> Jon M. Taylor writes:
> > On Sat, 23 Jan 1999, Richard Gooch wrote:
> >
> > > Jon M. Taylor writes:
> > > > On Sat, 23 Jan 1999, Richard Gooch wrote:
> > > > > I recall a quote from Dave Miller a year or two ago, when there was a
> > > > > thread about monolithic vs. microkernels. He pointed out that we
> > > > > haven't even come to the limits of monolithic kernel design, so
> > > > > considering monolithic kernels like Linux as "obsolete" is highly
> > > > > premature.
> > > >
> > > > I do not think that blindly adhering to an obsolete design until
> > > > the last scrap of performance has been wrung from it when a clearly
> > > > superior alternative is available and free is a very wise use of
> > > > programmer resources.
>
> This sentence above has the tone of you telling others what they
> should be working on. I would suggest that adding some words like "but
> I hasten to add that I'm not telling people what to write" would have
> avoided a negative interpretation.
When in doubt, never assume a negative connotation of something
ambiguous someone says. Especially on the net, where less emotional
subtlety makes it through.
> > > Rubbish! This is what evolution is all about! Species do not instantly
> > > become extinct when a better one (occupying their niche in the
> > > ecology) comes along.
> >
> > Of course not.
>
> I'm glad you agree with me.
>
> > > They become extinct when they are driven into
> > > the ground by superior competition or a changing environment.
> >
> > Sure.
> >
> > > They
> > > hang on until the bitter end.
> >
> > This is where the analogy breaks down. OSes are not life
> > forms. They exists to serve our needs, cannot reporoduce without
> > our help, and the form of their genetic code, so to speak, is
> > entirely within our control. I see no reason whatsoever why a
> > *free* OS should hang on to the bitter end.
>
> 1) You can't control volunteer work
It is not all volunteer work now and is likely to become less so
with time as big money moves into OSS. The small fry will continue to
innovate at the fringes everywhere, but *relying* on volunteer labor for
support of needed kernel features is not a business option.
> 2) while people are interested in working on something, let them, no
> matter how misguided you think they are. They might come up with
> something really interesting. Their continued efforts are not
> necessarily unwise
I never said otherwise.
> > > Sometimes the "obsolete" species makes
> > > an evolutionary jump changes the world forever.
> >
> > Then it is not the same species anymore. Just because FluxOS
> > is made up of Linux sausage does not make it Linux, even if you call
> > it Linux. By that token, I guess "Linux" will be around forever as
> > long as there is an OS that goes by the name of "Linux". But this
> > is semantic nonsense. When I say that Linux will be obsolteted by
> > another OS within two years, I mean more-or-less the current
> > monolithic POSIX-compliant Unix clone architecture of Linux 2.2 (and
> > Linuxly also 2.3/2.4 as well).
>
> Linux may make an evolutionary jump that gives it a tremendous
> advantage over (possibly then superion NGOS's), *while remaining
> monolithic and POSIX-compliant*.
Specifics please, and info as to why this design would be any
better than other NGOSes out there. I am basing my predictions on OSes I
know exist today, with running code in some form.
> I'm not saying it will do this, but
> it *may* happen. Saying that Linux (monolithic and POSIX-compliant) is
> doomed to obsolescence is an unprovable statement.
Sure, you can only make a probabilistic statement. I think it is
very *likely* that within two years, one of the currently-existing NGOSes
will be a competitior to Linux in some area of computing. In fact, that
is already happening today. Flick, the high-security cousing of the Fluke
PVM kernel, was developed in a DARPA-sonsored project. The military is
interested in Flick because it is an interrupt-model kernel which,
provides an inherently cleaner and more watertight design than a
traditional process-model kernel like Linux.
> > > One of the great strengths of the Open Source movement is that it is
> > > driven by evolutionary development. To say that development of some
> > > project should be terminated because you think some other project is
> > > better misses the point.
> >
> > I never said that anything should be terminated. Please do not
> > put words in my mouth.
>
> As I showed above, that was the tone of what you said. You didn't say
> so explicitely but you implied it.
