lkml.org 
[lkml]   [1999]   [Jan]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: User vs. Kernel (was: To be smug, or not to be smug, that is , the question)
On Thu, 21 Jan 1999, Guy Sotomayor wrote:

> On Thu, 21 Jan 1999, Jon M. Taylor wrote:
> > That is IMHO a bad idea. Linux, because it follows the
> >traditional Unix monolithic kernel model, is already so obsolete that it
> >really isn't worth putting serious time into incompatible API
> >improvements. Its value right now is that it is a free, well-written and
> >POSIX-compatible open source Unix clone. Please leave it as such. Just
> >treat Linux as what it is: a legacy-compatible *interim* OS that provides
> >a nice, mostly-stable and well-known environment in which to develop
> >device drivers and OS guts (and apps, GUIs etc too).
> >
> > Soon, probably within a year or two, a decent, open-source next
> >generation OS like Hurd (Microkernel), FluxOS (polymorphic virtual
> >machines/nested processes), Dolphin (exokernel), The Cache Kernel, or
> >another NGOS will be developed to the point where it can step in and
> >cannibalize the guts out of Linux to produce a fully-functional NGOS which
> >is suitable for mainstream use. With the Univerity of Utah OSKit having
> >been released, this process is well underway and accelerating rapidly. At
> >that point, then, we all can dump 25+ years of accumulated Unix API
> >brokenness and move on to a clean, modern, well-designed OS.
> >
> > Linux will then exist only as some sort of NGOS legacy
> >compatibility wrapper, like WINE, and the real Linux can gracefully fade
> >away into history, its "springboard" role having run its course. This is
> >the natural way of things. If you want to "fix" Unix/Linux, do it right
> >and rework everything ground-up from first principles. Just MHO of
> >course.
>
> I don't want to start a flame war but I have to comment on this...
>
> I've been the part of a couple of NGOS projects at a previous employer. The
> fact of the matter is that the OS guts don't matter to most folks except us
> techno-geeks. What folks want are applications that work and good hardware
> support (ie device drivers). I have seen superior OS designs (and
> implementations) wither away and die because there was not a critical mass of
> applications and device support. Without *both* of those an OS is just an
> interresting footnote.

Which is why you have to have something like Linux first. The
Hurd languished because its creators tried to do the drivers and the
common algorithms at the same time as the NGOS design. The OSKit is
essentially linux and freebsd source all chopped up into components with
COM interfaces and mixed together. This allows good NG designs to be
implemented quickly and optimized over time.

> The reason that Linux/Unix has been around for 25+ years IMHO is because it has
> had a reasonably stable API set so a large body of applications could be
> developed. Given how long Unix has been around I don't expect Linux to
> disappear in short order.

Me either, but it will happen. The mores successful NGOSes are
not desireable for nothing. If they provide a clear and present benfit,
they will be used.

> Also remember that Unix did not spring into being with its existing API set.
> Many APIs have been developed as lab prototypes that have stood the test of
> time and become mainstream features that noone would consider not having. Like
> or not the Linux API set is evolving if for no other reason to support new
> types of devices. As new device paradigms become more standard we may see new
> APIs to allow them to be better exploited.

I'm not talking about API particulars, I'm talking about the
fundamentals of the OS design. A nested process/PVM OS is nothing like a
traditional Unix kernel.

Jon

---
'Cloning and the reprogramming of DNA is the first serious step in
becoming one with God.'
- Scientist G. Richard Seed

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:49    [from the cache]
©2003-2011 Jasper Spaans