Debian GNU/Hurd 2017 released | Hacker News Reader

Debian GNU/Hurd 2017 released

lists.debian.org

116 pointsjrepinc18 comments

How feasible would it be to have Duke Nukem Forever run on it (under Wine)?

(HURD and DNF have both been synonymous with vaporware for a long time, sometimes combined to "Yeah, X is going to be finished right after Duke Nukem forever runs on the Hurd". It would be a fun gag to make it happen.)

I keep hoping GNU/HURD will eventually morph into something usable and compelling... but I've been hoping that for nigh on a dozen years already. Certainly progress is being made and this release is an impressive advance in several quarters, but unless it gets traction in some manner that compels corporate sponsors to contribute to a supersonic spurt in development, it seems destined to lag increasingly far behind other open source OSes such as Linux and the *BSDs. And that's a real pity, because a usable, well-implemented micro-kernel architecture widely deployed and supported would be something very much worth everybody's while.

I don't think the kernel itself, especially a micro kernel, is all that big of deal these days.

There are already plenty of decent Unix-style and other kernels out there, mostly relegated to academia and research.

Any of these would be decent enough to lay the foundations of a free stack.

The real problem is the APIs, services, and the hardware drivers.

The locked garden style of hardware and subscription models are becoming more and more prevalent.

And even using Debian or some other Linux doesn't amount to squat if you're using a back-doored Intel chip with its 'management engine', running a browser which is making it easy for 200 trackers to follow you on every site, and a smart phone, designed by an ad company, has turned you into a broadcasting station.

> There are already plenty of decent Unix-style and other kernels out there, mostly relegated to academia and research.

The Hurd is the only multi-server microkernel system with a POSIX persona, though. The Hurd is a very flexible system, it just needs all those Linux developers to make progress more quickly.

The Hurd needs someone to actually use its tools. For example to use subhurds to build the next great virtual hosting facility.

Or to harden your own system by running your browser with minimized rights, only granting it the capability to access the filesystem outside its cache if and while the user actually wants to access it, and only for the right directories. The Hurd already provides that option.

If you want to test the Hurd quickly, just run this:

    wget http://people.debian.org/~sthibault/hurd-i386/debian-hurd.img.tar.gz; tar xf de*hu*gz; qemu-system-x86_64 -net user,hostfwd=tcp:127.0.0.1:2222-:22 -net nic -m 1G -drive cache=writeback,file=$(ls de*hu*g)

or, if you want to stick to 140 chars in your commandlines:

    wget http://people.debian.org/~sthibault/hurd-i386/debian-hurd.img.tar.gz; tar xf de*hu*gz; qemu-system-x86_64 -hda de*hu*g -m 1G

(via http://www.draketo.de/english/free-software/howto-hurd-140-c... )

For official info, see https://www.gnu.org/software/hurd/hurd/running/qemu.html#ind...

Does anyone know if Hurd will ever be production worthy? Impressive progress nonetheless from what I understand. I know it must be difficult to get developers as there are so many people using Linux & BSD. Fuchsia is supposed to be a micro-kernel too right? I wonder if there could be a libre fork of that.

Speaking of microkernels, MINIX 3 is open source, unlike the version of MINIX that inspired Linus to write Linux.

MINIX 3, unlike earlier versions, aims to be used in production rather than being purely a research OS.

MINIX 3 has incorporated code from NetBSD making it possible to run a lot of software on MINIX 3 from the NetBSD ports collection.

I have not used MINIX but I remain interested in it and think it deserves a mention.

Thanks for the reply! I have one of Tanenbaum's old books. Perhaps time to read it.

Google Fuchsia is based on Google Magenta micromernel. But they are slightly different story. Google is trying to make a whole new OS for embedded system. I mean it is not aiming POSIX-compability.

> It is now possible to run subhurds as unprivileged user, thus providing easy lightweight virtualization.

That would be amazing to achieve on Linux. Unfortunately UML is not quite the same. I wonder if they'll ever get to the place where Hurd could be a coreos-like system with better isolation.

At this point, is GNU/Hurd just for people who like to hack around with OS kernels?

I mean, it's interesting to see the progress now and then, but I can't imagine seeing it on anyone's short-list for OS's that could be deployed in a production environment.

Obligatory XKCD link https://xkcd.com/1508/

Does Hurd use systemd?

Also, how's the performance? The last time I tried Hurd the performance was atrocious.

No, it has an initialisation server (helpfully called init): https://www.debian.org/ports/hurd/hurd-doc-server#init

I'm curious to know what made you wonder whether it did, and whether you would consider that a positive or a negative (and why).

No, it has an startup server (helpfully called startup), since 2015: http://git.savannah.gnu.org/cgit/hurd/hurd.git/tree/NEWS#n86

I stand corrected. Apparently things have changed since I last used it. Apologies to all.

No, and systemd would not add compelling features to the Hurd, since the Hurd already provides most of the advertised advantages of systemd, but cleanly integrated into the system — and has been doing so for decades.

Some of the reasons for that are discussed in this article: http://www.draketo.de/light/english/free-software/some-techn...

If you want performance from the Hurd, run it in a qemu VM so it profits from the caching algorithms of the Linux host. Then performance is actually pretty good — and you still gain most of the advantages of using the Hurd as OS.

Reply on news.ycombinator.com