[boost] Re: a new version of geometry classes

From: Dave Abrahams (abrahams_at_[hidden])
Date: 2000-01-07 10:06:20

>> > Oh, here's another feature I like for my points: indexability (I like
>> > to be able to access the coordinates by index). That allows
>> > programmatic selection of orientation for things like scrollbars.
>> > Indexability is of course unusable when the types of x and y differ.
>> > OTOH this is definitely a corner case. I have no strong opinion on the
>> > topic.
>>
>> We can provide generic version of classes with two template parameters
>> (and without indexability) and a specialization for case T1==T2 with
>> operator[].  I think it will be reasonable solution.
>
> I have some sympathy with the overdesign comment, but it may be worthwhile
> considering the following for the generic version.  It overloads the []
operator
> and uses function overloading to achieve programmatic selection on orientation
> etc.  Obviously, as it's a compile time thing, it can't be used for looping
over
> (x .. y) - but that shouldn't be a problem for just two coordinates.
>
>     struct x_coord {} x;
>     struct y_coord {} y;
>
>     template <typename X, typename Y = X> class point
>     {
>     public:
>
>        X operator[]( x_coord ) { return x_; }
>        Y operator[]( y_coord ) { return y_; }
>
>     private:
>
>        X x_;
>        Y y_;
>     };
>
>     class latitude  {};
>     class longitude {};
>
>     void process_coord( latitude )  {}
>     void process_coord( longitude ) {}
>
>
>     void test()
>     {
>        point<latitude, longitude> p;
>
>        latitude  lat = p[x];
>        longitude lon = p[y];
>
>        process_coord( p[x] );
>        process_coord( p[y] );
>     }
>
> Gavin
>
> --
> Gavin Collings
> gcollings_at_[hidden]

Yep, this is a fine design. The question remains, though: should we restrain
ourselves, and say "if you need this specialized application, you should
make your own wrapper?" After all, this kind of orientation flipping is just
a simple case of generalized transformation using matrix multiplication.
Since our geometry proposal is strictly 2D and doesn't include any other
facilities for transformation so far. The more I think about it, the more
I'm sorry I mentioned it.

-Dave