timk@xenitec.on.ca (Tim Kuehn) (03/01/90)
I'm working on a program and am having a problem with the atof() function.
Every time I run it with a certain option turned on I get an error:
scanf : floating point formats not linked.
The following code sample generates the same error. I've RTFM'd and there's
not a HINT of anything else that should be brought in in terms of declerations,
etc.
Anybody have any ideas?
-------code sample here-------
#include <stdio.h>
#include <math.h>
#include <conio.h>
char arg[10];
main()
{
clrscr;
printf("start \n");
strcpy(arg, "-f3");
printf("%s \n", arg);
printf("%lf \n", atof(&arg[2]));
}
+-----------------------------------------------------------------------------+
|Timothy D. Kuehn timk@xenitec.on.ca |
|TDK Consulting Services !watmath!xenitec!timk |
|871 Victoria St. North, Suite 217A |
|Kitchener, Ontario, Canada N2B 3S4 (519)-741-3623 |
|DOS/Xenix - SW/HW. uC, uP, DBMS. Quality SW Guaranteed |
+-----------------------------------------------------------------------------+srt@maui.cs.ucla.edu (Scott Turner) (03/01/90)
In article <1990Feb28.210122.24511@xenitec.on.ca> timk@xenitec.UUCP (Tim Kuehn) writes: >I'm working on a program and am having a problem with the atof() function. >Every time I run it with a certain option turned on I get an error: >scanf : floating point formats not linked. In the future, you should say what compiler you're using, what machine, etc. Fortunately this problem is so common that it isn't necessary. The problem is that TurboC doesn't link in the floating point lib unless it thinks it is needed. And basically TurboC thinks it will be needed if it sees a floating point function being used. So to force TurboC to link in the floating point library, put the pointless line: (void) exp(1.0); somewhere in your code. That should do it. Scott R. Turner UCLA Computer Science "It's so bad it's not even wrong" Domain: srt@cs.ucla.edu UUCP: ...!{cepu,ihnp4,trwspp,ucbvax}!ucla-cs!srt
mac@idacrd.UUCP (Robert McGwier) (03/02/90)
From article <32397@shemp.CS.UCLA.EDU>, by srt@maui.cs.ucla.edu (Scott Turner): > In article <1990Feb28.210122.24511@xenitec.on.ca> timk@xenitec.UUCP (Tim Kuehn) writes: >>I'm working on a program and am having a problem with the atof() function. >>Every time I run it with a certain option turned on I get an error: >>scanf : floating point formats not linked. > > The problem is that TurboC doesn't link in the floating point lib > > (void) exp(1.0); As I am fond of saying, when you pay $99 for a compiler, what do you expect? ;->). I have heard myriad rumors (so what's new?) that Turbo and Waterloo are (FINALLY) working on an optimizing compiler with Turbo's front end and Waterloo's output (or some rational combination thereof), this might be a Turbo C product that I would finally believe is worth at least $99. Can anyone lend credence to this rumor? Bob -- ____________________________________________________________________________ My opinions are my own no matter | Robert W. McGwier, N4HY who I work for! ;-) | CCR, AMSAT, etc. ----------------------------------------------------------------------------
timk@xenitec.on.ca (Tim Kuehn) (03/02/90)
In article <1990Feb28.210122.24511@xenitec.on.ca> timk@xenitec.UUCP (Tim Kuehn) writes: >I'm working on a program and am having a problem with the atof() function. >Every time I run it with a certain option turned on I get an error: >scanf : floating point formats not linked. > >The following code sample generates the same error. I've RTFM'd and there's >not a HINT of anything else that should be brought in in terms of > declerations, etc. > >Anybody have any ideas? > >-------code sample here------- >#include <stdio.h> >#include <math.h> >#include <conio.h> > >char arg[10]; >main() >{ >clrscr; >printf("start \n"); >strcpy(arg, "-f3"); >printf("%s \n", arg); >printf("%lf \n", atof(&arg[2])); >} > I've gotten a number of replies on this one, and it seems that TC 2.0 has a problem of being TOO zealous in optimizing spacewise, namely in that it optimizes links to the floating point emulation pkg right out if it doesn't see something like i9 = exp(1.0); or any other mathematical type equation in the program. (Use of the atof() function doesn't count as a mathematical function requiring floating point evidently) Thanks to all those who wrote, adding the above line (and a double i9 declaration) fixed the problem quite nicely. +-----------------------------------------------------------------------------+ |Timothy D. Kuehn timk@xenitec.on.ca | |TDK Consulting Services !watmath!xenitec!timk | |871 Victoria St. North, Suite 217A | |Kitchener, Ontario, Canada N2B 3S4 (519)-741-3623 | |DOS/Xenix - SW/HW. uC, uP, DBMS. Quality SW Guaranteed | +-----------------------------------------------------------------------------+