Logical const

[Andrei Alexandrescu]

On 11/28/10 1:09 PM, Peter Alexander wrote:
> On 28/11/10 4:37 PM, Andrei Alexandrescu wrote:
>> The library solution to logical constness would gravitate around a union
>> with a const (or immutable) and a regular field of the same type:
>>
>> union {
>> T before;
>> const T after;
>> }
>>
>>
>> Andrei
>
> I must be missing something. Surely, when inside a const method, you
> still can't modify a union, whether you access a const member of the
> union or not?
>
> Could you elaborate?

Sorry, that was wrong. I was thinking of a lazily initialized field 
within an otherwise non-const object.

Andrei