Replying off list, because I don't think it matters much for the RT discussion.
>>>>> "Russ" == Russ Allbery <r...@debian.org> writes: Russ> Yes, I'm probably understating the difficulty of making this Russ> change in practice inside image building software as it's Russ> currently constructed. Russ> My concern about changing mkfs options is that I worry that Russ> this would be a constantly-changing target that has to be Russ> synchronized across multiple pieces of software that are not Russ> currently well-aligned with file system development. Sadly, supporting a new distribution in image building software is a constantly changing target that requires small tweaks all the time. Carthage has a function debian_container_to_vm that uses FAI to turn a directory tree into a VM image. For bookworm, we end up needing to deal with systemd-resolved getting split. There was also some change in recent memory related to apt-transport-https and when it was needed. It's mostly small stuff, but you end up having a series of tweaks for the guest system, no matter how much you wish you didn't. There are similar tweaks at the debootstrap level, at the FAI level, and even in the container frontends to a small extent. I absolutely agree mkfs compatibility options would be great, but chasing down how to call mkfs is par for the course. I haven't looked much at vmdb2 because I disagree with the design philosophy, but I think it has tweaks too.