Comparison of Imaginary types

[Damian McGuckin]

On Saturday, 11 July 2026 at 11:04:34 UTC, Nick Treleaven wrote:

> I forgot to reply to that. Don actually said C and D are 
> different:
>
>> D can avoid the need for a pure imaginary type, because *D 
>> guarantees constant folding, whereas C does not*.

While it is only my opinion, ...

I would agree 100% s a computer scientist and 200% if I was a 
maintainer of a compiler. But I would argue that this is 
outweighed by the clarity that an imaginary type provides to an 
end user (and people writing documentation). I would also 
apologize for my opinion to the person writing the compiler 
because it means more work. D certainly has lots of advantages 
over C. That said, in one of Walter's recent presentations, he 
notes that an attribute of Elegant Code is orthogonality. And 
having real types, imaginary types and complex types is an 
example of that. As an important aside, I will note that the line 
containing the link to that presentation on Walter's home page, 
the link to the Elegant D symposium has a typo, i.e. instead of 
dlangsymposium.com, it says dlangsymposiumlcom. Oops. Someone's, 
maybe Walter's, fat fingers typed the 'elle' key rather than the 
'dot' key which is just below it on the keyboard. I regularly 
type m instead of a comma for the same reason. Sorry, I digress.

I could not understand why Don thought the C99 imaginary type is 
a hack. I know one of the authors of the document supporting its 
inclusion very well and I have rarely had cause to question his 
insight. And the other author is a legend and a winner of the 
Turing award (not that such an award makes him infallible, just a 
lot wiser than I).

That use of UFCS to allow one to write "4 + 2.i" highlights the 
power of D. Nice one.

Thanks heaps for the feedback and spending time replying to my 
questions. Coming back to D after 15+ years is a steep relearning 
curve.