DIP 1019--Named Arguments Lite--Final Review

[drug]

27.08.2019 16:47, aliak пишет:
> On Monday, 26 August 2019 at 22:45:16 UTC, Walter Bright wrote:
> 
> So just to test out what people would say, I went around my team with 
> this (individually, they didn't hear each other)
> 
> struct S { int a = 1, b = 2, c = 3, d = 4; }
> 
> void main() {
>      S w = { d: 9, a: 9, 9 };
>      writeln(w);
> }
> 
> The answers:
> 
> 1. a 9, d: 9,... b...2? wait no .... what? Ok I don't know.
> 2. ambiguous? compile error! why would this be allowed?
> 3. does the last 9 set all the unset vars??
> 4. Can't tell, but the first highest non set address?
> 5. compilation error
> 6. What is this. What is the usefulness of this. Nothing happens. It's 
> bad. It should crash. I hate life.
> 7. this can't compile. If it does I would assume it's thrown away?
> 8. should crash?
> 
> Please don't allow the same logic to go in to named arguments.

That is rather simple logic - named arguments are defined by name, 
unnamed ones are defined by position starting from the last named 
argument, i.e. the first unnamed argument follows the last named one.