Hello, It seems like ever since useradd.bbclass was written it added an RDEPEND on "shadow". See commit a82885db00956734922291d8a17eb135461204fa for an ancient reference. I'm wondering if this is really necessary?
Recent commit 84efd72d48616405dbe4d73ec95917077144ed09 added systemd's udev to USERADD_PACKAGES so that a "GROUPADD_PARAM:udev" took effect. In our initramfs we have a package that depends on udev (in order to deal with the emmc device containing the rootfs), but we also explicitly 'PACKAGE_EXCLUDE = "shadow"' in the initramfs recipe. It seems like this was done to avoid all the PAM dependencies that shadow would drag in. This now ends up un-buildable without adding 'USERADD_PACKAGES:remove = "udev"' to our systemd_%.bbappend. I briefly considered using eudev in the initramfs, and systemd-udev in our rootfs, but unfortunately there is also 'CONFLICT_DISTRO_FEATURES = "systemd"' in eudev. I suppose the shadow-RDEPEND is necessary for the case where the package is installed via RPM/IPK? Is there any mechanism to specify that this RDEPEND doesn't need to be satisfied in a static image that I'm not aware of? -- Patrick Williams
signature.asc
Description: PGP signature
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#173493): https://lists.openembedded.org/g/openembedded-core/message/173493 Mute This Topic: https://lists.openembedded.org/mt/95120913/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
