Re: port-vax/59265

["Jan-Benedict Glaw via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR port-vax/59265; it has been noted by GNATS.

From: Jan-Benedict Glaw <jbglaw%lug-owl.de@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: port-vax/59265
Date: Fri, 6 Nov 2026 17:21:57 +0100

 --mUiJJltLYLLTRGHj
 Content-Type: text/plain; charset=utf-8
 Content-Disposition: inline
 Content-Transfer-Encoding: quoted-printable
 
 Hi!
 
 This has two sides: First of all, VAX is pre-752 fp and does not know
 about negative zero. And then there is:
 
 $ cat lib/libc/arch/vax/gdtoa/arith.h
 /* $NetBSD: arith.h,v 1.1 2006/01/25 15:33:28 kleink Exp $ */
 
 #define VAX
 #define NO_HEX_FP       /* XXX */
 
 
 =2E..which disables this feature completely. So "-0x0" is a successfully
 parsed sign, a successful zero (this combination, given that VAX has
 no negative zero is a bad situation by itself---however, this seems to
 be treated as a leading zero, so no actual issue), but then we abort on
 'x' as that's not recognized at all. That we're getting a zero out of
 the double is pure coincidence.
 
   OTOH, gethex() is there and I think the differences are not too bad
 here. That can be implemented, matching the usual differences (nan,
 inf, signed zero) as reasonable as possible.
 
 As for the signbit() check, I'd go for skipping them on VAX.
 
 MfG, JBG
 
 --=20
 
 --mUiJJltLYLLTRGHj
 Content-Type: application/pgp-signature; name="signature.asc"
 
 -----BEGIN PGP SIGNATURE-----
 
 iF0EABECAB0WIQQlDTvPcScNjKREqWEdvV51g5nhuwUCau3+ogAKCRAdvV51g5nh
 u6DmAJ9WwPj3oKlz19KQoIHw589wcLASaACeOPIUSHJFKHYTF1ZgABjhucrzBvk=
 =cuoz
 -----END PGP SIGNATURE-----
 
 --mUiJJltLYLLTRGHj--