Gunnar, "quite often" sounds worrysome -- this smells like a dpkg bug then? Did you see more duplicates of this?
This is an utterly complex and probably brittle workaround. If we need such a workaround, then the pre*inst* should call "dpkg -L <pkgname>" (this needs to be generated by debian/rules), iterate over all entries, and remove any which are a symlink. If it's a dpkg bug it would really be better to fix it there (needs a reproducer then), if it's really a hardware bug then no amount of software workarounds will help against those :/. But if you see many duplicates, it's more likely a dpkg bug.
Gunnar, "quite often" sounds worrysome -- this smells like a dpkg bug then? Did you see more duplicates of this?
This is an utterly complex and probably brittle workaround. If we need such a workaround, then the pre*inst* should call "dpkg -L <pkgname>" (this needs to be generated by debian/rules), iterate over all entries, and remove any which are a symlink. If it's a dpkg bug it would really be better to fix it there (needs a reproducer then), if it's really a hardware bug then no amount of software workarounds will help against those :/. But if you see many duplicates, it's more likely a dpkg bug.