hbo@sbphy.physics.ucsb.edu (Howard B Owen) (02/26/91)
I have a problem that I seem to recall was discussed in this group.:
Environment: Mips cc v 2.1/Ultrix v4.0/DECstation 5000
Problem: Output of doubles is garbled when mixed with other i/o statements.
Example:
#include <stdio.h>
main()
{
FILE *outfile = fopen( "quantum.dat", "w" );
double energy = 0.01;
printf("%f\n", energy);
printf("%f\n", energy);
fclose (outfile);
}
This program, compiled with "cc bug.c -o bug", produces the following output:
0.010000
19991592789737117000000000000000000000.000000
If the fopen and fclose statements are removed, or if a value is
assigned to "energy" after the first printf, Then the output is as
expected. Also, the following program displays a similar problem:
#include <stdlib.h>
#include <stdio.h>
main() {
float a= -0.1;
double mu = -0.1;
printf("%15.3e\n",mu);
printf("%15.3e\n",a);
printf("%15.3e\n",mu);
}
The output from this program, compiled like the last one, is:
-1.000e-01
-1.000e-01
-2.353e-185
I can't believe such a severe bug could be part of a correctly
configured system. We have
FORTRAN v 2.1 installed, as the following excerpt from setld -i shows:
F77BACKEND021 installed V2.1 Compiler Back End
F77BASE021 installed F77 Programming Environment
F77MAN021 installed F77 Manual Pages
F77UNSUP021 installed F77 Unsupported Utilities
Any help you could offer would be appreciated.
Thanks,
--
Howard Owen, Computer Systems Manager internet: hbo@sbphy.physics.ucsb.edu
Physics Computer Services BITNET:
HBO@VOODOO.BITNET
University of California, Santa Barbara DECNET Internet:
45180::HBO
"I am not a pay TV service!" Telephone:
805-893-8366 (work)