systemd packages -> image

systemd packages -> image

[Andreas Müller]

Hi,

Imagine a recipe is included into an image by a task and later gets systemd-
native support.
How about forcing the packages in SYSTEMD_PACKAGES packed into images in case 

IMAGE_INIT_MANAGER  = "systemd"

is found in the image definition?

Andreas

Re: systemd packages -> image

[Anders Darander]

* Andreas Müller <schnitzeltony@gmx.de> [111109 23:17]:
> Imagine a recipe is included into an image by a task and later gets systemd-
> native support.
> How about forcing the packages in SYSTEMD_PACKAGES packed into images in case 

> IMAGE_INIT_MANAGER  = "systemd"

> is found in the image definition?

Sure, that sounds like a nice idea. Although, in that case, SysV-init
files should probably also be packaged in separate packages and be
installed in a similar manner.

Cheers,
Anders

-- 
Anders Darander
ChargeStorm AB / eStorm AB

Re: systemd packages -> image

[Paul Eggleton]

On Thursday 10 November 2011 08:47:17 Anders Darander wrote:
> * Andreas Müller <schnitzeltony@gmx.de> [111109 23:17]:
> > Imagine a recipe is included into an image by a task and later gets
> > systemd- native support.
> > How about forcing the packages in SYSTEMD_PACKAGES packed into images in
> > case
> > 
> > IMAGE_INIT_MANAGER  = "systemd"
> > 
> > is found in the image definition?
> 
> Sure, that sounds like a nice idea. Although, in that case, SysV-init
> files should probably also be packaged in separate packages and be
> installed in a similar manner.

Why are we going to such an effort to provide both sysvinit and systemd in the 
same distro? Why not make it a distro policy decision? This would save a huge 
amount of work.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre

Re: systemd packages -> image

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Op 10-11-11 10:47, Paul Eggleton schreef:
> On Thursday 10 November 2011 08:47:17 Anders Darander wrote:
>> * Andreas Müller <schnitzeltony@gmx.de> [111109 23:17]:
>>> Imagine a recipe is included into an image by a task and later gets 
>>> systemd- native support. How about forcing the packages in
>>> SYSTEMD_PACKAGES packed into images in case
>>> 
>>> IMAGE_INIT_MANAGER  = "systemd"
>>> 
>>> is found in the image definition?
>> 
>> Sure, that sounds like a nice idea. Although, in that case, SysV-init 
>> files should probably also be packaged in separate packages and be 
>> installed in a similar manner.
> 
> Why are we going to such an effort to provide both sysvinit and systemd
> in the same distro? Why not make it a distro policy decision? This would
> save a huge amount of work.

So I can start sending patches to add the systemd files to ${PN} in oe-core
recipes now? That's the big hold up: lack of vision from the OE-core folks.


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)
Comment: GPGTools - http://gpgtools.org

iD8DBQFOu6VyMkyGM64RGpERAsQzAKCiZU39orO0DUQ7sUw6y5Dk/pbk9ACggabc
BcceXFVsVOS2GqnFvC2KbZ4=
=MYbw
-----END PGP SIGNATURE-----

Re: systemd packages -> image

[Paul Eggleton]

On Thursday 10 November 2011 11:20:34 Koen Kooi wrote:
> So I can start sending patches to add the systemd files to ${PN} in oe-core
> recipes now?

I would think if these additions are conditional upon systemd being in 
DISTRO_FEATURES, I can't see a problem. I think the ideal situation would be 
to have a generic "init" bbclass (or some better name) which does all the work 
for you regardless of whether sysvinit or systemd is used (i.e. it would 
effectively replace update-rc.d.bbclass), and save having to have packaging 
boilerplate in every recipe.

My main point was, yes there are both people who want to stay with sysvinit 
and people who are keen to move to systemd, but I'm not convinced both of 
those need to be catered for within the same distro policy.

> That's the big hold up: lack of vision from the OE-core folks.

This is a community project, if you have ideas or suggestions, just start the 
discussion...

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre

Re: systemd packages -> image

[Steffen Sledz]

On 10.11.2011 11:35, Paul Eggleton wrote:
> On Thursday 10 November 2011 11:20:34 Koen Kooi wrote:
>> So I can start sending patches to add the systemd files to ${PN} in oe-core
>> recipes now?
> 
> I would think if these additions are conditional upon systemd being in 
> DISTRO_FEATURES, I can't see a problem. I think the ideal situation would be 
> to have a generic "init" bbclass (or some better name) which does all the work 
> for you regardless of whether sysvinit or systemd is used (i.e. it would 
> effectively replace update-rc.d.bbclass), and save having to have packaging 
> boilerplate in every recipe.

What about upstart in this context? An generic "init" bbclass should support this too.

> My main point was, yes there are both people who want to stay with sysvinit 
> and people who are keen to move to systemd, but I'm not convinced both of 
> those need to be catered for within the same distro policy.
> 
>> That's the big hold up: lack of vision from the OE-core folks.
> 
> This is a community project, if you have ideas or suggestions, just start the 
> discussion...

-- 
DResearch Fahrzeugelektronik GmbH
Otto-Schmirgal-Str. 3, 10319 Berlin, Germany
Tel: +49 30 515932-237 mailto:sledz@dresearch-fe.de
Fax: +49 30 515932-299
Geschäftsführer: Dr. Michael Weber, Werner Mögle;
Amtsgericht Berlin Charlottenburg; HRB 130120 B;
Ust.-IDNr. DE273952058

Re: systemd packages -> image

[Anders Darander]

* Koen Kooi <koen@dominion.thruhere.net> [111110 11:21]:
> Op 10-11-11 10:47, Paul Eggleton schreef:
> > On Thursday 10 November 2011 08:47:17 Anders Darander wrote:
> >> * Andreas Müller <schnitzeltony@gmx.de> [111109 23:17]:
> >>> Imagine a recipe is included into an image by a task and later gets 
> >>> systemd- native support. How about forcing the packages in
> >>> SYSTEMD_PACKAGES packed into images in case

> >>> IMAGE_INIT_MANAGER  = "systemd"

> >>> is found in the image definition?

> >> Sure, that sounds like a nice idea. Although, in that case, SysV-init 
> >> files should probably also be packaged in separate packages and be 
> >> installed in a similar manner.

> > Why are we going to such an effort to provide both sysvinit and systemd
> > in the same distro? Why not make it a distro policy decision? This would
> > save a huge amount of work.

> So I can start sending patches to add the systemd files to ${PN} in oe-core
> recipes now? That's the big hold up: lack of vision from the OE-core folks.

Why not, it'd be nice to not have to install a bunch of extra packages
just to get the systemd files installed. (OTOH, no big deal, but then we
also could consider splitting out the SysV-init-files...)

And we should also move systemd into oe-core. That's atleast my wish... 

/Anders

-- 
Anders Darander
ChargeStorm AB / eStorm AB