Debian GNU/Hurd 2015 released | Hacker News Reader

Debian GNU/Hurd 2015 released

lists.debian.org

I am writing this comment from a modern graphical webbrowser running on an OS with the GNU kernel. Who thought this day would ever come?

Bare metal or VM?

VirtualBox. I don't have a spare machine lying around, but still cool...

Relevant question I would say is that even if you did have a spare machine lying around, how well would Hurd support its hardware?

It will be interesting to see if the Debian influence will promote the availability of a wide variety of architectures, including the options to load non-free firmware. From the GNU/FSF standpoint, this is at its core a GNU project, and GNU promotes freedom, free software and the rejection of non-free code. If a wide array of hardware support emerges, it will not come from the GNU portion of the project, where only either "freed" hardware (via coreboot, etc) or hardware with open specs is the target. Please note that "hardware" does not mean "system" - likely you can install on any number of systems that include lots of "closed" and unsupported hardware, but where you go from there is up to you...

http://www.nongnu.org/thug/gnumach_hardware.html

I don't know how well known it is, but Debian was started when Ian Murdock was an FSF employee.

Hey, Matt :-) Viva la Libre.fm!

I love Ian and I think of the folks that have drifted from _strict_ FSF/GNU practices, I think he's the most conscientious engineer who still keeps those values in mind.

I have my GNU-only systems, but due to my career in commercial and government software, I've just become comfortable with mixing free and non-free software.

Nothing has pleased me more than seeing this update to GNU Hurd.

I'm looking forward to updating my screencasts of GNU.

I may also put it on a ThinkPad.

Is it the year of Debian GNU/Hurd on the desktop?

I have it installed on a VM but can't seem to get Cinnamon running properly. Haven't messed with it too much but I was wondering what Desktop/Window Manager you installed to get to the point where you could use a "modern graphical web browser"?

I'm not picky and I just kind of randomly picked cinnamon during the install. Then Tasksel failed so when I restarted that part of the install I left the Desktop stuff out and everything worked fine. Once booted up I used aptitude to grab it, updated xinitrc to launch cinnamon-session but when I ran startx (after fiddling with enabling the hurd console that is) I was greeted with "Cinnamon Session Failed".

Anyway, I don't have a ton of time to mess with it but I'd like to try some options and see what I can get working. Any insight you have is appreciated!

Edit: When I said VM above, I mean VirtualBox if that matters...

I used fluxbox as the window manager. I simply installed iceweasel, xinit, xterm and fluxbox, ran startx and had working environment.

I'm extremely excited to check this out! I really like the idea of translators, and low and behold when I log in there's a tutorial on them in root's home directory! I might switch to this if the ui is bearable enough.

Link: http://www.gnu.org/software/hurd/hurd/documentation/translat...

What are your main motivations to use GNU Hurd? Not wanting to troll, fishing for a link for good background reading

Advantages:

http://www.gnu.org/software/hurd/advantages.html

http://www.gnu.org/software/hurd/community/weblogs/ArneBab/t...

Just read the advantages, some parts seem a bit old. Touting object oriented design as something enabling refactoring, whereas the trend for the last fifteen years has been toward interface-oriented design (with golang being the most extreme case).

Also, in a world of unikernels and app containers (trending toward the most static and restricted deployment unit possible), having an os being able to change its parts while running seems a bit useless.

But, who knows ? maybe we'll see new patterns emerge. That's a fantastic time for people interested in OSes.

the Hurd has an object-oriented structure that allows it to evolve without compromising its design. This structure will help the Hurd undergo major redesign and modifications without having to be entirely rewritten.

Yeah, good luck with that. OO is great but I'm not sure major redesigns are any easier in OO compared to any other arch.

Having a completely different OS arch in play does sound good though, perhaps opening up possibilities we havent realised yet. Perhaps someone should re-write that 'advantages' page to encompass the massive changes in Software thinking since it was first written in 2001.

> Yeah, good luck with that. OO is great but I'm not sure major redesigns are any easier in OO compared to any other arch.

It depends on what they mean by OO. I can't find a source at the moment, but I've heard of Smalltalk applications that have been running the "same" incrementally modified image[1] since the 1970s or 1980s.

http://en.wikipedia.org/wiki/Smalltalk#Image-based_persisten...

Hurd is a bunch of servers on top of Mach (there were attempts to get it ported to L4, Coyotos and some other kernels, but they all sadly fizzled out), so it is dated in many extents.

The main deal is the concept of translators, which are similar to 9P file servers and namespaces, and how all the Hurd servers are implemented. This design also enables easy persistence.