I'm sorry if you feel that way. I certainly did not intend that.
> > > Besides which it's arrogant. Volunteers are
> > > free to work on what they like: you have no right to tell them what
> > > they should be working on.
> >
> > I didn't do that! Why are you accusing me of this? All I ever
> > said was that I thought that this would happen as part of the natural
> > evolutionary course of free OS development.
>
> See above. People are not robots, and they respond to the tone of the
> message as well as the explicit content. I'm glad you've made you
> position clear in the above sentence. I just think that your original
> message could have been less inflammatory.
Fair enough. I think I have a little Bruce Perens in me sometimes
|->.
> > > Why not forward your troll to the *BSD mailing lists after doing:
> > > s/Linux/BSD/g
> > > s/NGOS/Linux/g
> > >
> > > and see what a warm welcome you receive.
> >
> > You need to maybe proof your replies once before shooting them
> > off, hm?
>
> I'm happy with my message as it stands. But thank you for your
> concern. I would ask, however, that you refrain from a condescending
> tone.
Yeah. That came off kinda bad. Sorry.
> > > > > Much later, with 2.2 on the doorstep, I pause and consider his
> > > > > words. 2.2 has brought us many performance improvements (networking,
> > > > > dcache, SMP).
> > > >
> > > > Few of which are specific to the Unix API, and most of which could
> > > > be pilfered for use in an NGOS easily. Exactly why is this relevant?
> > >
> > > Read what I said again. The point I was making is fairly obvious: the
> > > Linux kernel shows no sign of a slowdown in development. Your claim is
> > > that Linux is, or will very shortly be, obsolete.
> >
> > Not *very* shortly. More misstatements.
>
> Let me clarify. By very shortly, I mean in the timescale of a year or
> two. That being the same timescale you expect Linux (monolithic and
> POSIX-compliant) to become obsolete.
That isn't 'very' short by the standards of the OSS development
world. And things are only going to get faster now that the OSkit is out.
Maybe I didn;t play that point up enough, but that's what I mean. You
should see the number coll, wacky research OSes that have been built on
top of the OSkit in a very short (by your standard) timeframe. It is
pretty impressive, IMHO.
> > > To be painfully
> > > explicit, I refute your claim by drawing attention to the continued
> > > development of Linux. Until people run out of ideas, it will not be
> > > obsolete.
> >
> > People have not even run out of ideas for horse-and-buggy
> > contrivances. Are you saying they are not obsoleted by the
> > automobile?
>
> Read further, and will see the connecting sentence.
>
> > > Of course if 90% of Linux users switch to something else, then you
> > > also could consider it obsolete. But I doubt that will happen while
> > > new ideas keep coming.
> >
> > The buggy vs. car analogy appies here too.
>
> I think 90% of buggy users switched to cars before my father was
> born. So, yes, the analogy applies. Linux has not seen 90% of it's
> userbase switch to a NGOS kernel.
>
> Let me be more explicit: *if* 90% switch from Linux, *and if*
> developers run out of ideas, then I would consider Linux
> obsolete. I'll go futher, if 50% switch to a NGOS kernel and if Linux
> kernel developers run out of ideas (to compete with NSOS's), then the
> writing will be on the wall.
> BUT: while linux Kernel developers keep having new ideas, I don't
> expect to see a mass exodus, because it's likely that Linux
> (monolithic, POSIX-compliant) will be performance competitive.
It isn't an issue of performance, it is an issue of capabilities.
Hardware keeps getting faster, and I doubt any competitive NGOS would
exhibit a large enough performance gap with Linux to have it be a factor.
Look at Java. It is proof that a good idea can come along and cause a
mass exodus if it offers clear benefits.
> And, like I said above, Linux may yet evolve, remaining monolithic and
> POSIX-compliant and still compete with NGOS's. I would not be so bold
> as to make a prediction either way.
I would. Linus has said many times in the past that he doesn't
see a case for microkernels, which in some form will be the foundation of
most NGOS designs.
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/