It would be nice if single-case discriminated unions was serialized directly:
type PersonId = PersonId of int
type Person =
{ Id: PersonId
Name: string }{"id": 123, "name": "Foo"}instead of
{"id": {"PersonId": 123}, "name": "Foo"}
Edit: I realize there's some cases where this will do something very different from what's intended; e.g. if you have
type AuthType =
| ApiToken of string
With the intention of adding more cases in the future, and then you'll break the serializer when you add the next case.
So I'm propising that we only do the single-case DU "optimization" when: There's one and only one case of course, and that case only has 1 field(?), and one or both of the following:
- The name of the case is the same as the type. So
type UserId = UserId of string. If it's type UserId = PersonId of string, then that's a no-match. (?)
- When the argument is a "primitive" – string, number, long, etc. I can't immediatly think of a case where you want a single-case DU for something else, but I might be wrong about that (?).
I think I like 1 more, even though it's slightly more "magical", but I think it's most pragmatic way of doing it. What do you think?
It would be nice if single-case discriminated unions was serialized directly:
{"id": 123, "name": "Foo"}instead of
{"id": {"PersonId": 123}, "name": "Foo"}Edit: I realize there's some cases where this will do something very different from what's intended; e.g. if you have
With the intention of adding more cases in the future, and then you'll break the serializer when you add the next case.
So I'm propising that we only do the single-case DU "optimization" when: There's one and only one case of course, and that case only has 1 field(?), and one or both of the following:
type UserId = UserId of string. If it'stype UserId = PersonId of string, then that's a no-match. (?)I think I like 1 more, even though it's slightly more "magical", but I think it's most pragmatic way of doing it. What do you think?