Add test to verify that deserialization of sub type doesn't wrongly use BigDecimal

[richturner]

Search before asking

• I searched in the issues and found nothing similar.

Describe the bug

New issue created from comment in #3251 (which seems related).

Deserializing JSON with floating point numbers will incorrectly use BigDecimal instead of Double when using sub types (even though DeserializationFeature.USE_BIG_DECIMAL_FOR_FLOATS is false).

This bug is fixed in 2.15.2 but is present in the mentioned version (2.14.3) but a test should be added to prevent future regression if this fix was not deliberate.

I can also confirm the value of @JsonTypeInfo#include does not affect this behaviour.

Version Information

2.14.3

Reproduction

Subtype class definition

@JsonTypeInfo(property = "type", use = JsonTypeInfo.Id.NAME)
@JsonSubTypes({
    @JsonSubTypes.Type(ValueTest.ValueTest1.class)
})
public class ValueTest {

    @JsonTypeName("valueTest1")
    public static class ValueTest1 extends ValueTest {
        Object[] allowedValues;

        @JsonCreator
        protected ValueTest1() {
        }

        public ValueTest1(Object... allowedValues) {
            this.allowedValues = allowedValues;
        }

        public Object[] getAllowedValues() {
            return allowedValues;
        }
    }
}

JSON string to deserialize

JSON.readValue("{\"allowedValues\": [ 1.5, 2.5 ], \"type\": \"valueTest1\"}", ValueTest.class)

Result

Hydrated ValueTest instance will have BigDecimal objects in the allowedValues property.

Expected behavior

Hydrated ValueTest instance should contain Double objects unless DeserializationFeature.USE_BIG_DECIMAL_FOR_FLOATS is true

Additional context

No response

[cowtowncoder]

Added test based on suggestion; passes for 2.16.0-SNAPSHOT and should guard against regression now.