Date: Mon 24 Aug 92 19:26:35
From: ronis AT ronis DOT chem DOT mcgill DOT ca (David M. Ronis)
To: djgpp AT sun DOT soe DOT clarkson DOT edu
Cc: ronis AT ronis DOT chem DOT mcgill DOT ca
Subject: Bug & fix in %g format

Hi DJ,

Here is the patch for the bug I reported with %g format.  Recall that
trying to print something like 1.000001 using %13.5g format resulted
in 1.000001, not 1., (the latter I believe is correct).  The problem
is in the routine cvt in the file libsrc/c/io/doprnt.c.  In short,
the 0's between the two 1's were not counted as significant.  

If I call the fixed and original versions doprnt.new and
doprnt.old, respectively.  Here is the result of running 

		      diff doprnt.old doprnt.new


======================================================================
601,603c601
< 		if (fract) {
< 			if (prec) {
< 				do {
---
> 				while (prec && fract) {
606,609c604
< 				} while(!tmp);
< 				while (--prec && fract) {
< 					fract = modf(fract * 10, &tmp);
< 					*t++ = tochar((int)tmp);
---
> 					prec--;
611d605
< 			}
615d608
< 		}

======================================================================


David Ronis