I suppose I have my package according to packaging guidelines, so could you remove these errors from rpmlint output? at.src:136: E: hardcoded-library-path in %{buildroot}/lib/systemd/system/ at.src:137: E: hardcoded-library-path in %{buildroot}/lib/systemd/system/atd.service at.src:194: E: hardcoded-library-path in /lib/systemd/system/atd.service For *-sysvinit sub-package does exist error, which probably can't be fixed in rpmlint: at-sysvinit.x86_64: E: init-script-without-chkconfig-postin /etc/rc.d/init.d/atd at-sysvinit.x86_64: E: init-script-without-chkconfig-preun /etc/rc.d/init.d/atd But this warning might be fixed: at-sysvinit.x86_64: W: incoherent-init-script-name atd ('at-sysvinit', 'at-sysvinitd') Initscript name is correctly atd, but it's in sub-package at-sysvinit.
Are you using %{_unitdir} instead of /lib/systemd/system/ ? The hardcoded-library-path test isn't very smart, I don't think it will go off if you're using the macro.
(In reply to comment #1) > Are you using %{_unitdir} instead of /lib/systemd/system/ ? > > The hardcoded-library-path test isn't very smart, I don't think it will go off > if you're using the macro. Ha, new macro! Thanks. I'm not sure how this warnings should be solved but they should be somehow.
As a general rule, I'm a lot less concerned about resolving Warnings, than I am Errors. The intent of rpmlint is to identify potential areas of concern, then to let the human packagers examine them and determine if they are valid or not. I'm not sure of a good way to teach rpmlint that a "$foo-sysvinit" package will have a different initscript name, because any of the ways I can think of will make the check far more complicated than it is currently. So, I'm going to close this bug out, unless the hardcoded-library-path error triggers when you use the %{_unitdir} macro (and if it does, please reopen the bug). Thanks.
(In reply to comment #0) > at-sysvinit.x86_64: E: init-script-without-chkconfig-postin > /etc/rc.d/init.d/atd > at-sysvinit.x86_64: E: init-script-without-chkconfig-preun /etc/rc.d/init.d/atd This could be filtered out in Fedora's rpmlint config with something like this: addFilter("-sysvinit.* init-script-without-chkconfig") But is it actually intentional that when the at-sysvinit subpackage is installed, chkconfig will not be run for it? I haven't looked into guidelines concerning *-sysvinit packages, but not running chkconfig for them sounds fishy to me. > at-sysvinit.x86_64: W: incoherent-init-script-name atd ('at-sysvinit', > 'at-sysvinitd') > Initscript name is correctly atd, but it's in sub-package at-sysvinit. http://rpmlint.zarb.org/cgi-bin/trac.cgi/changeset/1875 By the way, these errors for at-sysvinit-3.1.12-10.fc16.x86_64.rpm look like something you probably want to look into: at-sysvinit.x86_64: E: non-executable-script /etc/rc.d/init.d/atd 0644L /bin/sh at-sysvinit.x86_64: E: init-script-non-executable /etc/rc.d/init.d/atd