantage to get anywhere. FreeBSD may have security or stability enhancements over Linux, but the licensing is a burden. ReactOS is a great idea but most people just want something that works. If people want a free way to run Windows programs, they'll use wine or they'll just pirate Windows.

Obviously, the people working on this have skill. But the way I see it, it's better to use that skill toward something everyone will benefit from or enjoy, rather than create something that will inevitably be abandoned some day. It just feels like a wasted effort on good manpower. There are other things you can code "for fun" that won't be completely overshadowed.

[Michael_S]

DragonflyBSD is porting drivers, as are the other BSD I believe. You don't always have to rewrite from scratch.

Porting the drivers from where, Linux? My understanding of licensing is that if you use a piece of code as a starting point, you have to use the same license or a compatible license even if you rewrite large portions of it. So I don't understand how any of the *BSD kernels can get drivers from Linux without shifting their license to GPLv2.

And in fact, one of the criticisms of PowerNex in one comment Reddit is that some of the code was originally taken from the Trinix project. Trinix has source code available but is not open source. The PowerNex author rewrote the Trinix code, but arguably he should be using their license for that code instead of MPLv2.

[profoundWHALE]

I can't possibly understand why the fuck would you waste your life like this working on such useless shit.

Because they can. Also, one man's trash is another man's treasure!

A neat start that will go nowhere when you will realize how hard it actually is to create a new kernel. And I'm not talking hard as in hard engineering, I'm talking hard as in to get others to write drivers for it.

That's a legitimate criticism. Whenever there is something new, even if it is a lot better, requires a lot of convincing and/or marketshare to typically get people to make things for it. Similar to Windows and their viruses.

DragonflyBSD is porting drivers, as are the other BSD I believe. You don't always have to rewrite from scratch.

[Min1123]

For anyone who wants to play with it, it's easier to get the ISO (legacy BIOS CD image) from the releases page than to compile the source with the supplied toolchain.

It's a small thing right now with the ability to boot via GRUB and a couple of inbuilt commands in an inbuilt shell, but it's a neat start.

[waxhead]

Great! Another fucking kernel. That will go after Hurd that is a turd. Great work guys! Don't help with what already works! Be original! Reinvent the PC! Fuck Intel! Make your own chips!

You are not a very bright individual are you? I am sure some Windows guys said the same thing when Linux came along and these days that "fucking kernel" is now running just about anything you can imagine. As Linux takes over more and more and will probably eventually be just as mainstream as Windows, the big exception is that Linux's license is a bit nicer to everyone. Why don't you check out Minix v3 as well? it's actually really cool and it is such a "fucking kernel" that it eats bugs for breakfast as well!

[justmy2cents]

i simply love the NIH, first it was that everyone wrote file managers, then they went on everyone writing DE, followed by everyone writing their own init, now the amount of new kernels says fashion changed into that direction.

can't wait for next step where everyone is doing their own CPU

[Michael_S]

im just not sure how anyone can get super excited about a garbage-collected systems language

1. Garbage collected systems languages like D or Go still generally provide much faster execution and lower memory overhead than scripting languages, and they make it easier to write low-defect code than C or C++.

But to be fair, the Go team has been working very hard to optimize their garbage collector and they've achieved performance on par with Java. The D garbage collector lags behind either for efficiency.

2. Garbage collection in D is optional, and the main language designers (Walter Bright, Andrei Alexandrescu) are working on a rewrite of the standard library so that you can use it with or without garbage collection. If they can make that work, I think it's a best of both worlds option - but on the other hand, unless they figure out how to do it in a minimally invasive way, migrating existing D code to the new version might be a nightmare.

[ssokolow]

The GC is optional, i prefer D over the old and ugly C/C++ languages

But, to be fair, the standard library does rely heavily on it, so going GC-less is only really practical for kernel developers and people willing to reinvent the bits of the standard library that they use.

(That "or people willing to reinvent the bits of the standard library they use" part was the reason a D fan gave when they made a thread on /r/rust/ while inquiring about the completeness and maturity of Rust's gamedev ecosystem. They were tired of being the only person who understood their D code.)

[Scellow]

im just not sure how anyone can get super excited about a garbage-collected systems language

The GC is optional, i prefer D over the old and ugly C/C++ languages