[std-interval] passing by value vs reference

[Sylvain Pion]

Guillaume Melquiond wrote:
> Le mercredi 05 avril 2006 à 11:32 +0200, Sylvain Pion a écrit :
> 
> 
>>3- leave freedom for the implementation to choose
>>4- leave freedom for the implementation and have it provide
>>    a standard specified way to obtain this information.
> 
> 
>>I would prefer 3 to 4.  I think
>>the complexity of implementing 4 is not worth it.
> 
> 
> I don't see any additional complexity in 4 with respect to 3. As already
> mentioned by Bill Clarke, you could simply have a typedef member in
> interval<T> to express the calling convention of the functions of the
> interval library: typedef interval<T> const &param_type;
> 
> This requires all the functions of the library to use the same calling
> convention. But I don't think it is too strong a requirement since there
> is no function that uses more than two interval arguments.

I meant complexity for the proposal.  I think this issue has the
potential to raise questions I would prefer to avoid discussing
if possible, like:
- should this be a mecanism applicable to the whole std library
   types instead?
- should it be fine grained for each argument of each function?
- is such a type-based mecanism enough for all users needs
   (or is a macro necessary for example)?
   For example: how would a user defined template function on
   intervals be defined to play nicely with this implementation
   "preference":
     template < typename T >
     void f(std::interval<T>::param_type i)
   does not work.

So, let's first hear if there is a need for such a mechanism at all,
or if the "unspecified" approach, 3, is good enough.

-- 
Sylvain