[meta-freescale] Packagegroup unexpectedly builds recipe

Jonas Rydow jrydow at gmail.com
Wed Oct 8 05:51:46 PDT 2014


Hi,
I have tried to get an image recipe to work with packagegroups. The problem
is that a recipe is built that in my opinion should not be built, in this
case portmap. I have narrowed it down to basically the example in the yocto
dev manual:
A packagegroup recipe packagegroup-pg-image.bb:
-------------------
DESCRIPTION = "My Custom Package Groups"
inherit packagegroup
LICENSE = "MIT"
PACKAGES = "\
    packagegroup-custom-apps \
    packagegroup-custom-tools \
    "
RDEPENDS_packagegroup-custom-apps = "\
    dropbear \
    psplash"
RDEPENDS_packagegroup-custom-tools = "\
    portmap \
"
-------------------

An image recipe, pg-test-image.bb:
-------------------
SUMMARY = "Package group test image definition."
DESCRIPTION = "This image does not do anything useful"
LICENSE = "MIT"
inherit image
IMAGE_FEATURES_append = " package-management"
IMAGE_INSTALL = "\
    packagegroup-custom-apps \
"
-------------------

When bitbaking, the portmap recipe is also baked, is this expected?
If looking at the debug log from:
bitbake -DDDD pg-test-image > pg-test-image.log the following is seen
--------------
DEBUG: Added runtime dependencies ['portmap', 'dropbear', 'psplash'] for
/home/jryd/..../meta-sre-freja/recipes-core/packagegroups/
packagegroup-pg-image.bb
---------------
It does not only get built but also populates sysroot, but it does not show
up in the manifest file for the image.

This is of course a constructed example, my real problem is that one
package gets built and installs things on sysroot which another package
should provide. This means that the last of the two recipes get to provide
the file(s). This causes build failures and or warnings.

I use daisy 1.6.1, to build for the sabresd.

Best regards
Jonas Rydow
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.yoctoproject.org/pipermail/meta-freescale/attachments/20141008/181e7f16/attachment.html>


More information about the meta-freescale mailing list