First Draft: Nominal Sum Types via `enum union` and `switch` Expressions

[Meta]

On Thursday, 24 September 2026 at 12:40:30 UTC, Dejan Lekic wrote:
>> And it's now implemented:
>> ```d
>> enum union Shape
>> {
>>     case Circle(double);
>>     case Rect(double, double);
>>     case Point();
>> }
>>
>> double area(Shape s)
>> {
>>     return switch (s)
>>     {
>>         static foreach (T; __traits(allVariants, Shape))
>>         {
>>             static if (__traits(identifier, T) == "Circle")
>>             {
>>                 case Circle(r) => 3.14159 * r * r,
>>             }
>>             else static if (__traits(identifier, T) == "Rect")
>>             {
>>                 case Rect(w, h) => w * h,
>>             }
>>             else
>>             {
>>                 // Optional trailing delimiter before right 
>> curly
>>                 case Point() => 0.0
>>             }
>>         }
>>     };
>> }
>>
>> Shape sh = Shape.Rect(3.0, 4.0);
>> assert(area(sh) == 12.0);
>> ```
>
> Is it me or this is just terrible, and this is just a tiny 
> example. For anything more serious (read - real world use 
> cases) the amount of ugly code like the one above would be 
> enormous...

This is just an example to show that static if/static foreach 
work. While it *is* possible to write code like this with my DIP, 
you will almost never see it in practice, outside heavy 
metaprogramming scenarios like serialization libraries, writing 
compilers/interpreters, bare metal, etc.