On Tue, 16 Jun 1998, Linus Torvalds wrote:
> and while some people have successfully bullied me into accepting a
> limited form of that into the kernel, I just don't want to have anything
> to do with anything like this. It's not something that the kernel header
> files should care about, because the kernel developers don't care about
> it.
>
> I think that kernel header files should be used for building the kernel.
> Nothing else.
>
> Linus
Ok, you want a blackbox, that is legal, hence no one should complain.
However, even a blackbox must have a well defined 'door' outside itself.
Until now _all_ of the kernel header files were your interface definition
for the outer world. Now as you abandon that, you should atleast have
'something' that defines your _exported_ interface in a clean relyable
manner, else user space people always would have to 'guess' what you
consider to be part of that interface.
So, why is that export header tree that Tytso and Richard proposed so bad?
It would not have to be linked in by libc, it simply could be copied.
I guess you need it for yourself as a place to look for what bits and
bytes _you_ must preserve to keep binary compatibility.
And no, I don't think you can delegate this work to someone outside the
'inner circle'. Its _your_ job to define the interface, regardless wether
its comfortable for you or not ;-)
Hans
<lermen@fgan.de>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu