Re: s6-rc blocks when oneshot service is still running

Alex Grabowski <[email protected]> writes:

> On 5/16/26 05:11, Alexis wrote:

>> Alex, have i got that right?
>>=20
>
> Short answer=C2=A0kind of yes.
>
> This email thread got me thinking about coupling between=20
> different
> programs in a system, that "script" is quite tightly coupled=20
> with
> program that starts it ("A"), because "A" needs to display some=20
> GUI
> confirmation about progress so far and is blocked by the=20
> script. So it
> is better to think about "script" as a part of that service "A"=20
> and
> not as a separate service, but indeed it provides a service (ie.
> something becomes available on the device (related to security)=20
> when
> the script is done).

*nod*

Yeah, part of the issue is that in casual use, the sort of=20
distinction i made isn't necessarily (or even usually) used. E.g.:

  "metalog provides a logging service"

And metalog can do so by being run directly, without a 'wrapper'=20
appropriate to the init /supervision / management system in use.

Further, a daemon can be a script - rather than something compiled=20
- which then gets called by what i'm calling the 'wrapper'=20
script. In such a case - to make up a silly example, 'chaird' - we=20
can then have a situation of:

  "The `chaird` service script starts the `chaird` daemon script,=20
  which provides a chair service."

%-}

i guess it's one of those cases where considering the context=20
can't be avoided.


Alexis.