Re: GNU tools netbsd vs netbsdelf

On Wed, Jun 16, 2021, at 3:16 AM, David Holland wrote:
> On Fri, Jun 11, 2021 at 07:46:43PM -0400, John Ericson wrote:
> > > You'll also find that the upstream configs are incomplete/outdated
> > > with respect to the ones in the NetBSD tree (because we tend not to do
> > > a very good job of coordinating with upstream) so I have some doubts
> > > about whether they can in general be expected to work without
> > > editorial intervention.
> > 
> > Good to know. I guess if anything that helps me? An expectation of
> > some downstream divergence means I wouldn't have to
> 
> Guess you didn't finish that chunk? :-)

Oops! Yes I just meant a reconciliation between upstream and downstream that is more async gives me some nice flexibility: everything doesn't need to work everywhere all the time.

> Anyway it's not so much that it's expected (it would be better if it
> didn't happen) as that upstreaming adjustments is a pain and tends not
> to happen.

Well, it's been a few weeks but things did move along upstream. netbsdpe, per the conversation in the other thread is slated for removal (there is a 1 release deprecation cycle for Bintuils). And the netbsd new-default change is tentatively approved with some conditions in https://sourceware.org/pipermail/binutils/2021-July/117378.html

The main condition is the nature of ns32k support in NetBSD:
> One problem with that - the NS32K target.  As far as I can see this target only supports aout format NetBSD, not ELF...
> 
> Possibly NS32K support in NetBSD is entirely deprecated.  If so, then we can add the ns32k-netbsd target to the obsolete list.  But I would like to have you confirm this first.

So, my main question (sorry this is a bit buried amid the quotes) is, what is the nature of NS32K support in NetBSD?
 * Is it deprecated/removed?
 * If not, does it in fact support ELF after all?
 * If not, should this be a show-stopper to the treatment of netbsd as netbsdelf instead of netbsdaout, or is it fine if ns32k-netbsdelf needs to be use?
> They aren't organized at all; they're completely ad hoc. Everything
> would benefit from systematizing the configury and shoveling out the
> mess of arbitrary #defines and #undefs and subtly order-dependent
> declarations that get spit out. But that's a huge job

Oof, I do empathize with that.

> and I get the feeling it would be pushing in the opposite direction from upstream.

Well, so far I feel my efforts have gone pretty well. Just the expected latency of coordinating disparate groups.

John