array initialization, is referencing a previous element ok?

Question

const QPointF points[] =
{
    QPointF(r.left() - i, r.top() - i),
    QPointF(r.right() + i, r.top() - i),
    QPointF(r.right() + i, r.bottom() + i),
    QPointF(r         


        

Answer 1

from http://www.comeaucomputing.com/pcgi-bin/compiler.cgi:

Copyright 1988-2008 Comeau Computing.  All rights reserved.
MODE:strict errors C++ C++0x_extensions

"ComeauTest.c", line 8: warning: variable "points" is used before its value is set
      points[0] // is this line valid (according to the C++ standard)?

Answer 2

_{C++03/C++11 answer}

---

No, it's not.

On the right-hand side of the =, points does exist¹ but the initialiser is only applied after all its operands have been evaluated.

• If points is at namespace scope (and thus has static storage duration and has been zero-initialized²), then this is "safe" but your use of points[0] there is going to give you 0, rather than QPointF(r.left() - i, r.top() - i) again.

• If points has automatic storage duration — it has not yet been initialised so your use of points[0] is attempting to use an uninitialised variable, where points[0] has an indeterminate value... which is bad³.

It's difficult to provide standard references for this, other than to say that there is nothing in 8.5 "Initializers" that explicitly makes this possible, and rules elsewhere fill in the rest.

---

¹ [n3290: 3.3.2/1]: The point of declaration for a name is immediately after its complete declarator (Clause 8) and before its initializer (if any), except as noted below. [ Example:

int x = 12;
{ int x = x; }

Here the second x is initialized with its own (indeterminate) value. —end example ]

² [n3290: 3.6.2/2]: Variables with static storage duration (3.7.1) or thread storage duration (3.7.2) shall be zero-initialized (8.5) before any other initialization takes place. [..]

³ [n3290: 17.6.3.3/2]: [..] [ Note: Operations involving indeterminate values may cause undefined behavior. —end note ]

Answer 3

Since there is no sequence point in the statement, the result is undefined, much like for the i=i++ example quoted at that wikipedia page.

Otherwise said, nothing specifies, whether the compiler should first evaluate everything, then assign, or do evaluate-assign for each element separately and in which order.

Answer 4

Old Answer (misses the point):

I checked the current C++0x draft, and there i found the sentence 8.5.1.17 which says:

17 The full-expressions in an initializer-clause are evaluated in the order in which they appear.

So while this sentence is not part of the C++ Standard from 2003, im quite sure that this should be working in any up to date compiler, if this is part of C++0x.

Edit:
The comments made me rethink this matter. This line only ensures that the QPointF objects are created in the order in which they occur in the array initialization (relevant if the element constructors have observable side-effects). The problem is, the value of points is indeterminate during its array initialization. So there cant be a guarantee for a valid value of points[0] either, at least not if you rely on the standard.