Re: pkgsrc: python310 issues -> no ca-certificates

[David Brownlee <abs%netbsd.org@localhost>]

On Tue, 18 Jul 2023 at 11:15, Jan-Benedict Glaw <jbglaw%lug-owl.de@localhost> wrote:
>
> Hi!
>
> ...and `nextafter` belongs to ${COMMON_SRCS} and is IEEE754 specific.
> So I wonder: Should we implement a VAX version? Can probably be done,
> though some care must be taken as D- and G-float can, in theory, be
> choosen as a gcc command line option. Or drop the declaration?

I think we can probably accept supporting a single VAX FP format (I
think the odds on someone porting NetBSD to the uVAX I and adding the
infrastructure for separate F+G and F+D builds... are low :)

>   Another route could be to accept that VAX FP is too dead and go for
> soft float all the way. :-/

As Johnny mentions, most of the VAX FP vs IEEE 754 differences are not
relevant for general use of most software. They mainly only trigger in
test cases, scientific and overly clever code. At this point we should
be happy with something "good enough" to get most software building
and running on VAX (with some patching to avoid some tests where
required).

Anyone with FAX FP-foo up for creating VAX variants of
lib/libm/src/s_nextafter{,f,l}.c ? :)

David