Bugzilla – Bug 1085143
X11:XOrg/xorg-x11-fonts requires some cleanup
Last modified: 2018-04-18 13:20:26 UTC
1. Remove handling for oS <= 12.2 (trivial, remove conditionals, remove %run_suseconfig_fonts) 2. use %ghost instead of %config - rpmlint dislikes config files outside /var, /etc - %config creates *.rpmsave files - *.rpmsave is not removed when package is removed, as a result, also the dir is kept - %config ships data inside the rpm which is altered/created with %posttrans anyway 3. track all files fonts.{scale,dir}, encodings.dir, .fonts-config-timestamp are created as a result of %posttrans resp. fonts-config. Track these files so on package removal the files and the containing directory can be removed 3a. ignore mode for .fonts-config-timestamp The file is only created when mkfontdir + mkfontscale are installed (optional). If the file does not exist, rpm -V reports a mismatching mode In case a package is removed, fonts.dir etc. may also be removed for non-empty dirs (e.g. truetype), but these are automatically recreated by %posttrans. This is inverse to the current problem the files will be overwritten during package install and only corrected during %posttrans.
Related submit request: https://build.opensuse.org/request/show/586106
Ok. Just had a try with your package. Appears to work, even if mkfontscale and mkfontdir are not installed. I would have expected a naked Xserver not being able to load the cursor font during startup, since it's not being referenced an empty /misc/fonts.dir, but it magically works. Xserver can even load without any font being installed. I'm not sure why/how this works, but it works. You can break it manually though. X -retro -fc cursor works, whereas X -retro -fc cursor2 does not. Anyway, I'm meanwhile convinced of your changes/cleanup. In a totally minimalistic system with only xorg-x11-fonts-core fonts package installed and recommends to mkfontdir/mkfontscale being ignored you may run into issues though with X clients/frameworks only using X core fonts. But I would say in such case you should better know what you're doing anyway. ;-)
Could you please redo the submit request and refer to this bugreport? I'm goint to accept it then. Thanks a lot!
Just accepted Stefan B.'s submit request. :-) Closing ...