Hi Steve, and thanks for your bugreport, Le 24.02.2012 10:05, Steve Langasek a écrit : > Please find attached a patch for lsb that adds a new helper function to > /lib/lsb/init-functions to support maintainer scripts that should behave as > no-ops because the package is upstart-aware and upstart is in use.
Okay. > I'm proposing this change in support of bug #591791, a bug against > debian-policy which aims to come up with coherent rules about the inclusion > of native upstart jobs in Debian. This is not a very high-level > abstraction, it still requires the init script to work out the correct > return value for each invocation. That's about as high-level as most of > the other lsb functions get anyway, but if you'd like something different > I'm certainly open to discussing. tl;dr: Is the "LSB support package" really the good place to support multiple init systems in Debian ? I'm mostly fine with the current patch; I just wonder if /lib/lsb/init-functions is really the good file/package to have all those Debian-specific initscript functions. As was highlighted by the #596529 bug, /lib/lsb/init-functions already ships way more functions that what the LSB mandates. I wonder if moving the Debian-specific (non-LSB) functions to an hypothetic /lib/debian/init-functions or alike wouldn't make things cleaner, e.g. maintained in a Debian-specific package, such as `base-files` or `debian-initsupport` or `whatever`. We could even consider that debian-initsupport package as the starting support for all the "let's support more than one init system" problem. On the other hand, as I expressed in my #596529 wontfix, many packages in Debian currently depend on functions implemented in /lib/lsb/init-functions and breaking that would certainly break wide parts of the archive and just continue to add more functions to /lib/lsb/init-functions is the easy way forward. What do you think ? Cheers, OdyX
signature.asc
Description: OpenPGP digital signature
