feat: accept function type argument #923
Conversation
|
From #628:
Can you elaborate with examples where this is used? I'm not seeing the use cases in the PR code shown so far, unless the answer to my first question about |
| _ = typeid(int); | ||
| _ = offsetof(t, a); | ||
| _ = :<T> (_: T) _ = sizeof(T::value);(std::true_type()); | ||
| _ = :<T> (_: T) _ = sizeof(type T::value_type);(std::true_type()); |
There was a problem hiding this comment.
Is there intended to be a difference between lines 7 and 8? Or is the type intended to be optional?
If there's no difference, can you show a case where it's required?
There was a problem hiding this comment.
The second one needs the disambiguating type.
See https://cpp1.godbolt.org/z/PY53n6Kvf (generated from https://cpp2.godbolt.org/z/KK15vf6Y5):
- Clang:
error: dependent-name 'T::value_type' is parsed as a non-type, but instantiation yields a type. - GCC:
error: missing 'typename' prior to dependent type name 'integral_constant<bool, true>::value_type'. - MSVC accepts.
There was a problem hiding this comment.
D'oh, now I see it. I missed the _type suffix that changed the meaning, sorry.
| @@ -5,7 +5,7 @@ testing_enabled: bool = false; | |||
|
|
|||
| outer: @print type = { | |||
|
|
|||
| object_alias: <T> T requires true == 42; | |||
| object_alias: <T> T requires true == sizeof(type T::value_type); | |||
There was a problem hiding this comment.
Is this just to exercise type for this pretty-print self-test, or is type now intended to be required here?
There was a problem hiding this comment.
The former.
This change is intended to exercise the type keyword in the pretty-print.
| @@ -5800,9 +5845,6 @@ class parser | |||
| //G postfix-expression | |||
| //G prefix-operator prefix-expression | |||
| //GTODO await-expression | |||
| //GTODO 'sizeof' '(' type-id ')' | |||
| //GTODO 'sizeof' '...' ( identifier ')' | |||
| //GTODO 'alignof' '(' type-id ')' | |||
There was a problem hiding this comment.
Right, these should go anyway as they already work without a special grammar. And throws-expression too, since throw(...) works (and I'm using that in reflect.h2's report_violation function).
This comment was marked as resolved.
This comment was marked as resolved.
Resolves #628 (comment).
Testing summary:
Acknowledgements: