To `toupper()` and `tolower()`: Must We Cast to `unsigned char`?

When Calling toupper() and tolower(): Casting to unsigned char

While a highly reputable commenter on Stack Overflow suggests casting char-arguments to unsigned char before calling toupper or tolower, this necessity is not explicitly mentioned by Bjarne Stroustrup in "The C Programming Language." This difference in opinion sparks the question: is such a cast necessary or is it an oversight?

Types and Representations

char, signed char, and unsigned char are distinct types. While char commonly represents either a signed or unsigned representation, the toupper function requires an int argument representable as an unsigned char. If the argument is not representable or equal to EOF, undefined behavior occurs.

Undefined Behavior

Suppose plain char is of a signed type. If name[0] denotes a negative value, using toupper(name[0]) risks undefined behavior. However, for the example provided by Stroustrup, the initialization guarantees non-negative values.

Unnecessary Conversions

Nevertheless, converting char to (unsigned)char does not resolve the issue since it may still result in a negative int value due to the implicit conversion.

Practical Considerations

While toupper could be implemented to handle negative values, it's not mandatory. Additionally, these functions must accept arguments equal to EOF (-1), which is typically a negative value.

Conclusion

Despite its acceptance of EOF, toupper requires the input character to be representable as an unsigned char to prevent undefined behavior. While Stroustrup's example may not demonstrate a need for casting, it's a recommended practice to ensure portability and correct functionality, particularly when dealing with negative or special characters.

The above is the detailed content of To `toupper()` and `tolower()`: Must We Cast to `unsigned char`?. For more information, please follow other related articles on the PHP Chinese website!