| Submitter | Radu Moisan |
|---|---|
| Date | Feb. 1, 2013, 9:22 a.m. |
| Message ID | <1359710522-26839-1-git-send-email-radu.moisan@intel.com> |
| Download | mbox | patch |
| Permalink | /patch/43799/ |
| State | New |
| Headers | show |
Comments
On Fri, Feb 1, 2013 at 10:22 AM, Radu Moisan <radu.moisan@intel.com> wrote: > enable/disable support as relevant > > Signed-off-by: Radu Moisan <radu.moisan@intel.com> > Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> What caused my sign-off - or is it a hint that meta-oe/meta-systemd need alignment with this? Andreas
On 1 February 2013 09:22, Radu Moisan <radu.moisan@intel.com> wrote: > enable/disable support as relevant Please write a more verbose comment, something along the lines of "systemd is auto-detected unless disabled explicitly, potentially contaminating sstate". > +# there is no --enable/--disable option for consolekit and it's not picked by shlibs, so add it to DEPENDS It's not picked up by shlibs because it's a dbus call not a library call, and that's *RDEPENDS* not depends. Ross
On 02/01/2013 11:39 AM, Andreas Müller wrote: > On Fri, Feb 1, 2013 at 10:22 AM, Radu Moisan <radu.moisan@intel.com> wrote: >> enable/disable support as relevant >> >> Signed-off-by: Radu Moisan <radu.moisan@intel.com> >> Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> > What caused my sign-off - or is it a hint that meta-oe/meta-systemd > need alignment with this? You were the only reference I had from meta-openembedded. Radu
On 1 February 2013 09:39, Andreas Müller <schnitzeltony@googlemail.com> wrote: > On Fri, Feb 1, 2013 at 10:22 AM, Radu Moisan <radu.moisan@intel.com> wrote: >> enable/disable support as relevant >> >> Signed-off-by: Radu Moisan <radu.moisan@intel.com> >> Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> > What caused my sign-off - or is it a hint that meta-oe/meta-systemd > need alignment with this? This patch is the meta-oe appends merged into oe-core - so yes once this has been integrated the meta-oe and meta-systemd bbappends can be removed. Actually looking at the commit log, it appears that JaMa did the important commit so credit to him should be given. Ross
On 02/01/2013 11:47 AM, Burton, Ross wrote: > On 1 February 2013 09:39, Andreas Müller <schnitzeltony@googlemail.com> wrote: >> On Fri, Feb 1, 2013 at 10:22 AM, Radu Moisan <radu.moisan@intel.com> wrote: >>> enable/disable support as relevant >>> >>> Signed-off-by: Radu Moisan <radu.moisan@intel.com> >>> Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> >> What caused my sign-off - or is it a hint that meta-oe/meta-systemd >> need alignment with this? > This patch is the meta-oe appends merged into oe-core - so yes once > this has been integrated the meta-oe and meta-systemd bbappends can be > removed. > > Actually looking at the commit log, it appears that JaMa did the > important commit so credit to him should be given. Noted.
On 02/01/2013 11:44 AM, Burton, Ross wrote: > On 1 February 2013 09:22, Radu Moisan <radu.moisan@intel.com> wrote: >> enable/disable support as relevant > Please write a more verbose comment, something along the lines of > "systemd is auto-detected unless disabled explicitly, potentially > contaminating sstate". > >> +# there is no --enable/--disable option for consolekit and it's not picked by shlibs, so add it to DEPENDS > It's not picked up by shlibs because it's a dbus call not a library > call, and that's *RDEPENDS* not depends. I did change that manually because I assumed it was a typo. However it make sense to add consolekit as a runtime dependency, but the documentation specifically says PACKAGECONFIG will add to DEPENDS. How is it different in this patch? The PACKAGECONFIG[systemd] is clearly specifying a build time dependency, why the PACKAGECONFIG[consolekit] is specifying a runtime dependency? How can I discriminate in this context? Radu
On Fri, Feb 01, 2013 at 12:22:01PM +0200, Radu Moisan wrote: > > On 02/01/2013 11:44 AM, Burton, Ross wrote: > > On 1 February 2013 09:22, Radu Moisan <radu.moisan@intel.com> wrote: > >> enable/disable support as relevant > > Please write a more verbose comment, something along the lines of > > "systemd is auto-detected unless disabled explicitly, potentially > > contaminating sstate". > > > >> +# there is no --enable/--disable option for consolekit and it's not picked by shlibs, so add it to DEPENDS > > It's not picked up by shlibs because it's a dbus call not a library > > call, and that's *RDEPENDS* not depends. > > I did change that manually because I assumed it was a typo. However it > make sense to add consolekit as a runtime dependency, but the > documentation specifically says PACKAGECONFIG will add to DEPENDS. How > is it different in this patch? The PACKAGECONFIG[systemd] is clearly > specifying a build time dependency, why the PACKAGECONFIG[consolekit] is > specifying a runtime dependency? How can I discriminate in this context? 3rd (DEPENDS) or 4th (RDEPENDS) parameter is the key difference.
