On Mon, Dec 08, 2014 at 12:48:46PM -0500, James Carlson wrote: > On 12/08/14 09:57, David Lamparter wrote: > > Now, for usage of the files bundled in Quagga: > > > > * .spec init systemd sysconf logrot. PAM > > Alpine - custom - - - - > > Arch - - custom - sed - > > Debian - custom - - custom custom > > Gentoo - custom - - used custom > > Mageia custom - custom used used used > > OpenSuSE custom custom - - used custom > > Rawhide custom - patch used used - > > Slackware - custom - - custom custom > > Note that even for those that have a "custom" variant, there may well be > someone downstream watching for diffs on the bundled files and manually > reinterpreting those as changes to be applied to the customized version > (or at least as a warning of important changes to note).
Right, thus the keeping one generic version (currently called "redhat") of init scripts around. I'm not arguing against that [anymore]. > I agree with Alexis Rosen's previous message. A "best effort" approach > makes a lot of sense. Keep what's there. Either take a blind stab at > applying changes in the rare case that they happen, or ignore the files > because you don't understand them and let 'em rot; either's fine. > Happily update if someone actually using that system contributes updates > "real" fixes. A readme that says "we think this might have worked at > one point, and we wish you the best of luck" should be sufficient. For most of the files, since they don't match any particular system (anymore - I guess they matched RedHat at some point), I'm not expecting a particular zeal from people in updating them to match a particular distribution. In fact, this is a bit of a self-truth: the point that they don't match any particular distro stems from people not caring to update them, which in turn creates even less interest in updating them. All amplified by no distro actually using them. Actually, keeping them updated for distro reasons (instead of Quagga reasons) generates false churn for other people who watch for changes as you mentioned in your previous paragraph. If we treat these scripts as generic and not poke them needlessly, the signal to noise ratio for packagers gets better... > A principled stand would be to make a platform-agnostic programming > interface. I.e., provide some sort of script or utility that can report > which daemons need to run and with what arguments based on the supplied > configuration, and then leave it up to a platform-supplied mechanism to > determine what to do with that list (feed to shell, tell systemd or > svccfg, or something else). But that's probably ETOOHARD. Well, the Debian init script is doing the (IMHO) best thing it can, by simply starting daemons if their specific config file or the all-daemons one exists. With the multi-instance patch, this would need to be extended to start multiple ospfd processes if configs with filenames "ospf-%d.conf" exist. No idea how to handle integrated config there... grep, probably... -David _______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
