For the id primary key field, should the ID type be used as the identifier instead of using a specific String or Int? #3608

For example, I used the ID type for the id field, which corresponds to the Long type id field in my Java entity. In one of my environments, the return value id type for this request is a string, such as "1", but in another environment, it returns the number 1 without quotes.

ok I think I understand more

I dont think this is a graphql-java thing

graphql.scalar.GraphqlIDCoercing#serializeImpl has this implementation

    private String serializeImpl(Object input, @NotNull Locale locale) {
        String result = String.valueOf(input);
        if (result == null) {
            throw new CoercingSerializeException(
                    "Expected type 'ID' but was '" + typeName(input) + "'."
            );
        }
        return result;
    }

That is the ID scalars coercing code ALWAYS serialises as a String. ALWAYS by specification and ALWAYS by implementation

That said its possible your JSON serialisation code might change the value "1234" into 1234 say as it goes into the network wire. This is a major guess but I do recall someone having this problem before

That is to say, ID=String, right? As for why such a thing happened, I don't have any good ideas. The environment has been running for many years. This is the first time I have found such a problem. I don't know if it has anything to do with the server/nginx.

Can you debug it - can you log the graphql java execution result BEFORE it goes into the network layer where it gets converted to JSON say? Are you using JSON or YAML or what on the wire?

No, the problem was discovered in the production environment, and there is no way to debug it