Support for OpenRC


Achara, Jagdish P <jagdishpachara@...>
 

Hi,

Currently we have the option to choose either sysvinit or systemd . Would, at some point, openrc be included in this list of options to choose from?

Jagdish


Richard Purdie
 

On Sat, 2020-09-05 at 08:15 -0700, Achara, Jagdish P wrote:
Hi,

Currently we have the option to choose either sysvinit or systemd .
Would, at some point, openrc be included in this list of options to
choose from?
It comes down to the demand for it, whether there are people willing
the maintain it, how much of the system its planned to support and so
on. It has implications for the testing matrix for example.

The hope is individual layers could add support for things like this
and that would let people use it and let us gauge demand too.

Cheers,

Richard


 

On Sat, 5 Sep 2020 at 16:28, Richard Purdie
<richard.purdie@...> wrote:

On Sat, 2020-09-05 at 08:15 -0700, Achara, Jagdish P wrote:
Hi,

Currently we have the option to choose either sysvinit or systemd .
Would, at some point, openrc be included in this list of options to
choose from?
It comes down to the demand for it, whether there are people willing
the maintain it, how much of the system its planned to support and so
on. It has implications for the testing matrix for example.

The hope is individual layers could add support for things like this
and that would let people use it and let us gauge demand too.
This does exist in https://github.com/jsbronder/meta-openrc, I haven't
tested it myself though.

--
Paul Barker
Konsulko Group


Khem Raj
 

On Sat, Sep 5, 2020 at 8:28 AM Richard Purdie
<richard.purdie@...> wrote:

On Sat, 2020-09-05 at 08:15 -0700, Achara, Jagdish P wrote:
Hi,

Currently we have the option to choose either sysvinit or systemd .
Would, at some point, openrc be included in this list of options to
choose from?
It comes down to the demand for it, whether there are people willing
the maintain it, how much of the system its planned to support and so
on. It has implications for the testing matrix for example.

The hope is individual layers could add support for things like this
and that would let people use it and let us gauge demand too.
init systems are quite taxing and intrusive to implement and hence I
agree with testing complexity
increase and in general higher maintenance work. meta-openrc seems a
good solution for now,
ideally, recipes should provide openRC scripts via bbappends and
perhaps create it as a DISTRO_FEATURE
but I think that's a good starting point to start with openrc and if
there are many users showing interest in
future we should definitely review it.


Cheers,

Richard


Rich Persaud
 

On Sep 5, 2020, at 13:55, Khem Raj <raj.khem@...> wrote:

On Sat, Sep 5, 2020 at 8:28 AM Richard Purdie
<richard.purdie@...> wrote:

On Sat, 2020-09-05 at 08:15 -0700, Achara, Jagdish P wrote:
Hi,

Currently we have the option to choose either sysvinit or systemd .
Would, at some point, openrc be included in this list of options to
choose from?

It comes down to the demand for it, whether there are people willing
the maintain it, how much of the system its planned to support and so
on. It has implications for the testing matrix for example.

The hope is individual layers could add support for things like this
and that would let people use it and let us gauge demand too.

init systems are quite taxing and intrusive to implement and hence I
agree with testing complexity
increase and in general higher maintenance work. meta-openrc seems a
good solution for now,
ideally, recipes should provide openRC scripts via bbappends and
perhaps create it as a DISTRO_FEATURE
but I think that's a good starting point to start with openrc and if
there are many users showing interest in
future we should definitely review it.

Devuan has also been testing in support of multiple init systems, which should improve upstream package readiness.  So far they have:

  - sysvinit
  - openrc
  - runit


Rich