Slashdot Mirror
GNU Hurd To Develop SATA, USB, Audio Support
An anonymous reader writes "Hurd, the GNU micro-kernel project that was founded by Richard Stallman in 1983, may finally be catching up with Linux on the desktop... Plans were shared by its developers to finally bring in some modern functionality by working on support for Serial ATA drives, USB support, and sound cards. There are also ambitions to provide x86-64 CPU architecture support. GNU Hurd developers will be doing an unofficial Debian GNU/Hurd 'Wheezy' release this year but they hope for the Debian 'Jessie' release their micro-kernel in Debian will make it as part of some official CDs."
Its fucking absurd that USB support and sound cards and SATA support is news in an operating system today.
Being able to run on a somewhat modern computer (they all come with SATA drives and USB ports nowadays - no support for those two basic technologies means your kernel just won't work on any hardware that's not totally obsolete by now), and being able to actually use all the hardware in that computer, is a fairly important feature of a useable OS, imho.
It's a microkernel, check Wikipedia.
Basically you will get clearly slower performance, but possibly much more reliability/stability, security, and all the benefits that go with modularity.
The point is that
a) computers will get so fast that the performance hit doesn't matter in standard programs
b) people hope to find ways of improving performance somewhat more into the direction of monolithic designs (=all the major platforms in use)
c) some application areas simply put additional stability over performance, so if we had a working microkernel... (no, Minix isn't good enough)
For now, best take it as a research project.
I hate to interrupt your Stallman bashing, but RMS isn't involved in Hurd development. He has been content to use Linux for many years now. Hurd development is driven mainly by other developers who are in it purely as a hobby, a way to play around with microkernel design, and they are not striving to reach a mass market.
Bill Gates didn't get to be, well, Bill Gates - by trumpeting Basic and DOS until people started saying, "Who?" He cut corners and compromised and, ahem, borrowed good ideas. It made him a gazillion dollars. And Richard, for all I agree with your ideals, and for better or worse, Bill Gates influenced the course of development of the personal computer more than you ever will.
What a shallow comparison! There are people whose main motivation does not come from how much money they can make or how much power they can gain over others. RMS's motivation does not even remotely have anything to do with Bill Gates' motives or 'comparing of penis length' type rituals such as 'Who has had most influence on PCs?'
People who are mainly motivated by power and greed tend to ridiculde and diminish the achievements of these people. But in the long run, their rantings doen't count. In two hundred years from now people will very likely still read the novels of Thomas Pynchon, but absolutely nobody will give a fuck about the iPhone 5. (Apple and Microsoft will probably not even exist any longer in 200 years. On the other hand, I'm pretty sure that the free software movement will be alive and well in 200 years from now, even if it might have been outlawed by then.)
Stallman is an eccentric personality who finds it difficult to relate to people and feels most comfortable around computers. I'd imagine that for him coding would be "the easy job", while taking on the role of public speaker and advocate for Free Software is probably a cross to bear rather than an escape from the hard work.
RMS coded GCC by himself - it was only later others got on board:
GCC history
And. of course, if it wasn't for RMS and GCC. Linus would not have been able to get a 'free' compiler for his project.
RMS is the seed of all of this. Don't knock him or his values. It is why we have a great 'free' OS (in all it's varieties) today.
Just to get things straight:
XNU (the Darwin kernel) utilises modules and message passing (signals), which is indeed a feature first pioneered in microkernels (by design). The current WinNT kernels do the same, and also call themselves "hybrid". Linux is almost there with modules and IPC signals.
All of this is, starting from a monolithic approach, 3% of the distance towards a microkernel (the 3% is arbitrarily thrown out, but you get the point: they are basically monolithic), and calling them "hybrid" is just trying to one-up everyone else.
And yep, OSX is nicely geared towards realtime, while in raw performance usually Linux is on top.
QNX is a real microkernel, but ad-hoc benchmarks ("real" benchmarks were never published) have shown its performance to be a fraction of modern operating systems. That's no problem though, as it has very low realtime latency, and that's what matters in its application area.
One of the most promising current microkernels is Fiaso.OC, a L4 fork I've been dabbling a bit with. It reaches 5%-50% of the Linux performance throughput in some classical applications, but can be faster in certain realtime scenarios.
Catching up to the last in the race is no achievement.
Wrong - catching up with the last in the race is a great achievement - you've just managed to bypass the rules of logic.
Or you're a whole lap ahead!
So if you're reading a guys post, and it shows that he created a Slashdot account in the 1990s, but since then, he hasn't been able to add even a basic amount of value to a modern thread, do you reply to him?
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
Seriously, the Hurd guys either need to get with it, or just quit. It is stupid to have something this completely out of date and keep pretending like it'll be relevant. No, if you want your kernel to have any chance at relevance it needs to support modern features. Yes, that means SATA, x64, and so on. None of these are new things, by any stretch of the imagination.
If they lack the resources or drive to get this kind of thing done in a timely fashion, then just let it go. There is no point to releasing a kernel 10+ years out of date (as the parent points out, SATA hit in 2003) particularly when there are plenty of options that ARE up to date.