X-Sybari-Trust:	0f73cef7 9ffcebbb 7a95d2f4 00000138
From:	Martin Stromberg <Martin DOT Stromberg AT epl DOT ericsson DOT se>
Message-Id:	<200303180839.JAA03639@lws256.lu.erisoft.se>
Subject:	Re: strto{d,f,ld}, inf and nan patch
To:	djgpp-workers AT delorie DOT com
Date:	Tue, 18 Mar 2003 09:39:15 +0100 (MET)
In-Reply-To:	<6137-Tue18Mar2003092204+0200-eliz@elta.co.il> from "Eli Zaretskii" at Mar 18, 2003 09:22:05 AM
X-Mailer:	ELM [version 2.5 PL3]
MIME-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

Eli said:
> I think we should at least verify that <anything> is a hex number.

Nope. Or at least probably not. Because the highest hex number we can
read is an unsigned long long. That covers the mantissa of a long
double. Where should the sign bit be indicated?

Lately I've been thinking "nan(s,m,e)" while printing and with the
",e" part optional while scanning (it'll be ignored anyway). s ==
sign, m == mantissa, e == exponent.

Or we could always omit "e". I just thought it nice to give the user
the whole story.

> Ideally, we should also generate the precise bit pattern as given by
> <anything>.  It cannot be too hard.

Yes. All we need to agree on the format and if somebody else does the
printing part, I'll do the strto{f,d,ld}() part.

> Btw, what does C9x say about <anything> that specifies a bit pattern
> which is not a NaN?

I don't have the standard here but I'm pretty certain it must
generate a NaN. That's why the "e" above will be ignored. And if "m"
is 0, set it to some other documented non-zero value (otherwise we'd
get Inf).


Right,

						MartinS

- Raw text -