问题
When an aggregate contains only one element, as below, positional notation results in a compilation error and we have to use named notation only. Why?
type singleton is record
v : integer;
end record;
v1 : singleton := (0);
results in the compiler message
check.adb:6:23: positional aggregate cannot have one component
check.adb:6:23: write instead "V => ..."
gnatmake: “check.adb" compilation error
whereas this is OK:
v2 : singleton := (v => 0);
回答1:
Parentheses round an expression are redundant, so (0) = 0 and it's an integer not an array aggregate.
So, for the special case of a one-element aggregate, named association is required to distinguish an aggregate from a simple value.
Contrast this with (0,0) which can only be an aggregate; therefore there is no ambiguity.
Even though in the context of the question, it's obvious to a human programmer which is intended, that will not always be the case.
Consider a one-element aggregate in a multi-dimensional array which is one field of a record; there can be ambiguities that the compiler cannot resolve (at least, before reading a whole lot more of the source file!) and would make life pretty difficult for anyone reading the program.
回答2:
You don't have to use named notation.
v1 : singleton := (others => 0);
This will assign 0 to all elements in v1 and the compiler will know that is not a number but an array instead.
If the record happen to have different types you could use
v1 : singleton := (others => <>);
来源：https://stackoverflow.com/questions/26955655/why-is-named-association-required-for-a-one-element-record-aggregate