[stmt.return,class.{ctor,dtor}] Clarify no return operand #4737
Conversation
| \indextext{conversion!return type}% | ||
| the \tcode{return} statement initializes the | ||
| glvalue result or prvalue result object of the (explicit or implicit) function call | ||
| by copy-initialization\iref{dcl.init} from the operand. | ||
| \begin{note} | ||
| A constructor or destructor does not have a return type. |
There was a problem hiding this comment.
If we claim that a constructor of a class does not have a return type, consider [dcl.init.general] p15
if the function is a constructor, the call is a prvalue of the cv-unqualified version of the destination type whose result object is initialized by the constructor.
[expr.call] p14
A function call is an lvalue if the result type is an lvalue reference type or an rvalue reference to function type, an xvalue if the result type is an rvalue reference to object type, and a prvalue otherwise.
These rules will be in conflict. How could the call of a constructor be a prvalue of the cv-unqualified version of the destination type?
Similarly, if a destructor does not have a return type, how does the following rule work?
If the postfix-expression names a destructor or pseudo-destructor ([expr.prim.id.dtor]), the type of the function call expression is void;
There was a problem hiding this comment.
I'm not seeing the quoted text in [dcl.init.general] p15.
A constructor invocation is not a syntactic function call as described in [expr.call] (it's an explicit type conversion instead), thus any statements about the type of the function call expression don't apply.
We can explicitly call a destructor in a function call expression, and the rule you're quoting overrides the general rule that the type of a function call expression is the return type of the called function.
I'm not seeing any conflicts.
There was a problem hiding this comment.
I'm not seeing the quoted text in [dcl.init.general] p15.
It is moved to dcl.init#general-16.6.3 in the current draft
if the function is a constructor, the call is a prvalue of the cv-unqualified version of the destination type whose result object is initialized by the constructor. The call is used to direct-initialize, according to the rules above, the object that is the destination of the copy-initialization.
There was a problem hiding this comment.
@jensmaurer Final thoughts?
There was a problem hiding this comment.
Constructors / destructors don't have a return type in the "function declaration" sense, and that absence makes "return something" syntactically ill-formed. I think the quoted stuff doesn't change that; it just means that a constructor call (in the context of copy-initialization, not in general) has (is considered to have) a certain type and value category. That has no bearing on the syntactic constraints on constructor declarations.
I thinks this should go in.
source/classes.tex
Outdated
| @@ -2078,8 +2079,13 @@ | ||
| the selected destructor may be deleted\iref{dcl.fct.def.delete}. | ||
|
|
||
| \pnum | ||
| \indextext{restriction!destructor}% | ||
| \indextext{constructor!address of}% |
There was a problem hiding this comment.
Thanks -- could you rebase please? Something weird seems to have happened to this branch.
Highlight that constructors and destructors do not have a return type and thus a return statement within a constructor or destructor cannot have an operand.
|
Rebased and force-pushed. |
Highlight that constructors and destructors do not have a
return type and thus a return statement within a constructor
or destructor cannot have an operand.
Supersedes #3647