Bug#826215: Bug#810018: procps pidof (was: Re: Processed: forcibly merging 851747 826215)

[Mark Hindley]

Craig,

Many thanks for the time and effort you have invested in this.

On Tue, Jan 13, 2026 at 10:09:38PM +1100, Craig Small wrote:
>    On Tue, 13 Jan 2026 at 22:04, Luca Boccassi <bluca at debian.org> wrote:
> 
>      The second point is correct
> 
>    Excellent, so a systemd init host doesn't need pidof then

Really? Pidof is used in some maintainer scripts. See dbus[1], for example.
Surely pidof remains Essential until all such usages are identified and Depends:
whatever-pidof-provider added?

I think it is also worth considering that the 2 pidof implementations are not
completely identical. A brief comparison of the manpages suggests that the
separator option is different (-d vs -S) and some of the other options only
exist in 1 or other implementation.  If they are not direct drop-in
replacements, how do we avoid/handle breakage for callers/users?

Please be assured that I am not averse to finding a solution to this, but it is
a difficult issue requiring considerable work for pretty marginal gains. At the
moment, I don't see just removing Essential: yes from sysvinit-utils, shipping
pidof from src:procps' and adding the usual Breaks/Replaces as being a complete
and adequate transition plan.

With best wishes

Mark



[1]  https://sources.debian.org/src/dbus/1.16.2-2/debian/dbus.postinst#L46