On Thursday, July 2, 2020 1:30:41 PM MST Brandon Nielsen wrote:
On 7/2/20 3:19 PM, Martin Jackson wrote:
>> 5-10 years? A better estimate would be 15-20 years. People aren't
>> going to
>> throw away perfectly fine systems and jump to new "cloud" platforms
>> because the OS they were using dropped BIOS support. They'll just stop
>> updating, and likely move to something that is still supporting BIOS,
>> if they
>> don't write their own installer and just continue using Fedora, given
>> this is an entirely artificial limitation.
> While I completely hear you on the fact that people will often sweat
> assets for years longer than accounting schedules suggest they should,
> do you really think they're going to write custom installers??? I think
> it's far more likely that they would move to other distros more amenable
> to supporting the hardware they have.
> There are many distros that cater to this kind of market already, some
> by design and some by inclination.?? I don't think we want to drive them
> For what it's worth, I do not think that removing legacy BIOS support
> from Fedora is the right thing to do.?? I don't see significant benefit,
> and I see lots of potential harm.
I don't think removing BIOS support _today_ is the right answer either.
I have BIOS only hardware kicking around, and quite a bit of my UEFI
hardware still supports legacy BIOS booting as well (though I don't use
However, I'm concerned about UEFI feature development / quality
assurance being held hostage by BIOS support for, based on above
comments, 5 to 20 years? Surely as a somewhat leading-edge distribution,
we need to start thinking about some kind of post-BIOS world.
Perhaps one small step toward that future would be enabling systemd-boot
on new UEFI installs, relegating GRUB2 to BIOS and upgrade installs
only? This split configuration could hang around until support for GRUB2
/ BIOS wanes to the point it can no longer stand under its own weight
(much like 32bit install media).
GRUB2, which is a UEFI bootloader as well, is a far superior bootloader to
systemd-bloat, and it supports usecases that are supported by Anaconda (the
Fedora installer framework) that systemd-bloat doesn't, as addressed elsewhere
in this thread by myself and several others. There is no way that supporting
BIOS can be a cause for UEFI feature development being "held back". It's got
nothing to do with UEFI stuff.
John M. Harris, Jr.