From: kunst AT prl DOT philips DOT nl
Subject: Re: Long floats
To: morphine AT whatever DOT cs DOT jhu DOT edu (Michael Phelps)
Date: Mon, 27 Jun 1994 10:41:32 +0100 (METDST)
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu (DJGPP users list)


Michael Phelps (morphine AT cs DOT jhu DOT edu) writes:

> Is there such number type as "long float" in djgpp?  I looked through
> the gcc docs and found a "long long" but not a "long float".  
> What I need is an extended (>double) precision floating point type.  
> I tried "long double" and the compiler compiled just fine; 
> the only problem was that I got a "floating point exception" when I 
> tried printing it with printf("%f") or "%lf".  Is it possible?

gcc supports the 'long double' type. You can do calculations with it.
However, printing 'long double' values using printf is not supported yet.
According to K&R (p.245) the conversion character is an extra 'L' in
front of the 'e', 'f' or 'g' conversion character, e.g. printf ("%Lg", ld);
Maybe 2.6.x will include this.

Pieter Kunst (kunst AT prl DOT philips DOT nl)