Editing C character classification (section)

== Implementation ==
For performance, the standard character classification functions are usually implemented as macros instead of [[function (programming)|functions]]. But, due to limitations of macro evaluation, they are generally not implemented today as they were in early versions of [[Linux]] like:

<syntaxhighlight lang="c">
#define isdigit(c) ((c) >= '0' && (c) <= '9')
</syntaxhighlight>

This can lead to an error when the macro parameter <code>x</code> is expanded to an expression with a [[Side effect (computer science)|side effect]]; for example: <code>isdigit(x++)</code>. If the implementation was a function, then x would be incremented only once. But for this macro definition it is incremented twice.

To eliminate this problem, a common implementation is for the macro to use table lookup. For example, the standard library provides an array of 256 integers {{endash}} one for each character value {{endash}} that each contain a bit-field for each supported classification. A macro references an integer by character value index and accesses the associated bit-field. For example, if the low bit indicates whether the character is a digit, then the <code>isdigit</code> macro could be written as:

<syntaxhighlight lang="c">
#define isdigit(c) (TABLE[c] & 1)
</syntaxhighlight>

The macro argument, <code>c</code>, is referenced only once, so is evaluated only once.