[boost] Re: typeof

From: Ian McCulloch (ianmcc_at_[hidden])
Date: 2004-05-08 05:56:01

David Abrahams wrote:

> Ian McCulloch <ianmcc_at_[hidden]> writes:
> 
>>> Probably ;-) It's all there right under "One definition rule" in the
>>> standard.
>>> 
>>> 3 Basic concepts 3.2 One definition rule
>>> 
>>> -5- There can be more than one definition of a class type (clause
>>>  class), enumeration type (dcl.enum), inline function with external
>>>  linkage (dcl.fct.spec), class template (clause temp), non-static
>>>  function template (temp.fct), static data member of a class template
>>>  (temp.static), member function [core 249: template of a class
>>>  template ] (temp.mem.func), or template specialization for which some
>>>  template parameters are not specified (temp.spec, temp.class.spec) in
>>>  a program provided that each definition appears in a different
>>>  translation unit, and provided the definitions satisfy the following
>>>  requirements. Given such an entity named D defined in more than one
>>>  translation unit, then
>>> 
>>> 
>>>    - each definition of D shall consist of the same sequence of
>>>      tokens; and
>>> 
>>>    - in each definition of D, corresponding names, looked up according
>>>      to basic.lookup, shall refer to an entity defined within the
>>>      definition of D, or shall refer to the same entity, after
>>>                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>
>>
>> To my reading of the standard, different typedef-names that are synonyms
>> for the same type are the same entity with respect to the ODR, so I don't
>> think any of this applies here.
> 
> If you limit the question to typedefs, you're right, but I'm not
> limiting the question to typedefs.  The problem I'm discussing is this
> one:
> 
>     namespace
>     {
>        template <class T>
>        typeof (T() + T()) add(T*, T*)
>        {
>           typeof(T() + T()) x;
>           ...
>           return x;
>        }
>     }
> 
> 
> If the declaration of x (or add, for that matter, I'm pretty sure)
> uses any names that refer to different entities in different TUs, it's
> an ODR violation.

Interesting.  The purpose of typeof (or BOOST_TYPEOF) is to name a type, and
that type will be the same in all TU's.  But the declaration itself might
involve a different type in different translation units, although the full
declaration will end up naming the same entity.  An interesting interaction
with typedefs and templates.  Or would you also claim that this would
happen even in the absense of templates?  If so, what makes the template
case different?

// header.h

namespace {
struct typeof_foo
{
   typedef int type;
};
} // namespace

typeof_foo::type bar();// ODR violation here???  
                       // Why is this different to previous examples?
                       // does anything change when templates are involved?

// end header.h

Cheers,
Ian