Warning C4800: 'int' : forcing value to bool 'true' or 'false' (performance warning)

Question

I have this problem in my code:

bool CBase::isNumber()
{
return (id & MID_NUMBER);
}

bool CBase::isVar()
{
return (id & MID_VARIABLE);
}

bool CBase::isSymbol()
{
return (id & MID_SYMBOL);
}

Counterforce

Did you try return (id & MID_NUMBER) > 0;? – Aretha

id or MID_SYMBOL must be of type BOOL. which is just #define to int type in Windows. Your return value is BOOL and it is converted into bool(true or false ). – Williamson

Why not !=0 instead of >0. And no, bool it not BOOL – Handiwork

(id & MID_NUMBER) != 0; is the general way to avoid the warning. – Jermyn

One of the warnings I usually turn off: #pragma warning(disable: 4800) – Pretext

I took your title and put it into Google, several of the first results explain the warning. – Isaacson

possible duplicate of What is the performance implication of converting to bool in C++? – Jermyn

Why not accept Marco's answer? – Ionosphere

Answer 1

FYI: Casts won't hide the warning by design.

Something like

return (id & MID_NUMBER) != 0;

should clearly state "I want to check whether this value is zero or not" and let the compiler be happy

Venusberg

Answer 2

Use the !! idiom eg

bool CBase::isNumber()
{
    return !!(id & MID_NUMBER);
}

Seabrook

Just to please the compiler !?? – Pretext

And why 2 ops instead of 1 (even if it is optimized)? – Handiwork

@Dieter Lücking: Non-standard workarounds are probably worse. – Handiwork

@Handiwork The warning itself is useless – Pretext

@DieterLucking: It can not be useless as it is warning. – Williamson

@Lücking: Why? Of course the compiler can generate the correct stuff without changed code, but maybe the coder has not intended a implicit cast and finds a semantic error because of the warning? – Handiwork

First inverts it so you get the inverse of what you actually wanted, second inverts that so you get what you actually wanted. Useful when you have a mix of BOOL (windows), Bool (xwindows) and bool (C++) – Seabrook

@Handiwork : The warning is not about semantics, it is about performance, which is specious. – Rarefied

I give you a +1. This was useful for me, since I have this warning in templated code where I cannot compare against 0 but !! works. – Mantellone

Answer 3

Where's the declaration of id and MID_NUMBER? Are you sure they are not windef-style BOOLs rather than (lowercase) bool's? BOOL's have been in windef for decades typedef'd as an int; they pre-date proper C++ bool's and a lot of developers still use them.

Romina