after Linux, what? in place of Hurd, Eros, Brazil,...?

[cbbrowne-HInyCGIudOg at public.gmane.org]

Toomas Karmo wrote:
> I hereby recant. It's now NOT clear that Hurd is the Next Great Thing. 
> 
> At this point I'd like to ask whether anyone can peer into a crystal
> ball and see what the Next Great Thing might be. Rob Brockway has
> referred in passing to Eros, Brazil, and Plan9 as interesting
> operating systems. Do any of these have passionate backers? Is anyone
> on the listserv keen to wade in and DEFEND Hurd?

I do not see any of these being the Next Great Thing.

- Eros is interesting, but its _proper_ use involves designing embedded
  systems atop it, things that aren't much "like Unix."  And while it
  has the "Eric Raymond Seal Of Approval" on it, I haven't seen anything
  more than talk about it.

- Brazil would be a proprietary Sun thing, and it is by no means evident
  that Sun will survive to make it more than a curiosity any more than
  they have had success creating useful "JavaOSes."

- Plan 9 is a proprietary Bell Labs thing; while interesting things
  could be gleaned from it, and have (Alex Viro's filesystem namespaces
  patch, for instance), it has quite clearly failed in becoming of
  general interest.  They dropped efforts on Plan 9 in favor of Inferno,
  which has considerable parallels to JavaOS and Brazil, but that, too,
  has pretty much fallen to being a curiosity.

These systems are now all years old (save perhaps for Brazil), and none
have the "spark of youth" that would be needed in order for a system to
burst into flame so as to "take up the torch" as a successor.  Hurd
suffers from the problem of dependancy on pretty old hardware; these
systems do too.  The last time I looked at Plan 9, they _refused_ to try
to support Adaptec SCSI adaptors, which meant I couldn't install it on
the hardware I had then.  This sort of fragmentary support for hardware
is quite typical.

The system I previously pointed at as a more plausible "successor" was
Dragonfly BSD, which is mostly a spinoff of FreeBSD.  I can't claim that
is a perfectly natural successor, but it has _some_ merits that I would
contend are necessary to a would-be contender:

 1.  It is a fresh project, attracting interest.

 2.  They have a strategy that permits moving much/all of the existing
     Unix software to run on it.

     Given LIBC, hierarchical filesystems, and an architecture that can
     be hoped to use to run XFree86, you can hope to run software from
     the FSF, GNOME, KDE, based notably on the wide array of stuff
     available in BSD Ports.

 3.  Unlike the other would-be successors, the intent is NOT to create
     something un-Unix-like.

     Hurd and Eros and Brazil are conspicuously steps _away_ from Unix.
     (Plan 9 isn't; it is intended as "more Unix-like than Unix...")

I don't think that people will necessarily leap from Linux to Dragonfly
BSD unless two things occur:

 1.  Linux kernel development stalls and/or goes wrong.

     There certainly have been challenges in the stability of the
     development process.  

     - There's near-open-warfare over the use of proprietary SCM
     software (BitKeeper) which frequently leads to major flame wars.
     There are people who cannot contribute to the Linux kernel because
     they are also involved with writing competing SCM code.

     - Linux has jumped through about three sets of memory management
     code since 2.3, and it is not evident that the major surgery is
     anywhere near completed.

     - There were plans to completely revise the IDE code; that got
     rolled back for 2.6; they'll retry for 2.7.

 2.  The choice of distributions gets too painful.

     MandrakeSoft's filing for bankruptcy wasn't particularly good for
     that distribution.

     "Red Hat Linux" has had the triple challenge of:
      a) Too many bad "dot zero" releases that were so "bleeding edge"
         as to cause considerable bleeding;
      b) The increasingly-proprietary "upgrade network";
      c) Now, they are dropping the former sequences in favor of
         having (pricey) RHAS/RHES that target stability alongside
         "Fedora Linux" that is likely to be, if anything, MORE bleeding
         edge than before.

     Debian has been having considerable pains in trying to release new
     "stable" releases.

     SuSE has gotten steadily more proprietary since about version 7.

It is by no means implausible that challenges of availability of updates
of all these sorts may lead people to look for something new.  It _is_ a
good thing to have the various *BSD systems around; they represent
meaningful "escape routes" for some of the problems that are possible.

(In addition, the mere _existence_ of "escape routes" discourages
corporations that control Linux distributions from doing anything _too_
dramatically stupid.  I have long argued that the existence of Debian,
complete with 'constitution' and 'democratic process,' however
imperfect, keeps away the reality the more paranoid fantasies people
have had about Red Hat.)
--
output = reverse("ac.notelrac.teneerf" "@" "454aa")
http://www.ntlug.org/~cbbrowne/languages.html
Signs of a Klingon Programmer - 3. "This  machine is GAGH! I need dual
Pentium processors if I am to do battle with this code!"
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml