I completely understood that. My point was that if the utilities are seen
to be sufficiently close to the kernel, we can assume that it will be less
than five years before the older versions are no longer being used with new
kernels.
For example, every distribution ships with lilo and ?fdisk as part of the
base package. The number of people who need to upgrade their kernel but who
still keep an old LILO for use with it is, I would submit, very small.
>We should be extremely slow with breaking old stuff.
>After all, we claim that Linux is a stable environment -
>not only in the sense that it does not crash, but also
>in the sense that it is not necessary to install the
>latest versions of everything every three months.
I don't think we are "breaking old stuff" too badly. There's already code
in setup.S to check who/what loaded us, so presumably this kind of exercise
has been done before. Sure, it should be phased in, but I don't think it's
beyond reason to fix LILO and FDISK, let the developers of other utilities
know where the library is so they can get _correct_ disk geometry (many of
them will be pleased to be able to incorporate your experience with disk
layouts into their programs), and announce that 2.3.mumble will be the last
version that can install a _later_ kernel with on old LILO and without
specifying linear geometry.
All software development is a tradeoff between functional advances and
backwards compatibility. LILO and FDISK are so fundamental that I reckon
they can be counted as almost part of the kernel.
I guess the decision to take functionality out is not something one takes
lightly (read: without Linus). But if it has to be done, five years from
tomorrow is nearer than five years from next month, I suppose.
Nick Brown, Strasbourg, France (Nick(dot)Brown(at)coe(dot)int)
__________________________________________________________
email address updates : @coe.int replaces @coe.fr
for more information, http://dct.coe.int/info/emfci001.htm
__________________________________________________________
-
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/