> I do not have the same programming record as Derek Atkins, but I modestly
> agree. Each major version drifts more. 1.0 stopped at 1.0.9, 1.2 at
> 1.2.13, now we are at 2.0.30 and still waiting for a new release since
> the whole ISDN stuff stopped working in 2.0.30!
I think we still have to get the engineering of "stable" releases right -
the whole 2.0 series has been a bit of a mess.
The general belief was that the stop-dead approach that was used with 1.2
wasn't too good, because 1.2.13 was starting to look very tired by the
time 2.0 was released. On the other hand, I agree that the "cure" of
continuing development on 2.0 does seem to be worse than the original
disease. Any change that breaks binary compatibility is a definite no-no
in a stable kernel - I can't think of any reason sufficiently compelling
to make such a modification. Performance tweaks can always be released as
seperate patches against 2.0 (like the ISS stuff was originally) but they
ought not to go into the mainstream kernels if they will have any impact
on compatibility.
p.