From: bd733 AT rgfn DOT epcc DOT edu (Jason M. Daniels)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Possible bug?
Date: 4 Jun 1997 16:48:19 GMT
Organization: The Rio Grande Free-Net, El Paso Community College, El Paso, TX
Lines: 27
Distribution: world
Message-ID: <5n468j$2jt@news.epcc.edu>
References: <Pine DOT BSI DOT 3 DOT 91 DOT 970603215937 DOT 1369A-100000 AT host DOT hawk DOT igs DOT net> <5n3dqh$s3g AT bagan DOT srce DOT hr> <5n3i63$145 AT star DOT cs DOT vu DOT nl>
NNTP-Posting-Host: rgfn.epcc.edu
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

Ruiter de M (mdruiter AT cs DOT vu DOT nl) wrote:
> Simun Mikecin (sime AT fly DOT cc DOT fer DOT hr) wrote:
> : Philip Ibis (pibis AT host DOT hawk DOT igs DOT net) wrote:
> : :   double n;
> : :   memset(&n,0xFF,sizeof(double));  /* contrived, sure */
> : :   printf("%f\n",n);
> :             ^^
> : Have you tried using %lf here?

> Why? According to the docs, %hf is for floats, %f is for doubles and
> %lf is for long doubles (what's that? same as doubles, for DJGPP?).
> Scanf differs.

> : P.S. I think that conforming to ANSI %f is for float, %lf for double and %Lf
> : for long double. But since the way these are implemented on x86 platforms it
> : ussualy works with any format.

> Ummm, I don't think so. But then again, who am I? :)

No, the way DJGPP impliments it is quite correct. This is, in fact, 
discussed in the comp.lang.c FAQ.

--
Jason Daniels -- bd733 AT rgfn DOT epcc DOT edu
---> BELIEVE THE LIE <---
Linux: The choice of a GNU generation.
Winblows 95: The world's best-selling computer virus.