> It would be nice if lintian would warn if maintainer scripts use the > init.d script directly and not via invoke-rc.d. Checking that the string > "/etc/init.d/" exists but "invoke-rc.d" does not might be a good enough > approximation to catch most such problems. I did a quick check using the attached shell script to find potentially bad postinst files. It found, in sarge/main/i386, 134 binary packages that call an init.d script (not always their own) with a start or restart argument, but don't use invoke-rc.d if it is available. I then checked the results manually and found that there were only two false positives: php4-tclink and uif. Both tell the user how to run the init.d script but don't do it themselves. I thus conclude that the test in the script is pretty good. I don't have the time right now to implement this as a lintian check, but I hope that may happen by someone else's efforts (or mine, at some future date). -- Those who do, decide.
Attachment:
find-lack-of-invoke
Description: application/shellscript