booster VS set

People should stop using plain passwords to protect data and switch to something better.oster/ + Yubikey

Change from mkinitcpio to booster. It's faster and produce smaller images.

Neither, use booster.

Booster is an initramfs - a type of software that runs early during the boot process and helps to setup system, e.g. perform disk unlock, load extra drivers, setup ZFS filesystem, and so on.

I don't have high hopes that this will change in mkinitcpio, but perhaps you could bring it up with booster (which copied the mkinitcpio hooks and removal scripts)

This is my personal preference, but consider switching to booster.

initramfs: booster - TPM2 support, way faster than mkinitcpio, autodiscovering root partition

Note I'm using booster, so assuming you're using the default mkinitcpio you probably want /boot/initramfs-linux.img. Also, don't forget to modify if your ESP isn't mounted at /boot like mine.

Booster initramfs generator supports TPM, Yubikey and Network binding so you can easily protect your data using the strategy you want.

GPT table detection has higher priority at booster code. But for some reason it does not identify your partition table as GPT. Filed a github bug for it https://github.com/anatol/booster/issues/119

You might want to look at https://github.com/nofeaturesonlybugs/set

I have just released v0.5.0 of package set available @ https://github.com/nofeaturesonlybugs/set

Sorry for two replies but I ran my TypeInfo benchmarks with and without caching if you're interested in also sharing the impact. The benchmarks are in https://github.com/nofeaturesonlybugs/set/blob/master/type_info_test.go

It's generally not well received to use reflect to step around Go's type system but there are also instances where the type system starts to violate DRY. I've put together a small package that uses reflect to work some magic on types @ https://github.com/nofeaturesonlybugs/set