On Sat, Jan 12, 2019 at 01:51:41PM +0000, Colin Watson wrote: >On Sat, Jan 12, 2019 at 12:58:52PM +0000, Steve McIntyre wrote: >> NB: Ubuntu doesn't have the depends/recommends here, so I can only >> assume that some other method is used to ensure that shim-signed is >> installed there. I asked Steve Langasek about this, but I've not had >> an answer yet. > >I guess I can see the argument about having a recommendation here, since >GRUB does make use of shim if it's installed. > >Ubuntu instead does this with a grub-installer change that explicitly >installs both grub-efi-amd64-signed and shim-signed in the amd64/efi >case. What do you think about going with that instead, or maybe in >addition to your approach? Unpicking the patch I think it's something >like this:
I'd possibly do both? grub-installer is one obvious place to make d-i add things (and the patch there also looks fairly obvious, roughly what I also considered myself), but that *only* works for d-i. If we add the Recommends: for shim-signed, people will get a reasonable package set by default if they just install grub-efi-amd64-signed. Basically, I think encoding that in the normal package relationship with the Recommends: is the right thing to do anyway. -- Steve McIntyre, Cambridge, UK. st...@einval.com < liw> everything I know about UK hotels I learned from "Fawlty Towers"
