[Gordon Farquharson] > Hi Petter > > So after a lot of reading on init scripts and insserv, it seems to me > that the init scripts provided by nslu2-utils are not supportable by > insserv. You had written the following in a previous email thread:
Note I am one of the maintainers of sysv-rc too, and it is not ment to support multiple identical entries in one runlevel, and definitely not start and stop symlinks on the same level. > I believe the sysv-rc script calling the boot and shutdown scripts > will not behave the way you expect if there are both start and stop > symlinks for a given runlevel. I am not sure what it will do, but > it is definitely not defined how it should behave. :) In short, only > start or stop at a given runlevel. This is still true. > I can't find any documentation to support the assertion that one may > not put the same init script in a runlevel twice (e.g. K05zleds and > S99zleds in rc1.d). Actually, the Debian Policy seems to support > this use: Well, it has been an underlying assumtion on the implementation of /etc/init.d/rc and its optimization. The fact that zleds is executed at all is pure luck. If the optimization had been implemented slightly differently, it would not. > The init scripts provided by nslu2-utils require exactly this > functionality. I therefore provide that we add empty LSB headers to > scripts in nslu2-utils to meet the release goal of having an LSB > header, but document that the nslu2-utils package will not work > correctly with insserv. How does this sound to everybody? What about splitting it in two instead, one to run early, and one to run late? Will it work. I believe it is a bad idea to base nslu2-utils on undefined behaviour, and would recommend finding a way that did not have the same script run both stop and start symlinks in the same runlevel. happy hacking, -- Petter Reinholdtsen -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
