Bug#752075: daemontools-run: Add systemd support

[Gerrit Pape]

On Mon, Jul 28, 2014 at 02:36:06PM +0200, Michael Biebl wrote:
> Since you are going to do it "your way" anyway I'm not sure why you ask
> for feedback from us?
> 
> It's like talking to a brick wall.

Hi, I'm sorry that you feel this way.  I also found our communication
far from ideal, and after I refused to add a dependency to the
daemontools-run and runit packages, it failed completely.  To me this
feels like do it your way, or no way.

Since you're the specialists, there's still hope to get feedback other
than
On Tue, Jul 22, 2014 at 03:03:02PM +0200, Michael Biebl wrote:
> There is a reason, why we added the logic in a single place.
On Tue, Jul 22, 2014 at 06:11:24PM +0200, Michael Stapelberg wrote:
> I'm not sure how we can state this more clearly:
> Use dh-systemd or you _will_ have bugs sooner or later. It is the most
> simple implementation that we could come up with.

There also are reasons for my design choice.

> There are still various issues with the maintainer scripts code [1]

Please tell me which ones.

> [1] aside from the fact that negative clauses make it unnecessarily
> hard to read.

The fact is that people are different.  To me this is most intuitively
to read in set -e scripts.  When does this script stop?:

set -e
true && true
false && false || true
false && false
false && true || true
false && true
false && true || false
true && true || false
true && true && false
true && false && false

> and I see you dropped the .path unit again?
Yes.  I don't understand the question.

On Tue, Jul 22, 2014 at 03:18:02PM +0200, Michael Biebl wrote:
> What about Ansgar's improvements/suggestions [1]?
> E.g. adding a Documentation= header, directly using svscan, etc?

Thanks to Ansgar, I looked at them but currently don't want to do "a bit
more integration into systemd".  KISS and provide the very same software
experience to users of these packages under sysvinit, runit, and
systemd.

Regards, Gerrit.