A follow on to the example in ARM 9.9, p. 190:

The example states that in:

typedef int c;
class X {
 int f() { return sizeof(c); }
 char c;
};

that the "char c" declaration is in error because c was used in the inline
of f().

However, the inline of f() should not be "interpreted" until the
whole of class X has been seen, so semantically, c is not "used"
until the end of class X and thus "char c" is fine.  It is true that
lexically speaking, f() seems to "use" c before the "char c;".
However, the important thing is how the "sizeof(c)" would be
interpreted, and clearly at the end of class X c refers to "char c".
As a result, it seems clear to me that "c" should not be considered
used if it appears in an inline fuction def.

This confusion is just a symptom of the underlying disease caused by
this cankerous rule :-}.


********************************************************
* Steven Kearns            ....uunet!softrue!kearns    *
* Software Truth           softrue!kearns@uunet.uu.net *
********************************************************