Craig, Thanks for this.
I think there is another route to achieving something similar: procps could employ dpkg-divert(1) to avoid a file conflict on /usr/bin/pidof with sysvinit-utils with sysvinit-utils still continuing to ship its own implementation. I see this approach as having some benefits, in particular for some non-systemd and initless installations. In such cases, having to add a dependency on procps at about 2.2M just to provide /usr/bin/pidof appears materially worse than using sysvinit-utils with an installed size of about 100k. Obviously, if a non-systemd or initless installation requires procps then that pidof implementation would then be used. In my opinion there is little to choose between the implementations and I emphasise that I don't particularly mind which is used. My real concern is how it is packaged and the consequences of that, particularly in terms of installed size. I am grateful for your thoughts on this. Best wishes Mark
