Re: removal of kill?

[Eric Blake]

On 11/09/2015 09:14 AM, Pádraig Brady wrote:

>> I would be inclined to retain it one way or the other, since some people
>> surely require it, either because they choose not to, or simply
>> cannot, install an alternate package that provides it.
> 
> Maybe. It would be good to see example though.
> I searched for usage of gkill for example, but only
> found that that is used on cygwin as it provides kill
> that handles windows pids

Cygwin's own kill lacks some GNU features, but does indeed have extra
abilities related to Windows pids.  But I seldom hear of people using
gkill as an alternative to kill.  Since I maintain the Cygwin build of
coreutils, I'm in the position to state that the Cygwin community won't
be adversely affected whether we change kill to not be built by default,
or even drop it entirely from coreutils.

But whereas I can understand why we completely dropped su (which
requires non-standard interfaces to implement, and is highly
system-specific; plus it is a burden to maintain a security boundary),
it's harder to justify completely dropping kill.  Rather, I'd class it
in the same category as hostname, which is another coreutils app that is
not always built by default, because several distros (including cygwin)
prefer picking up the util-linux or other hostname variant, but where
keeping it in coreutils doesn't hurt because it can still be implemented
using mostly standardized interfaces, so it is not too much of a burden.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature