[OE-core] [PATCH 0/2 V2] add rconflict for systemd/tiny-init

Adrian Bunk bunk at stusta.de
Thu May 30 12:05:43 UTC 2019


On Thu, May 30, 2019 at 12:40:12PM +0100, Richard Purdie wrote:
> On Thu, 2019-05-30 at 14:15 +0300, Adrian Bunk wrote:
> > On Thu, May 30, 2019 at 01:32:18PM +0800, changqing.li at windriver.com
> > wrote:
> > > From: Changqing Li <changqing.li at windriver.com>
> > > 
> > > 
> > > Changqing Li (2):
> > >   tiny-init: add rconflicts
> > >   systemd: add rconflicts
> > > ...
> > 
> > I wonder whether the conflicts between init systems could be handled
> > with
> >     RCONFLICTS_${PN} = "${VIRTUAL-RUNTIME_init_manager}"
> > 
> > This would avoid the problem that all n init systems would need 
> > RCONFLICTS with all n-1 other init systems.
> > 
> > In my testing this seems to work with all supported package managers
> > (including the self-conflict not causing problems).
> 
> I think we can support multiple init systems within one package feed,
> at least to some degree.

This should work with either solution, or do I miss anything?

> I'm also very nervous about the self-conflict, 
> I'm surprised that works with all the package managers as I thought I'd
> seen that fail.

In Debian that's a common pattern for exactly this purpose,
see the last section of [1].

> Was that with rootfs construction or just building the
> packages?

rootfs construction with opkg/deb/rpm [2]

> Cheers,
> 
> Richard

cu
Adrian

[1] https://www.debian.org/doc/debian-policy/ch-relationships.html#virtual-packages-provides
[2] with tar rootfs construction is generally impossible

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed



More information about the Openembedded-core mailing list