On Saturday 26 Feb 2005 11:55 am, Senthil Kumaran wrote:

...

I would like to know about GNU/Hurd ... Does anyone of you have GNU/Hurd cds?

i think its dead

Nope., its very much alive and it runs on my Laptop. :)

On Sun, Feb 27, 2005 at 11:39:33AM +0530, Nagarjuna G. wrote:

On Sun, Feb 27, 2005 at 05:05:53AM +0100, Joe Steeve wrote:

...

...

On Saturday 26 Feb 2005 11:55 am, Senthil Kumaran wrote:

...

I would like to know about GNU/Hurd ... Does anyone of you have GNU/Hurd cds?

i think its dead

Nope., its very much alive and it runs on my Laptop. :)

HURD is not dead, it is shaping up slowly in the womb of developers. We dont call an embryo dead, do we. Its a complex baby, gestation period is more. This period can be reduced if more developers join. If you are good in C contact anang@gnu.org.in.

the above address should be ab@gnu.org.in (Anand Babu). Please correct.

Nagarjuna

Kenneth Gonsalves lawgon@thenilgiris.com writes:

ok, maybe it is still just alive - but its soon gonna be dead - and no number of C programmers can revive it

Ok., i dont wish to start a flame war here. But technically ( any OS researcher would agree ) the Hurd is a much better design than the Linux kernel. The Hurd is `Extensible` and is capable of doing things which cannot be done in a clean (not dirty hacks) way on the Linux kernel. The Linux kernel is a step back in technology., because of its Monolithic design. Microkernels had their bad time., but not anymore. The future is there. Sticking with a legacy system is not a good idea.

And morover discouraging its developers by derogatory comments like above is not a very good thing. The developers of the Hurd are professional people who are playing with the best in OS technology. They ought to be respected.

The GNU/Hurd is very much usable. What lacks is support for drivers in the GNUMach microkernel. And the development of the GNU/Hurd is slow because of the lack of developers. Work is underway to make the Hurd run on the L4 microkernel which is more advanced than the GNUMach. The GNU/Hurd project needs people to test applications on it. You can basically recompile any POSIX application on the GNU/Hurd., but sometimes there are some issues.

So _definitely_ the Hurd is not a dead project., or a to be dead project., or whatever. It was started late., it did not gather momentum largely because the masses were happy with the `Linux kernel`.

Kenneth Gonsalves lawgon@thenilgiris.com writes:

i was just echoing linus torwalds comments on hurd. What he said is that it is over engineered and hence although it looks great on paper, it can never be properly implemented.

Linus maynot agree., and a lot of `Linux` fanatics may not agree.. but., its a known fact in the OS research area that a monolithic system such as the `Linux` kernel would obsolete itself soon. With the processing power we've today, we need a lot more features than what the Linux kernel provides.

Supercomputing is a area where Monolithic systems are just not suitable. `Distributed OSes`, are far more advanced than Clustering. Future supercomputing wont be based on `clusters`., they'll be based on distributed OSes which make a network of machines look like one huge machine at runtime. The OpenMosix project makes a network of Linux kernels into a Distributed OS., however its a hack. On the Hurd, a distributed system is a feature. The advantage is because of the superior microkernel technology underneath. In 1990., with the meagre processing power, talking about message passing might have been absurd. But today, when both processing power and microkernel technology have grown exponentially, sticking to old theory is kinda silly. It is basically limiting ourselves when we can see far better things.

Which is why very few people are interested in developing it.

The developers are not very easily attracted to the Hurd basically because of lack of awareness. When people know `Linux` and not `GNU/Linux`., obviously they never get to know about the `GNU` system., nor the GNU kernel `Hurd`. Just talk to `Linux` enthusiasts about calling `Linux` as `GNU/Linux`., they'll sneer at you. They are happy with what they have. I dont know how they'll react when they realize that their `beloved` system would be obsoleted.

On Tuesday 01 Mar 2005 12:38 pm, Philip Tellis wrote:

Sometime Today, JS cobbled together some glyphs to say:

...

Supercomputing is a area where Monolithic systems are just not suitable. `Distributed OSes`, are far more advanced than Clustering. Future supercomputing wont be based on `clusters`.,

snip

One question. Let's agree that the OS is GNU (as in the operating system named GNU) and for the time being it uses the kernel named

Linux.

So, we call it GNU/Linux. However, we all know that for the end

user,

the OS alone is not sufficient. End users want windowing systems

and

graphics and mouse control and multimedia and what not. These are,

of

course services provided by the X server and windowing system

through

direct kernel calls in most cases. The entire operating system (ie,

the

software that makes the computer system operational) from the end

user's

point of view is really X11 (or X.org)/GNU/Linux.

In today's age though, one also needs pretty graphical email and web clients, so we add Mozilla (because Opera isn't Free). When I say Mozilla, it would encompass Firefox/Thunderbird if that's what you prefer. The operating system is now Mozilla/X/GNU/Linux, and I

think

that's fair.

In fact, we could prolly do away with most of the GNU part. We need glibc, the dynamic linker and loader. No need of a compiler, no

need of

a shell (unless you have shell scripts). You'll prolly need a

python

and perl interpreter, but I can't really think of anything else

(correct

me if I'm wrong). So the operating system (again, from the end

user's

point of view) is now Mozilla/X/gnu/Linux.

I'd like to state again, so that there is no confusion. I'm looking

at

the operating system from the end user's point of view, and not from

the

computer scientist's point of view. I am (or at least have been) a computer scientist.

X, Mozilla, Apache as also important projects and equating them to the core GNU system is to confuse the issue. This is not a new argument. What this argument misses is that none of these appendages can be compiled or distributed as freely as on GNU systems without depending on commercial libraries. You should think of the situation where X, Mozilla, Apache, or for that matter Linux kernel get distributed without the GNU system, and get compiled with Borland, Visual Studio etc. Then people will relaize that GNU is not just an appendage, it is indispensable and others cannot even be implemented as free software in the real sense without GNU. Others are appendages, kernel and GNU are not. That is why there is no need to go on appending p/q/r/s/GNU/Linux. GNU + Linux is enough, and FSF's claim is legitimate.

So the point is, why dont people become fair and credit an indispensable contribution of FSF?

BTW, I also dont think that the kernel Linux will become obsolete. It is also getting adaptable to different environments. Another argument why Linux kernel will not be obsolete is because, history shows that bad technologies do not disappear just because they are technically bad. If that were the case vulnerable M$ would have perished long ago.

-- Nagarjuna

Philip Tellis philip.tellis@gmx.net writes:

GNU is enough. Adding linux to the end of it only serves to give credit to Linus Torvalds for doing no more than posting his code to usenet. However, if you are going to add Linux to the end of the GNU operating system, then you must add all the others as well. Give credit where it is due.

<...>

It's not crediting the contribution of the FSF. The FSF set out to write an operating system called GNU. They did that, and that's why we call the operating system GNU. Why add anything to it?

Nope., they did not yet. The `GNU` system was engineered to have the `Hurd` as the kernel., and Linux is just a temporary fit. In fact glibc was originally written to support the Hurd but was re-engineered for the sake of the Linux kernel. I think the FSF funded it., I'm not sure though.

One cannot call the `Debian GNU/Hurd` distribution as _the_ GNU system. Its debian's variant of the GNU system. All GNU/Linux distributions are simply variants of a `GNU/Linux` OS which is not _the_ GNU system. The GNU system is still under development. It'll have the Hurd as its kernel. It may not have the legendary `rc` scripts., talk is on to have a GUILE extensible startup program to start the system.

You'll find better info at gnu-system-discuss@gnu.org.