Same C code different results

jlliagre · 07-24-2004, 12:27 AM

OK, sorry, but as the input is discarded anyway, one getchar is simpler than gets, and is enough for the job.

itsme86 · 07-24-2004, 02:13 AM

Quote:

OK, sorry, but as the input is discarded anyway, one getchar is simpler than gets, and is enough for the job.

No, I'm sorry but there isn't always just one character left on the buffer. Here's proof:

Code:

#include <stdio.h>

int main(void)
{
  int num;
  char buf[20];

  printf("Number: ");
  scanf("%d", &num);
  fgets(buf, sizeof(buf), stdin);
  printf("%d:%s:\n", num, buf);

  return 0;
}

And when I run it:

Quote:

itsme@itsme:~/C$ ./scanftest
Number: 15abc
15:abc
:
itsme@itsme:~/C$

If you only used one getchar() it would pull the 'a' off the buffer but leave 'bc\n'.

You can't trust the user to type only what you're expecting them to type, so I stand behind my argument that using fgets() with a large enough buffer or a loop that calls getchar() until \n is reached is the best tool for the job and that a single getchar() is sadly insufficient. This is why I suggested just using fgets() in the first place and forgetting all about scanf() since you have to worry about pulling everything off the buffer afterward anyways. scanf() should only be used for trusted input.

jlliagre · 07-24-2004, 03:26 AM

itsme86, it seems you didn't read my replies.

I understand your point, which is valid in most case, but I persist thinking one getchar is enough to handle the "press any key" prompt in this particular example, as the buffer is discarded anyway after the main function is exited.

For the rest of your posting, you are just suggesting to do what I already did and explained. I see no contradiction here.