(background, Manoj asked for opinions on #debian) I think I agree that the stuff under /var/lib/initramfs-tools is internal state specific to initramfs-tools and should be purged by initramfs-tools. (as distinct from the initrd itself, which is what the user/linux-image/kernel-package asked to be created). Since it can be reproduced without kernel-package (using debian images or make; make install) it's definitely not filed against the right package atm. I'd have expected downgrading to minor and reassigning it back to initramfs-tools, but I'll leave that to the maintainers to sort out. Some snipped from IRC when we were discussing it: 17:24 < mjj29> I think minor against initramfs-tools seems reasonable 17:25 < Manoj> now, arguably, if the liunx-image runs update-initramfs on install, it should run it again on removal, but that is a separate bug 17:25 < mjj29> "the initramfs doesn't belong to initramfs-tools" <-- but the sha1sum does, and the initramfs won't stop working if it's removed 17:25 < mjj29> initramfs-tools might just get a little confused, but that's fine 17:26 < mjj29> install initramfs-tools; install linux-image-foo; purge initramfs-tools; install initramfs-tools 17:26 < mjj29> should leave you in the same situation as 17:26 < mjj29> install linux-image-foo; install initramfs-tools -- Matthew Johnson
Attachment:
signature.asc
Description: Digital signature