Re: [tlug] Peeling onions.

[Uva Coder <uva.coder@example.com>]

On Fri, 04 Mar 2005 16:16:05 +0900, Stephen J. Turnbull
<stephen@example.com> wrote:
> 
> Could you be more specific?  My take is that Linux has never used the
> latest features of GCC, in fact Linux often broke because GCC
> implemented some optimization based on assumptions that the kernel
> violates.  So I don't really think that which C compiler is used
> matters.

Well, I've been studying the 7th to 10th edition Unix, Linux, Plan 9,
Inferno, and V Kernel. Perhaps I'm seeing patterns where there are
none; that is very possible. I need to take a break and revisit it
next week.

It's amazing that the benefits of each system haven't made its way
into a newer operating system. I know I'm off on a tangent again, I've
got my head in the clouds thinking about a Fair Share Scheduler (FSS)
today. Sorry.
 
> On the other hand, while you clearly can write operating systems in
> C++, Modula-3, and Lisp, where are you going to get the
> reasonable-quality implementations except from GNU?

From Ken (Thompson) C, it's open source.

> So I don't see the toolchain as mattering that much, nor is it really
> replaceable.  Note that all the *BSDs gave up long ago.  I just got a
> flyer for RHEL on Opteron, they're selling some fancy compiler suite
> with it---and the UI is 100% drip-in compatible with GCC.

Yeah, but I see GRUB and LILO as being train wrecks. I only see them
as getting worse.

> I don't see how.  I think it's Linus who is the bottleneck.  Just
> compare the HURD with any of the actually useful OSS kernels; you'll
> see that "the GNU way" kills kernels.

I view the GNU way as a more restrictive paradigm than other closed or
open source projects. I don't see GNU as a coherent foundation where
all applications can share resources through a system protocol (i.e.,
P9 protocol). From my understanding, even V Kernel did the same thing
to allow for process migration. GNU addresses neither. But that is
just the tip of the iceberg.

I won't bore the ideas about namespaces and globally sharing system
resources. As for sharing data, the popular things nowadays are nfs,
scp, ftp, and http; how sad an existence we live. I believe we can do
much better. Why not a single protocol that extends across all
boundaries to access data the same way, securely and safely across
networks? A file is a file, right? If you are concerned about opening
a file the protocol doesn't do that but a utility.

I'll end here. I'm sure folks are getting tired of my ideas. 

-uva