How about adding more explict writing/reading converter interface?

[hqsz]

Hello, thanks for supporting spring framework :)

It appears that Spring Data's converter system currently distinguishes between writing converters and reading converters based on annotations.

		Class<?> type = converter.getClass();
		boolean isWriting = type.isAnnotationPresent(WritingConverter.class);
		boolean isReading = type.isAnnotationPresent(ReadingConverter.class);

Is there a specific reason for controlling this solely through annotations? If there isn't a particular reason, how do you feel about adding WritingConverter and ReadingConverter interfaces that extend the Converter interface to define converters more explicitly?

[hqsz]

I already reported to spring-projects/spring-framework#31260. but I moved here because they said this place was more suitable

[mp911de]

Spring Data follows the one-way nature of Spring Converters where each converter is specified by its own type. Annotations are only required to disambiguate converters if their type signatures cannot be mapped to store-native types or when we cannot determine the conversion direction (store to model, model to store).

Having a separated model makes it easier to implement only one side of the conversion as we have a lot of cases of one-way usage.
We also provide ConverterBuilder as middleground for a more API-oriented approach to configuring converters.

Taking a step back, what is the thing that you're looking for (i.e., what problem do you want to solve)?

[hqsz]

yeah i agree about your comment.

I encountered this issue with spring & mongo. so i want to avoid ambiguity of converter interface In terms of service developer. maybe adding validation to MongoCustomConversions can be a workaround solution

Thanks for answering! i'll close this issue. if you have any other opinion. feel free to reopen this issue. Have a good day 👍