Question

I’m working with some autoconf stuff, and there is a test that includes the following:

static int test_array [1 - 2 * !((((float)((int)((float)1.4))) == ((float)1.4)) >= 0)];

This fails with:

error: storage size of ‘test_array’ isn’t constant

but when I change this to:

static int test_array [1 - 2 * !((((int)((int)((int)1.4))) == ((int)1.4)) >= 0)];

it works fine.

I’m not very familiar with all the hacks the autotools employ, but both of these seem like the value should be determinible at compile-time to me.

Why does the first one fail?

Answer 1

Let us look at the standard (6.6 (6)):

An integer constant expression shall have integer type and shall only have operands
that are integer constants, enumeration constants, character constants, sizeof
expressions whose results are integer constants, _Alignof expressions, and floating
constants that are the immediate operands of casts. Cast operators in an integer constant
expression shall only convert arithmetic types to integer types, except as part of an
operand to the sizeof or _Alignof operator.

(emphasis mine).

You must not cast to float in an integer constant expression (except in arguments to sizeof or _Alignof), thus

1 - 2 * !((((float)((int)((float)1.4))) == ((float)1.4)) >= 0)

isn’t an integer constant expression.

The number of elements of a static array must be an integer constant expression evaluating to a positive integer.