On Fri, Mar 6, 2015 at 6:03 AM, Bottazzini, Bruno <[email protected]> wrote: > On Sex, 2015-03-06 at 09:23 +0100, Anders Darander wrote: >> * Bottazzini, Bruno <[email protected]> [150305 17:15]: >> >> > On Qui, 2015-03-05 at 15:28 +0100, Anders Darander wrote: >> >> > > Just a quick question before I look into the patch in more detail. >> >> > > Is the new setting of PACKAGECONFIG consistent with how systemd was >> > > built previously? I guess it is. >> >> > Hi Anders, >> >> > it is consistent with how systemd was built previously. If you apply the >> > patch and bitbake it. Systemd will be built and shipped normally. >> >> > But now it will give some options on how to customize it by excluding >> > packages you don't want to be with systemd. >> >> Nice, I'm really liking this! That's something I've planned on doing >> myself for a while. >> >> > > Another comment, you should remove the dependcies that gets added using >> > > PACKAGECONFIG from DEPENDS, e.g. acl etc. (Or are they required >> > > nevertheless?) >> >> > You mean I should do the following ? >> > - PACKAGECONFIG[acl] = "--enable-acl,--disable-acl,acl" >> > + PACKAGECONFIG[acl] = "--enable-acl,--disable-acl" >> >> No, I meant to remove them from the long >> DEPENDS = "kmod docbook-sgml-dtd-4.1-native intltool-native gperf-native acl >> readline dbus libcap libcgroup glib-2.0 qemu-native util-linux" >> line. (Unless I overlooked that part in your patch?) >> >> > If I get what you said correctly, yes they are required. >> >> What I meant, was thas unless e.g. acl is required even when building >> with --disable-acl, it's better to add the acl dependency in the >> PACKAGECONFIG like you to. Though, at the same time, remove acl from the >> long DEPENDS-line. (As otherwise we'll build acl anyway). > > Anders, > > You are right. > > It is not needed to specify the libs on depends. > > I have removed it and as soon as we review the whole patch I will send a > new version with this corrected >
On top of all this 'acl' is a distro feature so that should be accounted for too. -- _______________________________________________ Openembedded-core mailing list [email protected] http://lists.openembedded.org/mailman/listinfo/openembedded-core
