I’m not convinced that this is actually simpler. I’m also confused because Red Hat already was trying to simplify booting with systemd-boot. Are they no longer happy with their existing approach?
No more boot loader: Please use the kernel instead
Submitted 4 months ago by bot@lemmy.smeargle.fans [bot] to hackernews@lemmy.smeargle.fans
https://pretalx.com/devconf-cz-2024/talk/W3AVCT/
Comments
henfredemars@infosec.pub 4 months ago
pizza_the_hutt@sh.itjust.works 4 months ago
Speaking of systemd boot, I tried it on a recent install, and my system only booted properly about 20% of the time. I switched back to grub, and it’s been rock solid since.
timbuck2themoon@sh.itjust.works 4 months ago
I’d be curious what the issue was.
I’ve used grub for years but am curious about systemd-boot as it’s supposed to be far simpler/easier.
HakFoo@lemmy.sdf.org 4 months ago
The “looks for other operating systems” is huge for any desktop use with dualboot, recovery media, and plenty of systems have utilities like Memtest86 through the boot menu.
Grub always seemed weirdly complex yet incomplete- the idea that you had to pre-enumerate every bootable OS made it worse than rEFInd to me.
lvxferre@mander.xyz 4 months ago
Yeah, nah. This only throws more complexity under the rug.
Among other stuff, whatever is booting your computer needs to 1) find the kernels that you have, 2) find any other OS that you might have, and 3) allow the user to pick one of those. You can either use a specific tool for that (bootloader) or dump those roles into the kernel, but you can’t get rid of them without breaking a lot of stuff.
Regardless of the above, the owner of a device should be able to turn secure boot off; devices not allowing so are broken by design, to prevent your full ownership over it.
Part of the complexity is intrinsic, as explained. And if you’re concerned about the additional complexity from the implementation, the solution is a different bootloader, not ditching the concept altogether.
I stopped watching the video at 8:22, as she was talking about bugs.
The simpler solution is to simply pour more development into GRUB2, not to throw the problem into the kernel devs’ hands, as if it was some sort of hot potato.