There is no trend between "object-oriented" and "interface-oriented" design, they're both terms with lots of meanings. The actor model is a form of object model. It's not necessarily bad.

The Hurd actually has its own system for running multiple concurrent instances of itself, subhurds: https://www.gnu.org/software/hurd/hurd/subhurd.html

Application containers still depend on a host OS. In fact, they're basically a way to partition the OS resources and library namespace so that you can get around its deficiencies by limiting each logical software bundle to a sub-OS. Each individual container is not any more malleable than what the host OS provides. It's quite the hack, really.

Unikernels/libOS are great, but they're just another technique that harkens back to the old days of writing code to initialize the machine, except it's now reusable through virtualization and library linking. Full OS deployments can still be useful.

Very nice. I'll check it out when I get home. Certainly had this filed as "it might never happen" and haven't checked on the progress in quite some time. I hope someone will build an OS course around it eventually. Seems pretty good for teaching purposes.

Congratulations to everyone involved.

Does anybody offer a cheap (like $5 a month cheap) VPS you could run GNU/Hurd on? I am feeling a perverse need to run a website on it. Or maybe that's still a bad idea at this point?

Gopher would be less perverse.

2015 will be the year of Hurd on the Desktop

For the tone of the comments here, it'll be more the year of the Hurd Virtual Machine.

I have a snapshot from May 2014 running as a VirtualBox instance. Tried IceWM and a GNU build chain on it. I'll probably update soon, need to find out what open issues have been addressed in 0.6.

I just rolled GNU/Hurd 2015 into a VM. It's early, but the first thing I'll note is that I felt for the first time that GNU/Hurd is a deployable OS. Still, I need that 64-bit version... killing me, here!

Do you have any info or a website you could point me to on how to build the image into a virtualbox VM? Running VBoxManage from their guide [1] isn't working for me. I can create the VDI file but on boot I'm getting an error saying hd0s1 isn't found >_<

[1] https://www.debian.org/ports/hurd/hurd-install

Have you tried qemu/kvm? You also use the pre-rolled here:

http://ftp.debian-ports.org/debian-cd/hurd-i386/current/READ...

>Since the last snapshot release coinciding with "wheezy", the init system has been switched to sysvinit for a more Debian-like experience.

Hehehe, didn't Debian just switch to systemd with Jessie?

They changed the default to systemd. Sysvinit is still available.

Availability of Sysvinit does not mean you can use it as safe as systemd , there is no requirement for Sysvinit support in debian package's anymore , so if you want use Sysvinit then you maybe encounter some package which does not work on debian.

Debian/kFreeBSD also cannot run systemd -- it's been designed for Linux, rather than for portability.

I wanted to look at the source code. Just to save anyone else the time to find it, the link is:

http://git.savannah.gnu.org/cgit/hurd/hurd.git/tree/?id=v0.6

Why Savannah (in savannah.gnu.org I mean)?

Gnus live in the Savannah?

Edit: ninjaed by detaro

What are the advantages vs. Minix 3?

how far along is Minix 3?

if there's one thing I never expected it was to be downvoted for asking a legitimate question.

Sounds similar to hurd. 32bit only, graphical ui, bsd applications. Theoretically usable as a daily driver.

> graphical ui

Not currently.

> Theoretically usable as a daily driver.

Not likely.

Minix 3 doesn't have Debian userspace :)

It did back in 2011, at least on one Debian developer's computer:

http://www.braincells.com/debian/index.cgi/search/item=270

Hurd is based on Mach and available for x86 only. You can run X11 and Debian packages. Drivers for Linux have been ported. Debian/Hurd re-uses some of the Debian community infrastructure for development. Has a goal ("viable for everyday use, and gives users and programs as much control over their computing environment as possible"). The focus yields things at the intersection of research/pragmatic use (e.g., introducing novel features like translators).

Minix is less mature from a community standpoint and its development processes. It's developed and run by about half a dozen contributors (generous estimate). Mostly students. Lacks leadership. It runs on x86 and arm. The microkernel design is its own. It's been possible to run X on past versions, but despite other comments, X11 is not actually usable on the latest release (3.3.0, released 2014 September). No (graphical) web browser has ever been ported. Running primarily parts of the NetBSD userspace. Not so much interest in taking advantage of the modular design to introduce novel features; implicit focus on porting existing BSD packages instead.

wait... its not april 1st? awesome. :)

Not this time — but 4 years ago it was ☺

→ http://www.gnu.org/software/hurd/news/2011-04-01.html

Times change.

Reply on news.ycombinator.com