On Mon, 04.08.14 11:11, Umut Tezduyar Lindskog (u...@tezduyar.com) wrote: > >> diff --git a/units/ldconfig.service b/units/ldconfig.service > >> index 43c145b..09a2b74 100644 > >> --- a/units/ldconfig.service > >> +++ b/units/ldconfig.service > >> @@ -13,6 +13,7 @@ Conflicts=shutdown.target > >> After=systemd-readahead-collect.service systemd-readahead-replay.service > >> systemd-remount-fs.service > >> Before=sysinit.target shutdown.target systemd-update-done.service > >> ConditionNeedsUpdate=/etc > >> +ConditionFileIsExecutable=/sbin/ldconfig > >> > >> [Service] > >> Type=oneshot > > > > Ummmmmhh... Do we really want this? I presume you are not using glibc, > > are you? > > We are actually. glibc 2-15 but not x86 ISA. > > > > > I mean, whether ldconfig is around or not is not really an admin > > decision, it's an OS vendor decision. Because of that we should probably > > make shiping the ldconfig service file optional, rather then always ship > > it but then conditionalize it... > > > > I'd prefer to keep conditions rather minimal in general... > > I could also mask the service but it is good to deviate as little as > possible. Maybe a configure check to install the service or not?
Yeah, I think a --enable-ldconfig-service or so switch would be good, so that we can drop the conditionalization. Lennart -- Lennart Poettering, Red Hat _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel