Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!wasatch!cmos.utah.edu!jacobs
From: jacobs%cmos.utah.edu@wasatch.utah.edu (Steven R. Jacobs)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Turbo-C bug
Message-ID: <1956@wasatch.utah.edu>
Date: 30 May 89 23:02:11 GMT
References: <2556@Portia.Stanford.EDU>
Sender: news@wasatch.utah.edu
Distribution: na
Organization: University of Utah CS Dept
Lines: 28

In article <2556@Portia.Stanford.EDU> hobbes@Portia.Stanford.EDU (Andrew Zimmerman) writes:
>The following program doesn't run when compiled under Turbo-C 2.0.
>(It does compile under Turbo-C, and it runs under Zortech C.)
>Does anyone know if you can explicitly tell Turbo-C 2.0 to link in
>the floating formats?

I don't know about explicitly, but you can fake out the compiler as
shown in the modified example below.  I've seen problems similar to
this in Turbo-C 1.5 as well as 2.0, and they have problems getting
it right with structures that have float/double elements.

#include <stdio.h>
main()
{
    double a[3][3], tc_kludge;
    int i;

    i = 1;
    scanf("%lf", &tc_kludge);
    a[i][1] = tc_kludge;
    printf("a[1][1] = %lf\n",a[1][1]);
}

This trick need only appear at one point in the program, so luckily
you don't need to modify all 18,324 scanf() calls :-)


Steve Jacobs  ({ihnp4,decvax}!utah-cs!jacobs, jacobs@cs.utah.edu)