The Current State Of Debian GNU Hurd

[phoronix]

Phoronix: The Current State Of Debian GNU Hurd

DebConf 15 happened last month in Heidelberg where there was many interesting sessions, not just for Debian GNU/Linux but also Debian GNU/Hurd...

http://www.phoronix.com/scan.php?pag...urd-2015-State

[IsacDaavid]

Phoronix: The Current State Of Debian GNU Hurd

Need you be reminded that GNU Hurd is older than Linux but is still mostly i686 bound and lacks USB support.

http://www.phoronix.com/scan.php?pag...urd-2015-State

You call this subtle Michael?

Hurd development is only one year older than Linux, and it has never seen more than 5 or 6 contributors at one time. I'm impressed that a microkernel nobody cares about but everybody likes to make fun at is actually this usable.

[thebishop]

I really don't understand why this project is still alive. All power to the devs (solidarity!), but it's a lot of work, and I don't see who would want to use it vs existing Free kernel options.

[ejr2122]

I would use Debian/Hurd and GuixSD/Hurd. A full-featured OS with a microkernel inside could be a really cool.

[IsacDaavid]

I really don't understand why this project is still alive. All power to the devs (solidarity!), but it's a lot of work, and I don't see who would want to use it vs existing Free kernel options.

The same could be said about Clang, *BSD, etc.
Their devs obviously don't see these projects as the only option in their areas, but they think choice and a different perspective are worth the effort. I guess experimenting with a radically different kernel like Hurd might be a lot of fun.

I would recommend reading http://blog.darknedgy.net/technology/2015/07/25/0/ to anyone interested in knowing what Hurd is about on a technical overview.

[tajjada]

I would use Debian/Hurd and GuixSD/Hurd. A full-featured OS with a microkernel inside could be a really cool.

Well, I would too ... but only if it had sound, USB, SMP, x86_64, proper graphics drivers, ... for a start. I am not even mentioning Wayland, Vulkan, OpenGL, any filesystem newer than ext2, RAID, ssd TRIM/discard support, ...

[mark45]

You call this subtle Michael?

Hurd development is only one year older than Linux, and it has never seen more than 5 or 6 contributors at one time. I'm impressed that a microkernel nobody cares about but everybody likes to make fun at is actually this usable.

Usable? Then shit is edible because you can eat it and not die.
Getting a kernel to support a filesystem and whatnot is one thing, getting it ready to compete is a completely different thing.

[IsacDaavid]

Usable? Then shit is edible because you can eat it and not die.
Getting a kernel to support a filesystem and whatnot is one thing, getting it ready to compete is a completely different thing.

Competition, usability and technical merit are different aspects. Look at Windows: historically it's been shit, slowly getting technically better, yet it kills the competitors in terms of popularity.

Hurd is neither trying to replace Linux, nor is it as useless for ever day usage as you think.

[ejr2122]

I would recommend reading http://blog.darknedgy.net/technology/2015/07/25/0/ to anyone interested in knowing what Hurd is about on a technical overview.

Cool article.

[Luke_Wolf]

The same could be said about Clang, *BSD, etc.
Their devs obviously don't see these projects as the only option in their areas, but they think choice and a different perspective are worth the effort. I guess experimenting with a radically different kernel like Hurd might be a lot of fun.

I would recommend reading http://blog.darknedgy.net/technology/2015/07/25/0/ to anyone interested in knowing what Hurd is about on a technical overview.

No it couldn't. Clang and *BSD are well developed solutions that do provide certain concrete advantages over their competitors (in fact other than OpenMP workloads there's not that much point to using GCC instead of CLang as Michael's benchmarking has shown that they are at effectively equivalent performance (other than OpenMP workloads) and CLang compiles things in less time). Hurd has some interesting concepts, but at the end of the day has no concrete advantages, particularly as Hurd is not the only competitor in the Microkernel department.

Now the devs are free of course to do as they please, but don't compare things that are competitive or slightly behind, that have their own advantages to something that isn't even close to being in the same realm of completeness.