Description
A simple class with a c-tor like this:
class Result
{
public Result(string error)
{
}
}
and the json:
var json ="""
{
"error": "some error"
}
""";
var result = JsonSerializer.Deserialize<Result>(json, new JsonSerializerOptions(JsonSerializerDefaults.Web));
throws an exception at runtime saying I don't have a property or field defined in the class.
So if I add a property like this:
class Result
{
public string Error { get; }
public Result(string error)
{
}
}
the exception is not thrown anymore. I correctly receive the "some error" string in the c-tor. However, I don't care to set the property at all.
I know that if I don't want to have a property or field defined in the class, I need to implement a custom JsonConverter<Result>, so how to solve this is not my question.
My question is: Since the deserialization logic is perfectly capable of mapping the error JSON field to the error parameter of the c-tor, why is the property required?
You might ask, why would I NOT want to have a property or a field defined?
Because, in a scenario where the response is a bit more dynamic, I'd like to be able to map the c-tor parameter to something else, without having to have something more complicated like a custom JsonConverter:
class Result
{
// Useless property just to make the deserialization work
public Dictionary<string, Item> Data { get; }
public Item Item { get; }
public Result(Dictionary<string, Item> data)
{
Item = data.Values.Single();
}
}
class Item
{
public int Id { get; }
public Item(int id) => Id = id;
}
{
"data" : {
"dynamic_name": { "id" : 1 }
}
}