[RFC] Rich enum constants without parens #136
Conversation
| } // namespace fixed_containers::rich_enums_detail | ||
|
|
||
| template <typename RichEnumType> | ||
| class RichEnumConstantProxy |
There was a problem hiding this comment.
I generally approve of this idea. I am leaving my thoughts as a comment rather than as PR review such that replies are simpler.
Specifically, I like that this change makes the common use cases of RichEnums simpler. For example, I like that we can omit operator() when comparing enums such as:
TestRichEnum1::value_of(BE::C_ONE) == TestRichEnum1::C_ONEor in switch statements such as:
case TestRichEnum1::C_ONE:However, the current approach has some downsides in my eyes.
- The first downside in my opinion is that calling member functions of
RichEnumsnow requiresoperator->, which in my opinion is an operator which conveys pointer like semantics which imply nullptr possibility. However, since the pointer will never be null in this case, this is a purely aesthetic concern and unfortunately I don't currently have alternate suggestions. Example:
TestRichEnum1::C_FOUR->ordinal()- I am not in favor of the
// TRANSITION
constexpr const RichEnumType& operator()()In my opinion, this will result in having two different ways of doing the same thing and I expect both to stick around. From what I've seen, things in transition:: tend to stick around ;) . I suggest that we make this a breaking change that requires users to change their code. That is unless this repository has some convention of backwards compatibility.
| BackingEnum backing_enum_{}; | ||
|
|
||
| public: | ||
| explicit(false) constexpr RichEnumConstantProxy(const BackingEnum& backing_enum) |
There was a problem hiding this comment.
question from curiosity: I've noticed the pattern explicit(false) for a few ctors. What is the reason for it? Omitting it would have no effect since ctors are implicit by default, right?
There was a problem hiding this comment.
explicit(false) makes it explicit that the function is intended to allow implicit conversion and it is not an omission.
2 participants
No description provided.