On Wed, Jul 16, 2014 at 12:57:41PM +0200, Tom Gundersen wrote: > If this is merely for sysv compat, it feels wrong to add this in the > native unit files (but maybe the argument could be made regardless?).
That was indeed the original reason for the patch, but I believe it to be the right thing to do anyway. At Wednesday 16 July 2014 14:50:39 Zbigniew Jędrzejewski-Szmek wrote: > It does not provide anything for the services which are part of remote- > fs.target, they have to carry their own dependencies on local-fs.target if > necessary. Correct, or prefferably only on the mount-points in local-fs.target they actually need... > So the only thing it does is that it'll delay remote-fs.target if it were to > be reached before local-fs.target. Which in my oppinion is a good thing, as writers of service files, as well as legacy init scripts, should rightfully be able to assume that if they request all remote file systems to be mounted, local file systems will be too. Also I would have a hard time explaining why a change of a remote mount to a local one should be allowed to cause regressions... / Jon _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel