[OE-core] [PATCH 3/5] bootchart2: add initscripts-functions as rconflict

[richard.purdie at linuxfoundation.org]

On Wed, 2019-05-29 at 10:45 +0800, Changqing Li wrote:
> On 5/29/19 7:13 AM, Richard Purdie wrote:
> > On Tue, 2019-05-28 at 12:01 +0800, changqing.li at windriver.com
> > wrote:
> > > From: Changqing Li <changqing.li at windriver.com>
> > > 
> > > bootchart2 rdepend on lsb, lsb rdepend on lsbinitscripts,
> > > lsbinitscripts rconflict with initscripts-functions. so also add
> > > rconflict for this one.
> > > 
> > > Signed-off-by: Changqing Li <changqing.li at windriver.com>
> > > ---
> > >   meta/recipes-devtools/bootchart2/bootchart2_0.14.8.bb | 1 +
> > >   1 file changed, 1 insertion(+)
> > > 
> > > diff --git a/meta/recipes-
> > > devtools/bootchart2/bootchart2_0.14.8.bb
> > > b/meta/recipes-devtools/bootchart2/bootchart2_0.14.8.bb
> > > index 396d0ec..36092e6 100644
> > > --- a/meta/recipes-devtools/bootchart2/bootchart2_0.14.8.bb
> > > +++ b/meta/recipes-devtools/bootchart2/bootchart2_0.14.8.bb
> > > @@ -143,6 +143,7 @@ RDEPENDS_pybootchartgui = "python3-pycairo
> > > python3-compression python3-image pyt
> > >   RDEPENDS_${PN}_class-target += "${@bb.utils.contains('DISTRO_FE
> > > ATURE
> > > S', 'sysvinit', 'sysvinit-pidof', 'procps', d)}"
> > >   RDEPENDS_${PN}_class-target += "lsb"
> > >   DEPENDS_append_class-native = " python3-pycairo-native"
> > > +RCONFLICTS_${PN}_class-target = "initscripts-functions"
> > If lsb and initscripts-functions have the conflict, why do we need
> > this in bootchart2 as well?
> 
> Yes,  this is wrong for indirect rconflict,  especially that, this
> maybe 
> break someday when lsb changed.  Same as lsb/lsbtest in patch  4/5
> and 5/5.
> 
> 
> For patch 1/5, 2/5,  it is direct rconflict.  It will also report 
> rconflict during do_rootfs when both are  installed  if only one
> package 
> set rconflict.
> 
> but I think it is better to set rconflict for both packages, in this 
> way,  when user parse the recipe to get the recipe info, it can get  
> this rconflict info.

We only want to list direct conflicts, not indirect ones. If we tried
to map all indirect conflicts we'd have something totally
unmaintainable, it wouldn't scale.

Cheers,

Richard