Hello,

Can anyone explain why difftime return zero seconds difference between two different times?

Thank you,
Neil

#include <sys/time.h>

int main(void)
{
time_t s, f;

time(&s);
sleep(10);
time(&f);
printf("%s\n", ctime(&s));
printf("%s\n", ctime(&f));
printf("%f\n", difftime(f, s));
return 0;
}

If you compile the program with optimalization (gcc -O3 prog.c) you'll get the right result. It's strange, don't know why it works this way.

you should include time.h not sys/time.h, then you'll get the right result

Thank you both for your replies.

Including <time.h> instead of <sys/time.h> has fixed the problem. I guess I should have been following my
Kernighan and Ritchie more closely.

Compiling with -O3 and <sys/time.h> gives a non-zero, but incorrect answer of 7000 odd seconds
for an actual delay of 10 seconds! I've no idea what's going on here either. (Compiling with -O3 and
<time.h> works properly.)

So problem solved. Thank you very much for your help.

However, I need to use gettimeofday() from <sys/time.h>. (I need it as it gives access to the microsecond
accuracy timer.) I've included both <time.h> and <sys/time.h>, and the program compiles and runs properly.
However, I was worried that difftime() would be defined twice.

Looking at <sys/time.h> I see it includes <time.h> itself and does not separately define difftime() so
I don't think my including both header files is a problem. (In fact, according to "grep -r difftime *" run in
/usr/include, difftime only occurs in time.h) But that still leaves the puzzle of why difftime does
not work properly when only <sys/time.h> is included.