Re: [boost] 3-state boolean type

From: Fernando Cacciola (fcacciola_at_[hidden])
Date: 2002-08-02 10:24:42

----- Original Message -----
From: "Peter Simons" <simons_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Friday, August 02, 2002 11:49 AM
Subject: Re: [boost] 3-state boolean type

> Just curious: Shouldn't the property of a variable being unset or being
> set (and having a value in the latter case) be usable more generally?
> In a project of mine, I used the following template to achieve this:
>
>     template <typename varT>
>     class resetable_variable
>         {
>       public:
>         typedef varT value_type;
>
>         resetable_variable()
>                 : is_set(false)
>             {
>             }
>         resetable_variable(const value_type& val)
>                 : value(val), is_set(true)
>             {
>             }
>         resetable_variable<value_type>& operator= (const value_type& rhs)
>             {
>             value = rhs;
>             is_set = true;
>             return *this;
>             }
>         const value_type& data() const throw()
>             {
>             return value;
>             }
>         operator const value_type&() const throw()
>             {
>             return value;
>             }
>         bool empty() const throw()
>             {
>             return !is_set;
>             }
>
>       private:
>         value_type value;
>         bool is_set;
>         };
>
> This was used when parsing configuration files in order to know which
> values had been set in the file and which had not.
>
Yep, I've found this quite useful myself.
Look at the Files Section for 'class optional'. It is conceptually quite
similar to resetable_variable.
I had held formal submission of the optional<> class waiting for a
definitive solution to the 'stack-aligned storage' problem, because
optional<> doesn't default contructs its object when it is uninitialized,
but for this it requires a properly-aligned uninitialized raw storage as
part of itself (to avoid a heap allocation).
Now we do have aligned storage, but I haven't resumed the submission setup
(even though I have incorporated it in my own optional<> code which I use
daily), mainly since it wasn't clear when I first drop the idea here if it
raised sufficient interest.
Now that you bring this up, I think I might brush it up and make a formal
submission.

Nevertheless, a tribool type is useful in it self, beyond this.

Fernando Cacciola
Sierra s.r.l.
fcacciola_at_[hidden]
www.gosierra.com