karl@??? writes:
> Rainer Weikusat:

[...]

>> For religious reasons, I refuse to use busybox for anything. I spent
>> some years working with and fixing this code for an embedded system at
>> the time when these were still fairly small. The corresponding product
>> (not project) is long dead, hence, I'm rid of the associated
>> problems. Unless I'm forced to, I'll never again touch this code(base).
>
> Ok, I havn't seen thoose problems in my embedded things, but embedded
> now can host regular devuan, and also the busybox situation might have
> improved.
>
> ///
>
> To solve this at a dist level, we need to assign people to maintain
> . a separate kernel (not needing initrd) for a specific group of systems
> . a slimmer init
> . a fsck.ext4 (perhaps static)
> . a mount (perhaps static) that makes it possible to mount /usr
> . hwclock possible run after mount -a
>
> Sound interesting ?

Not to me. As I already wrote: My idea how to deal with that, based on
the software I want use, would be to create a package which turns this
from a hard into a soft failure, ie, instead of ldd bombing out with
libpcre not found when the kernel tries to execute init, this should
print a warning (or do nothing at all) when booting on a system without
SELinux support (preferably, one package to handle all programs which
have acquired a SELinux dependency in the meantime). If that was
interesting to someone but me, I'd be willing to put some work into it
(in the little spare time I have).

How general this could be made would need to be determined (ie, if the
idea is workable at all, which might not be the case). As the answer to
the 'Is someone interested ...' in question would seem to be no, I'll
let the matter drop.