lianep at eng.sun.com wrote: > I think the more useful RFE here would be to list the manifest file(s) > which contributed to the service definition in the repository. Allow > the question to be answered directly rather than trying to answer > it by convention.
Agreed completely. Although obeying convention is not bad either, I would expect to find milestone manifests in /var/svc/manifest/milestone. At one time there still was tftp in inetd.conf and I would run inetconv to demo the conversion and it made a file called tftp-udp6.xml (I think) but the FMRI was network/tftp/udp6:default. Drove me crazy. > The only annoying part is if the service is imported from a file that no > longer exists on the filesystem (or is on an unmounted filesystem), but > that's an OK answer for an admin too, I'd think. Yes, if I can query the repository and get something like "initial import:/path/to/xml", I can find out for myself if that file is still there or has gone away for whatever reason. > Still, aside from idle curiousity, I'm not sure why this is important > to someone not trying to fix a bug in the manifest. Do remember that > a very valid answer to the question is "there is no manifest. this > service was created by software." Fine too. But I'd like to know. > Let's work through cases where > locating the manifest is deemed necessary, and make sure we have RFEs > filed against the repository to ensure that those cases can be resolved > without resorting to a tool that's not always available to you. This question was raised by someone curious ... we admins have lived & died with tangible files for <hyperbole>hundreds</hyperbole> of years, don't take it away from us. :) CT
