yank unary '+'?

[Andrei Alexandrescu]

Michiel Helvensteijn wrote:
> Andrei Alexandrescu Wrote:
> 
>>> What will removing it gain you?
>> Sancta simplicitas.
> 
> Hm.. I don't really buy that argument.
> 
> I see you and Walter removing/witholding things (incomparability
> operators, logical operator overloading) from the language, because:
> "I can't imagine a use for it and removing it makes the language
> simpler."

I disagree, and I think "not imagining a use for X" is a very weak 
argument for removing X. When a feature is made to walk the plank, we 
have much better arguments than that. Take typedef: it was ill-defined, 
defining it properly would have been a major effort, and all benefits 
could actually be emulated with a struct simpler and cheaper.

> Meanwhile, you're keeping C syntax for function-pointers around, and
> I'm missing syntactic sugar for my tribool.

C syntax for pointers to functions is there for a reason, but we're 
considering removing it. For one thing, TDPL doesn't mention it.

> The fact that you or I think there isn't a use for a feature, doesn't
> mean there isn't one. Programmers keep finding new and unintended
> uses for language features, which is a good thing. And if you want to
> simplify the language, I wouldn't start with the unary + when you've
> still got all that C stuff around.

I was mostly talking about overloading operator +. Operator + has a long 
history of being available for overloading in C++, which we can use to 
our advantage. It has been used to emulate DSLs, but I think we have 
much better means to define DSLs in D than to redefine all operators to 
mean some convention-chosen things.


Andrei