They are supported, the problem is that you rely on implicit construction of an A instance which does not exist in D.

The correct way to write this is

struct A
{
	int a;
}
int f(A a...){ return a.a; }
static assert(f(A(4))==4); 

I think this one can be closed as invalid.

They are supported, the problem is that you rely on implicit construction of an A instance which does not exist in D.

Implicit construction is supported for class arguments of a typesafe variadic function.

class A
{
    this(int a) {this.a = a;}
	int a;
}
int f(A a...){ return a.a; }
static assert(f(4) == 4);

See: https://dlang.org/spec/function.html#typesafe_variadic_functions

For other types [than arrays and classes], the argument is passed by value.

So yes, the current compiler behavior is correct, but it makes a variadic struct argument completely useless. You might as well just write:

-int f(A a...){ return a.a; }
+int f(A a){ return a.a; }

Asking why arrays and classes are supported here, but not structs, is a valid enhancement request.

However, Walter has tried to deprecate typesafe variadic arguments before, since D has so many different kinds of variadic functions that it's getting really complex. I think most D users don't even know implicit construction of classes exists, I haven't seen them being used. Perhaps instead of adding typesafe variadic structs, we should deprecate typesafe variadic classes instead. (Removing typesafe variadic arrays received some pushback, since it is being used)

That is wild!

we should deprecate typesafe variadic classes instead.

Yes.

Implicit construction is supported for class arguments of a typesafe variadic function.

sorry if I mislead people with my attempt to get this issue closed but in first place, why ? usually implicit construction is not a thing.

Perhaps instead of adding typesafe variadic structs, we should deprecate typesafe variadic classes instead

yes, in a first time, then eventually later the whole "type safe variadics"-thing.

Anyway, after playing with that feature, that's just silly... as you have to define a ctor. I dont understand why this was added to the language. Is that a D1 fossil ?