Implement enum to string #89
Conversation
|
Why is Windows build failing? P.S. |
|
I think the basic implementation is OK for a PoC quality; I'll be mentioning some functional issues in review. |
saki7
added
the
upstream-workaround
|
|
||
| template <class E> | ||
| requires std::is_enum_v<E> | ||
| constexpr std::string_view enum_to_string(E value) |
There was a problem hiding this comment.
- The basic name for this functionality should be
to_identifier, notto_string. We neednoexceptand[[nodiscard]]; we can treat this non-throwing version as the default (preferred) function. Since this version has the fallback value, it should be namedto_identifier_or. - In addition to the current version, we need another version that throws
class bad_enum_access : std::exceptionfor nonexistent values . Note that, even if the predefined enum members can be statically determined by reflection, there are still dynamic values for flag-like enums. This throwing version should be namedto_identifier. - Additionally, I think we can provide yet another version that delegates to
std::to_charsin case of nonexistent value. This version can be named as e.g.to_identifier_or_underlying.
| return std::meta::identifier_of(enumerator); | ||
| } | ||
| } | ||
| return "<unnamed>"; |
There was a problem hiding this comment.
This should be customizable via function parameter; something similar to std::optional::value_or is needed
| } // namespace yk | ||
|
|
||
| #endif | ||
|
|
There was a problem hiding this comment.
We need some utility for adapting user-defined enums.
Such functionality should enable std::formatter and operator<<, and it should require minimal effort on the user's part.
There was a problem hiding this comment.
I'm considering yk::enum_serializer like std::formatter, defaults to use yk::enum_to_identifier. What do you think?
|
I think we can also provide a more rich version that resolves combination of flag-like values (e.g. |
|
Since we have this serialization feature now, I think we should also have the opposite one, i.e. enum parser. We should determine the priority for the parser implementation; I'm not sure if it should be included in the initial PR. |
| requires std::is_enum_v<E> | ||
| constexpr std::string_view enum_to_string(E value) | ||
| { | ||
| if constexpr (std::meta::is_enumerable_type(^^E)) |
There was a problem hiding this comment.
Is this condition really needed? Do there exist such case that std::is_enum_v<E> == true && std::meta::is_enumerable_type(^^E) == false ?
There was a problem hiding this comment.
is_enumerable_type returns false if E is incomplete type, which results to prevent enumerators_of being constant expression.
There was a problem hiding this comment.
Understood. I think we need a test case for this.
closes #88
https://godbolt.org/z/Erq7bfevf