Bug#911165: debian-policy: drop requirement to ship sysvinit init script with same name

[Michael Stapelberg <stapelberg@debian.org>]

Hi,

I have recently been made aware that this policy is still around when
someone filed a bug asking for a package to be made compliant. I had
assumed that the requirement would have been dropped by now, so let me
echo/add a few thoughts.

When I have previously voiced discontent about the work required to
support multiple architectures in Debian, porters have reassured me that
they would help with that. In the same spirit, I have effectively been
relying on contributions for sysvinit support in all of my packages.

It was almost 10 years ago (!) that I have first read about systemd. I
have been using systemd exclusively for many many years. My working
knowledge of sysvinit has degraded over the years, and I have no
motivation or time left to even test sysvinit support in any capacity.

At this point, there are many computer users who have not ever used
anything but systemd. I worry that _demanding_ (instead of encouraging)
sysvinit support via policy is too much effort for package maintainers,
for very little practical gain. I can say with certainty that my time is
better spent elsewhere, and I think many others feel the same way.

My personal suggestion is to change policy to require support for
Debian’s most relevant init system and architecture combination
(i.e. package maintainers should ensure systemd/amd64 works). Anything
beyond that should be encouraged, but kept optional, so as to unburden
the package maintainers.

-- 
Best regards,
Michael