I agree with what you are saying but I think the issue that is important
is that 1.3.xx is not a release kernel so it is implied that if a user
decides to use it, he may experience problems like changes in the "API".
> I do not have any such problems, but sometimes my colleagues have:
> "It worked yesterday now it is broken." If Linux wants to be mainstream,
> it got to be less destructive. I have now converted an industrial group
I dont think it has be less destructive. The problem now is it is
mainstream enough to be causing the problems you are speaking of. Since
Linux is freely available, I don't think the kernel developers must
put historic compatibility ahead of improvement. Look at it as
growing pains.
> As in "kernel-space--user-space" interaction? Or as in API -
> APIs are sacred, let people know about changes IMMEDIATLEY if
> anything changes? I've written a device driver for a since
> deceased co-processor board (it WAS fast in 1993) and I have
> sevreral generations of patches to cope with API changes.
Agreed but I would rather have a tight kernel than have several
generations of obsolete support. When an API is broken, everyone
on the kernel related lists ARE aware immediately. The ones who
are not (I am guilty myself) are the ones who either lapse on
following the development or don't follow the list at all. Those
are also the ones who should not be using a development kernel.
They should wait until the distribution they are using is updated.
-Melvin S.