What the...? casting bug?

Technical Programming
1

'splain this, somebody:

#define STRAIGHT_DRIVE_DIRECTION ( 127 + 30 )
 
      static int steeringValue = 0;
      . . .
      steeringValue = (int) ((STRAIGHT_DRIVE_DIRECTION));
      printf( "nav: DTC  %3d ", steeringValue  );

That printed a value of -99! Why?

Changing the assignment to:


      steeringValue = (int) ((unsigned char) (STRAIGHT_DRIVE_DIRECTION));

made it print “157” as expected the first time.

Is this weird, or is there something about type casting I don’t understand?

2

Ohhhhhhh!

Funny how you understand something as soon as you post it.

The compiler takes (127 + 30) and performs signed 8-bit arithmetic, yielding -99. Casting as a signed (int) duly kept the sign: -99. Casting as an unsigned char reinterpreted the value as 157, which was then cast as a signed int: 157.

My error was in assuming that constants were treated as unsigned chars, but they are treated as signed chars.

3

That is exactly what happened. :slight_smile:

4

Boy, you guys are fast. :slight_smile:

Norm, I think if you had used


#define STRAIGHT_DRIVE_DIRECTION ( 127u + 30 )

you would have been ok.

5

There’s a compuler option for that. Project>Build Options…>Project. Go to MPLAB C18. check “Treat ‘char’ as unsigned” (under Category General). The command-line flag is ‘-k’