On Thu, Apr 7, 2022 at 5:35 PM Samuel Sieb <samuel(a)sieb.net> wrote:
On 4/7/22 14:51, Jared Dominguez wrote:
> On Thu, Apr 7, 2022 at 3:49 PM Samuel Sieb <samuel(a)sieb.net
> <mailto:samuel@sieb.net>> wrote:
>
> On 4/7/22 08:02, Jared Dominguez wrote:
> > This is a proposal. Nothing has changed yet. The choice is now
> whether
> > to go forward with it or come together with a cohesive
> > alternative, including one of the two listed in the proposal. But we
> > need a solution that accounts for the existing maintainers not
> having
> > capacity to continue maintaining legacy code. I've seen responses
> from
>
> I haven't yet seen a clear answer about what code is "rotting"
and
> which
> legacy code is too hard to maintain. Is there something actually
> broken
> right now?
>
>
> For one, syslinux hasn't seen an update in 3 years and a release in 7
> years, and it has outstanding bugs. Legacy boot isn't where grub2 is
> getting development attention. The current maintainers in Fedora won't
> have capacity to continue maintaining legacy boot support in Fedora. As
> grub2 continues to be developed for UEFI systems (ARMv8-9 and x86-64,
> not to mention non-UEFI ppc64le and s390x), there is added risk of
> regressions on legacy x86 boot that won't be getting developer attention.
I don't understand why we're still using syslinux instead of grub for
legacy boots, especially since I think now you can use the same grub.cfg
file for both.
syslinux (really isolinux) is only used on installation media.
This comment from 2012 suggests we probably should have dropped
isolinux in favor of BIOS GRUB a while ago, but c'est la vie. I think
it's in scope to exchange syslinux for GRUB in the F37 cycle, pending
review of the effort required to do it.
https://mjg59.dreamwidth.org/11285.html?thread=345621#cmt345621
There is always a risk of regressions, but if there is
no current problem, then why is there this push to obsolete a lot of
active hardware? This is not comparable to the 32-bit removal where it
was only a few really old systems. This is going to affect decent
systems that are less than 10 years old. I have a work HP laptop from
2012 that has "experimental" EFI support that really doesn't work well
and possibly a newer one as well, but I can't check it right now.
Given that Microsoft has had an exception for UEFI being required on
server hardware up until barely two years ago, I suspect RHEL folks
are going to have customers asking for reconsideration.
--
Chris Murphy