[OE-core] Correct way of having the PN package depend on

Richard Purdie richard.purdie at linuxfoundation.org
Tue Sep 20 21:30:29 UTC 2011


On Tue, 2011-09-20 at 14:00 -0700, Chris Larson wrote:
> On Tue, Sep 20, 2011 at 1:24 PM, Daniel Lazzari <dlazzari at leapfrog.com> wrote:
> >> On 9/20/11 2:03 PM, Daniel Lazzari wrote:
> >> > I noticed a problem with the alsa-utils package where the alsa-utils package does not depend on all of the little extra packages it's supposed to pull in. I'm not sure what the best way to fix this is though. Can anyone give me a little guidance or point me at a recipe that does this correctly already?
> >>
> >> If you are using the RPM backend, I'm currently working on something that may
> > resolve this issue for you.  Hopefully ready today.
> >>
> >> Otherwise, switch to ipk or deb and see if you get the same behavior.
> >
> > I'm using ipk already.
> >
> >> I don't see why this is a problem, personally. If someone wants a recipe's main package to depend on the other packages it emits, they're free to make the recipe do so. But the ${PN} package is not defined to include everything from the recipe, it never has been.
> >
> > I understand that. With the alsa-utils package, the recipe specifically mentions
> > # This are all packages that we need to make. Also, the now empty alsa-utils
> > # ipk depends on them.
> >
> > Right above adding packages such as "alsa-utils-alsamixer" to the PACKAGES variable. I need all of those more specific packages and if I understand correctly, I should be able to get them by just adding the "alsa-utils" package to my custom image.
> 
> Sounds like it's just a recipe bug then, if it claims to do something
> which it doesn't do.

Its defintely a bug in the recipe with the "alsa-utils" package just
being broken. We need to do something like:

ALSAUTILSPKGS = "alsa-utils-x \
                 alsa-utils-y \
                 alsa-utils-z"

PACKAGES += "${ALSAUTILSPKGS}"
RDEPENDS_${PN} += "${ALSAUTILSPKGS}"

which should make the recipe do what it says it should do.

Cheers,

Richard






More information about the Openembedded-core mailing list