c++ bool anomaly -- why is this possible?

Question

I just stumbled across this: When postifix incrementing a bool twice, the integer value still is 1 instead of the expected 2. I'd like to know why this happens.

bool test = false; // results in integer 0
test++; // results in integer 1
test++; // results in integer 1

// but there still is at least one byte of memory used:
// results in integer 137
*(unsigned char*)(&test) = 137;

Without looking at any machine code produced by this, I would assume that the compiler ignores all but the least significant *bit* when you've told it to work on a `bool`. It's also possible that `bool + 1` gets turned into `bool & 1` as an optimization or to prevent overflows, which would cause the behavior you're seeing. The compiler could be doing some even more clever; you should really include the assembly for this bit of code. — ssube, Jun 24 '13 at 16:52
A `bool` can only be either `true` or `false`, setting it to 137 is undefined behavior — David Rodríguez - dribeas, Jun 24 '13 at 17:10

score 8 · Accepted Answer · edited Jun 20 '20 at 09:12

8

This is how the ++ operator is specified. See C++11 §5.2.6[expr.post.incr]/1 (emphasis mine):

The value of a postfix ++ expression is the value of its operand. [Note: the value obtained is a copy of the original value —end note] The operand shall be a modifiable lvalue. The type of the operand shall be an arithmetic type or a pointer to a complete object type.

The value of the operand object is modified by adding 1 to it, unless the object is of type bool, in which case it is set to true. [Note: this use is deprecated, see Annex D. —end note]

(The prefix ++ operator has similar language to allow it to be applied to a bool value.)

Modifying the bool object through a pointer or reference of a type other than bool yields undefined behavior.

edited Jun 20 '20 at 09:12

Community

1
1

answered Jun 24 '13 at 16:52

James McNellis

348,265
75
913
977

I'm guessing applying `--` to a `bool` would set it to `false`, then, regardless of its prior state? – JAB Jun 24 '13 at 16:53
@JAB: No. Neither the prefix nor the postfix `--` may be applied to an operand of type `bool`. – James McNellis Jun 24 '13 at 16:54
1

Aliasing through `unsigned char` is allowed, though. The UB doesn't come from just modifying it, it comes from writing a value not in the value representation of `bool`. This is fine: `bool d, s = true; std::memcpy(&d, &s, sizeof(bool)); /* loop char by char */` for example. – GManNickG Jun 24 '13 at 16:59

score 0 · Answer 2 · answered Jun 24 '13 at 16:50

0

why is this possible?!

Undefined behaviour is just that - undefined. Anything at all could happen.

if enough memory for 137 is allocated, then why the hell test++; results in 1?

The language specification defines that behaviour.

answered Jun 24 '13 at 16:50

Carl Norum

219,201
40
422
469

score 0 · Answer 3 · answered Jun 24 '13 at 16:51

0

Because you've invoked undefined behavior by casting a pointer to a different type and performing read/write on it. UB = anything can happen, including nasal demons.

answered Jun 24 '13 at 16:51

Edward Strange

40,307
7
73
125

c++ bool anomaly -- why is this possible?

3 Answers3