[libvirt] configure --prefix=$HOME --with-init-script buglet

Right now, on a Fedora system, if you configure libvirt to install into $HOME (because you don't have permissions to install into the default /usr/local or even into the system /usr), --with-init-script defaults to auto mode, detects that /etc/redhat-release exists, and tries to install the init script anyways. configure.ac needs to be patched to also check that $sysconfdir is /etc before changing with-init-script=auto over to with-init-script=redhat, so that the init scripts are only installed during a system install (where they make sense), and not during a /usr/local or $HOME install. I don't have time to patch this today, if someone wants to beat me to it. -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org

On Tue, May 17, 2011 at 10:26:54 -0600, Eric Blake wrote:
Right now, on a Fedora system, if you configure libvirt to install into $HOME (because you don't have permissions to install into the default /usr/local or even into the system /usr), --with-init-script defaults to auto mode, detects that /etc/redhat-release exists, and tries to install the init script anyways.
configure.ac needs to be patched to also check that $sysconfdir is /etc before changing with-init-script=auto over to with-init-script=redhat, so that the init scripts are only installed during a system install (where they make sense), and not during a /usr/local or $HOME install.
Why? I've never seen a package that avoids installing some files according to prefix. It certainly doesn't make any sense if with-init-script is explicitly set to redhat. With with-init-script set to auto, I can see a bit of sense in this but it still looks very weired that changing prefix has any impact on detected features. Jirka
participants (2)
-
Eric Blake
-
Jiri Denemark