Re: [boost] Re: Interval Library and comparison operators

From: Sylvain Pion (pion_at_[hidden])
Date: 2002-09-05 14:23:53

On Thu, Sep 05, 2002 at 12:49:46PM -0600, Dave Gomboc wrote:
> > >   1) Don't support any comparison operators, or
> > >   2) Use comparison policies
> > 
> >     3) Pick a fixed meaning for the comparison operators, and provide
> > function (object)s for the other kinds of comparison.
> 
> Judging from the variety of different, conflicting definitions that other
> interval arithmetic libraries use, and the varied proposals that Boosters
> have brought forth, it seems that no particular comparison operation has a
> clearly most valid claim to operator<= and the like.  Therefore, I don't
> think that assigning one meaning to them is appropriate: no matter the
> definition selected, mnay users would find it counter-intuitive and expect
> an alternate definition.  I think it would be best if all 13 relations
> Joel Young suggested be provided (meets, precedes, etc.) and either of the
> following done:
> 
>     a) prohibit definition of relational operators such as <= entirely
>        (declare them private and do not define them)
> or
>     b) allow the library user to overload the relational operators
>        as they see fit for their application (which is easily done if
>        implementations of the 13 relations are already provided).

You missed Douglas Gregor's initial arguments in another mail.  Quoting him :

> It seems there are really only two options here:
>   1) Don't support any comparison operators, or
>   2) Use comparison policies
> 
> If we were to choose the former, users will write their own versions of 
> operators <, <=, >, >=, ==, !=. Inevitably, we will come across a project 
> that has two different meanings of operator< for intervals (but the function 
> signature will be the same), leading to ODR violations and confused 
> programmers.

The ODR rule is what kills you, because some applications might need more
than one comparison policy.

-- 
Sylvain