Re: [tlug] Peeling onions.

[ijw@example.com]

>>>>>> "Uva" == Uva Coder <uva.coder@example.com> writes:
>
>     Uva> I wonder what other possibilities are available if GNU was
>     Uva> dropped and we started with new compilers and used
>     Uva> different tools/utilities?

That could be taken in many different ways.

Do you mean that Linux systems use so much runtime GNU software
(shellutils, bash, textutils, libc etc.)?

This isn't true - rescue disks often use a small non-GNU libc and busybox.
 And there's always the option of using the BSD runtimes, which wouldn't
take that much porting.  I don't see how this would change anything,
though (apart from shutting up RMS about GNU/Linux, which is a laudable
aim in itself).


Do you mean that compiling Linux requires GNU software?

We'll ignore the commercial compilers for the moment, but even then TinyCC
clearly works:

http://fabrice.bellard.free.fr/tcc/tccboot.html

The only thing the kernel really depends on if you were to be actively
developing is GNU make.

> 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?

Pay money? ;-)

>     Uva> It seems to me that the GNU way of doing things restricts
>     Uva> Linux kernel development to some extent.
>
> 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.

Hurd's problem was that it was a huge project.  Also, while it may have
been well defined in the mind of the intiators, it was never very well
described to people.  In short, it was badly marketed, which meant that
few people were interested in joining the development.

Linux ran from the first release, and had a working compiler on it fairly
promptly, which made it far more interesting.

> Nor do I see the BSDs innovating the way Bell Labs has done, first
> with Unix then with Plan 9.  (And lots of other goodies in between.)

The secret to innovation, in my mind, is to have something so small that
you can rewrite it easily.  Linux is now too big.  The core bits that need
serious experimentation, such as the MM system, have a steep learning
curve.  Better to experiment with a different platform:

- L4 and EROS seem to be more interesting from a core-OS-out perspective;

- The HURD, for all its faults, is a great environment for experimenting
with user-level APIs and filesystems, because you can effectively replace
OS functionality.  It's nice to be able to write a new FS and not crash
the system, or experiment with authentication systems.  (Yes, Plan9 is the
same in this respect, but bastardising C was a bad idea and the licence is
crap.)

-- 
Ian.