Skip to content

CWG2873 [over.over] Is it intended that when a function is overloaded with a function template then its address cannot be taken unambiguously #516

Description

@ranaanoop

Full name of submitter: Anoop Rana

Reference (section label): [over.over]

Link to reflector thread (if any): https://stackoverflow.com/questions/78170991/in-c-why-cant-i-take-the-address-of-a-function-that-also-has-a-templated-vers

Issue description:

I came across the following code that is rejected by all three compilers and seems to be ill-formed as per current wording in over.over#3.

template <typename T>
void AcceptAnything(T&&);

void Foo();

template <typename T>
void Foo();

void Bar() {
  AcceptAnything(&Foo);
}

I think this is ill-formed because(as explained here) because the specialization from the function template cannot be formed here so nothing can be added to the set and so it is ill-formed(I may be wrong here but as far as I understand the topic this is what makes it ill-formed). But I don't think the whole program should be ill-formed here as there is another non-template function available that can be used. Instead this inability to form a specialization from the function template can be ignored as a non-template function that can be used is available.

Suggested resolution:

Maybe something like the following can be added:

The program is ill-formed if this is part of the final result of overload resolution.

Activity

  1. jensmaurer commented on Mar 16, 2024

    @jensmaurer
    Member

    Hm... That's odd. We do say "if any"; the inability to produce a template specialization shouldn't cause a hard error for the entire set, I think.

  2. ranaanoop commented on Mar 16, 2024

    @ranaanoop
    Author

    @jensmaurer Yes, at first I also though that the "if any" should make the program well-formed. But all compilers rejecting the program seems to suggest that it should be made more clear perhaps. Though it is possible that as per current wording("if any") all compilers are wrong but I wanted to confirm what is intended here. I will update the answer afterwards.

  3. ranaanoop commented on Mar 16, 2024

    @ranaanoop
    Author

    @jensmaurer over.over#5.sentence-2 seems to make this well-formed. Although there will be no specialization that can be generated from the function template here so nothing will be eliminated from the set. So the one candidate is the non-template function which should be used. Even if somehow a specialization could've been generated, the non-template function should've been used as per over.over#5.

  4. jensmaurer commented on Mar 17, 2024

    @jensmaurer
    Member
  5. changed the title [-][over.over] Is it intended that when a function is overloaded with a function template then its address cannot be taken unambiguously[/-] [+]CWG2873 [over.over] Is it intended that when a function is overloaded with a function template then its address cannot be taken unambiguously[/+] on Mar 17, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions