Re: [tlug] Just curious... how much impact does a kernel update make?

["Stephen J. Turnbull" <stephen@example.com>]

Ian Wells writes:

 > There were significant differences between 2.4 and 2.6 by the time
 > they'd finished.

There were significant differences between 2.4 and 2.6 before 2.*5*
was even forked.  That is, Linus knew that there would be substantial
ABI incompatibilities (GNU_MODULE or whatever the "I'm a GPL module"
thingee is, for example) and reorganization of things like the kernel
module hierarchy on disk.

 > 2.6, however, has shown that adding architectures, drivers,
 > filesystems and schedulers doesn't seem to cause enough change to
 > bother with bumping the major version numbers.

That's sort of known a priori, isn't it?  What we've learned from 2.6
is that the POSIX ABI plus Linux extensions already in 2.6 is flexible
enough to accomodate those things efficiently without additional extension.

 > Unless someone needs application binary incompatibility, meaning
 > all your applications would need recompiling and retesting, I don't
 > see why they'd go for 3.0 (or even 2.8).

2.8 could happen if the vendors who are currently clamoring for
stability find a few ABIs they need that aren't in the current kernel
and can't be reasonably handled by current extension mechanisms.
We've learned from sad experience that backward compatibility is just
a euphemism for incompatibility in many circumstances.

Putting on my pointy Nostradamus cap with stars and moons and wizardly
signs, I predict that Linux will go to 3.0 when the threading problem
(ie, theoretically safe and pragmatically efficient use of shared
state) gets solved.  I'm willing to bet that will involve new
primitives, possibly including a new version of POSIX (or SUS if you
prefer that terminology).

I could imagine other such events, but I'd have to do some research.
Ie, don't go buying stock because I said that. :-)  Seriously, after
the discussion with Curt re: STM in Haskell, I think this one could
happen soonish, and it seems like a solution is on the horizon.  The
fact that even so nobody but SPJ really understands that stuff
(specifically, I doubt very many kernel hackers do, and the way that
Linus has been beating up on some senior lieutenants recently suggests
that certain forms of brain damage may pervade the hierarchy -- slb,
you really should post those gems to a wider forum, they're quite
amusing) suggests that it is *very* unlikely that POSIX has the tools
to handle it, and Linux really is a very POSIX system.