Feature add variant_cast #129
Conversation
Added 4 overloads of the method "variant_cast" In order to move values out, to return values by reference or to return a pointer to the underlying values, when the type does match: template<class T> T variant_cast(variant&& operand); template<class T> T variant_cast(variant& operand); template<class T> T variant_cast(const variant& operand); template<class T> const T* variant_cast(const variant* operand) RTTR_NOEXCEPT; template<class T> T* variant_cast(variant* operand) RTTR_NOEXCEPT; Additionally: - added non-const version of variant::get_value() - fixed copy-right year in root CMakeLists.txt file Fixes rttrorg#108
and fixed moveable test
|
@gabyx |
| { | ||
| variant var = 12; | ||
|
|
||
| auto value = variant_cast<int>(&var); |
There was a problem hiding this comment.
Is that correct: value is still bound to the var. meaning if var gets destroyed, value is dangling as well.
There was a problem hiding this comment.
yes, that is correct. Same as with variant::get_value<T>()
src/rttr/variant.h
Outdated
| * \code{.cpp} | ||
| * | ||
| * variant var = std::string("hello world"); | ||
| * std:string* a = variant_cast<std::string>(&var); // performs an internal type check and returns extracts the value by reference |
src/rttr/variant.h
Outdated
| * std:string* a = variant_cast<std::string>(&var); // performs an internal type check and returns extracts the value by reference | ||
| * int* b = variant_cast<int>(&var); | ||
| * std::cout << "a valid: " << a != nullptr << std::endl; | ||
| * std::cout << "b valid: " << b != nullptr << std::endl; |
There was a problem hiding this comment.
Probably add "// b is nullptr"
| { | ||
| const variant var = 12; | ||
|
|
||
| auto value = variant_cast<int>(&var); |
There was a problem hiding this comment.
so the const semantic in variant_cast<...> is neglected. meaning we dont write variant_cast<const int>.
Wouldnt that be better? I see the benefits of the current implementation. I would leave it like that
There was a problem hiding this comment.
@gabyx
What would be better?
You cannot have a const-by-value inside a variant. However, you can have a const int*
|
|
||
| REQUIRE(var.get_type() == type::get<int>()); | ||
|
|
||
| auto value = variant_cast<int>(var); |
There was a problem hiding this comment.
;) -> auto has the semantic of always stripping references == &. So I think when you write auto value this is always an lvalue (not an lvalue reference).
So it seems that this test is obsolete :-), but better leave it included :-)
There was a problem hiding this comment.
I know, but also this case can happen, thats why I added the test.
Its also in the section: SECTION("by value")
| * \return A reference to the stored value. | ||
| */ | ||
| template<typename T> | ||
| T& get_value(); |
There was a problem hiding this comment.
I like that one. So such a functionality was not yet there. Its up to the user to call that with the right type.
| @@ -652,7 +652,7 @@ struct RTTR_API variant_data_policy_empty | |||
| } | |||
| case variant_policy_operation::GET_VALUE: | |||
| { | |||
| arg.get_value<void*>() = nullptr; | |||
| arg.get_value<const void*>() = nullptr; | |||
There was a problem hiding this comment.
was that wrong before? and did you already had a get_value function before but not in the interface in variant (I suppose arg is a variant?)
There was a problem hiding this comment.
I go "down" with a const void*, that's why I adjusted it.
https://github.com/rttrorg/rttr/pull/129/files#diff-196fbfbb881e65b49f3262cd9ef3bf19R421
Added global "variant_cast" method
Added 4 overloads of the method "variant_cast"
In order to move values out, to return values by reference
or to return a pointer to the underlying values, when the type does match:
template
T variant_cast(variant&& operand);
template
T variant_cast(variant& operand);
template
T variant_cast(const variant& operand);
template
const T* variant_cast(const variant* operand) RTTR_NOEXCEPT;
template
T* variant_cast(variant* operand) RTTR_NOEXCEPT;
Additionally:
Fixes #108