> Richard Earnshaw <rearnsha@arm.com> writes:
> > Erm, radical suggestion time.  Rather than trying to fix up what is 
> > increasingly apparent as a "botch" job, why don't we migrate the ARM port 
> > to using ELF.  I'm now in a position where I can make details of the ARM 
> > ELF format available and we can do the job properly and (maybe) become 
> > more compatible with the rest of the arm world at the same time.
> > 
> > Comments?
> 
> "And that would be done when?!"  (Yes, that's just rehetorical.)
> 
> That's well and good 6 weeks, or 2 or 3 months from now when it's
> actually done.  I can imagine that it could be done sooner, but i
> think a quality job, to make sure it's all tested, etc., would
> probably take at least that long.  And if you're really going to try
> to become compatible -- not just slightly more compatible -- with the
> rest of the ARM world (structure alignment is what i'm thinking of),
> then it'll probably take longer than that.
> 
> In the mean time, everybody who builds -current from fresh
> NetBSD-current sources loses.  Sure, they've been losing for a long
> time, but it's _really_ silly for them to be told "lose a while
> longer, we have a good fix in hand but don't want to use it."
> 
> 
> I would bet that there's no real need for people to even install from
> binary snapshots, if they're willing to build two iterations of the
> world (one all static, then one with all the new tools)...

Yeah, Yeah, all valid points, but my point is: is it really worth 
expending a lot of effort unravelling the existing mess (especially if it 
creates an incompatibility) if a few further hacks would keep the old 
world running until a new one (elf based) can replace it in its entirity?

I'm not in a position to make the decision for anyone, just trying to 
cover the options.

Richard.