On 1 February 2013 10:22, Radu Moisan <radu.moisan@intel.com> wrote: > I did change that manually because I assumed it was a typo. However it make > sense to add consolekit as a runtime dependency, but the documentation > specifically says PACKAGECONFIG will add to DEPENDS. How is it different in > this patch? The PACKAGECONFIG[systemd] is clearly specifying a build time > dependency, why the PACKAGECONFIG[consolekit] is specifying a runtime > dependency? How can I discriminate in this context? I spoke to Radu online already, but I'll repeat again here in case anyone else is confused by PACKAGECONFIG. The latest documentation is here: http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-P The key point is that the 3rd argument is DEPENDS, and the fourth is RDEPENDS. Many instances of PACKAGECONFIG will leave out the fourth, but in this case the third is empty (note the extra commas) so consolekit is added as a runtime dependency. Ross
On 02/01/2013 12:27 PM, Burton, Ross wrote: > On 1 February 2013 10:22, Radu Moisan <radu.moisan@intel.com> wrote: >> I did change that manually because I assumed it was a typo. However it make >> sense to add consolekit as a runtime dependency, but the documentation >> specifically says PACKAGECONFIG will add to DEPENDS. How is it different in >> this patch? The PACKAGECONFIG[systemd] is clearly specifying a build time >> dependency, why the PACKAGECONFIG[consolekit] is specifying a runtime >> dependency? How can I discriminate in this context? > I spoke to Radu online already, but I'll repeat again here in case > anyone else is confused by PACKAGECONFIG. > > The latest documentation is here: > http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-P > > The key point is that the 3rd argument is DEPENDS, and the fourth is > RDEPENDS. Many instances of PACKAGECONFIG will leave out the fourth, > but in this case the third is empty (note the extra commas) so > consolekit is added as a runtime dependency. > > Ross Thanks guys for the clarification. Very useful info, I didn't notice the extra comma. Radu
Patch
diff --git a/meta/recipes-extended/polkit/polkit_0.104.bb b/meta/recipes-extended/polkit/polkit_0.104.bb index 2c0aaca..08e016a 100644 --- a/meta/recipes-extended/polkit/polkit_0.104.bb +++ b/meta/recipes-extended/polkit/polkit_0.104.bb @@ -7,8 +7,13 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=155db86cdbafa7532b41f390409283eb \ DEPENDS = "expat glib-2.0 intltool-native gobject-introspection-stub" -PACKAGECONFIG = "${@base_contains('DISTRO_FEATURES', 'pam', 'pam', '', d)}" +PACKAGECONFIG = "${@base_contains('DISTRO_FEATURES', 'pam', 'pam', '', d)} \ + ${@base_contains('DISTRO_FEATURES','systemd','systemd','consolekit',d)}" + PACKAGECONFIG[pam] = "--with-authfw=pam,--with-authfw=shadow,libpam,libpam" +PACKAGECONFIG[systemd] = "--enable-systemd,--disable-systemd,systemd" +# there is no --enable/--disable option for consolekit and it's not picked by shlibs, so add it to DEPENDS +PACKAGECONFIG[consolekit] = ",,,consolekit" PR = "r9"