What I think I understand from this case, is that part of rationale for
this work, is that there are emerging platforms where supporting *any*
32-bit userland is problematic or impossible. Part of the goal, then,
is to utilize SPARC to enforce that we always have a 64-bit version of
all of the userland tools (or at least all of the core tools).
I've confirmed this via IRC with Roland.
In order for that to be meaningful, I think we'd have to have a new
policy requiring that at least tools required for single user (and
probably a significant set beyond that) be 64-bit clean. That sounds
like a new Big Rule, requiring a written opinion to me.
So, I'm pressing the derail button.
Project team (Roland), *please* don't despair. Derailing doesn't mean
that the case will not be successful, it only means that we need to
understand the full scope of the problem, discuss it, and write a
written opinion, possibly resulting in the creation of a new Big Rule.
What I'd like to see in full case materials to be presented is:
1) the above rationale made explicit
2) some definition/clarification for deciding whether the Big Rule
applies to a project or not (the boundary probably extends beyond single
user mode, but how far beyond?)
3) an initial list of tools (not in e-mail, but an itemized list
deposited in the case directory)
4) analysis of big-picture performance issues, if any (does 64-bit run
faster, or slower? maybe run a some test of boot time analysis.)
5) a rough estimate in the increase of the size of the media/miniroot
for the SPARC port
6) some thought given to future views on amd64 -- will such a direction
ever make sense for amd64 as well?
7) if you are going to use a phased approach to delivery (fixing a few
binaries at a time), as I suspect you will want to (just to make the
problem tractable), then a statement to that effect. (that will avoid
you needing to come back to ARC again for this same issue.)
I realize that's a non-trivial amount of effort, but I think at the end
of the day, it will result in higher quality decision making.
-- Garrett
