Instead of proceeding with blinders on, like many other aspects of systemd, please don't ignore your out of focus items. Sure, don't implement for them, but do some investigative plan for dealing with them eventually if should be needed.
Project Managers call that "risk management". I sometimes get the vibe that Systemd calls that "wontfix".
I've been using bootloader specifications in a number of embedded system projects and I really like it.
The only configuration that's required for you in the bootloader is to select a boot medium and bootspec
does everything else. barebox takes it one step further and supports a linux-appendroot option, which
automatically appends a file-system specific root= option to the kernel command line. That way you can
have multiple slots with the same bootspec (e.g. for A/B booting scenarios) and you can transparently
boot the same rootfs whether it's locally on SD, NAND or even on NFS. Good stuff.
(Disclosure: I do get paid to work on barebox. This post is on my own dime though :^)
I have worked on creation of custom boot menu entries and this is basically a huge pain in existing technologies, like grub2. If you have multiple OS, every one of them must know, even a little bit, about every other one. This means that innovation in the early boot section is much harder than it needs to be, unless you want to take over entire computer.
It would be great if someone else had come up with the bootloader with “drop-in files” mechanisms, and pushed it to every major distro, but that did not happen in the last 30 years, so it is unlikely to happen anytime soon.
And yeah, I wish Poettering would make few new repos instead of pushing all under “systemd” umbrella... but it’s not a big deal, given he is finaly solving this major annoyance.
4 comments
[ 2.4 ms ] story [ 33.8 ms ] threadProject Managers call that "risk management". I sometimes get the vibe that Systemd calls that "wontfix".
(Disclosure: I do get paid to work on barebox. This post is on my own dime though :^)
I have worked on creation of custom boot menu entries and this is basically a huge pain in existing technologies, like grub2. If you have multiple OS, every one of them must know, even a little bit, about every other one. This means that innovation in the early boot section is much harder than it needs to be, unless you want to take over entire computer.
It would be great if someone else had come up with the bootloader with “drop-in files” mechanisms, and pushed it to every major distro, but that did not happen in the last 30 years, so it is unlikely to happen anytime soon.
And yeah, I wish Poettering would make few new repos instead of pushing all under “systemd” umbrella... but it’s not a big deal, given he is finaly solving this major annoyance.