Inscribe vs serialize

[cmlsharp]

It's not super clear to me what the difference between Serialize and Inscribe is and why the latter is preferred. I get that inscribe allows you to add in additional contextual information not stored in the struct, but serde also lets you do this. Is the idea that Inscribe just makes this nicer syntactically?

In other words, are the following two implementations effectively equivalent (not in that they produce the exact same transcript, but in the guarantees being provided)?

#[derive(Serialize)]
struct Data {
    x: i32,
    y: i32,
    #[serde(serialize_with = "Data::add_context")]
    context: () // note this is a zst so there is no runtime cost to this member

#[derive(Serialize)]
struct Context {
    // ... 
}

impl Data {
    fn new(x: i32, y: i32) -> Self {
        Self {x, y, context: () }
    }
    fn add_context<S: Serializer>(_: &(), s: S) -> Result<S::Ok, S::Err> {
        let ctx = // ...
        ctx.serialize(s)
    }
}

// ...
let data = Data::new(x,y);
transcript.add_serial(label, data)

and

#[derive(Inscribe)]
#[inscribe_addl("add_context")]
struct Data {
    #[inscribe(serial)]
    x: i32,
    #[inscribe(serial)]
    y: i32,
}


#[derive(Inscribe)]
struct Context {
    // ...
}

impl Data {
    fn new(x: i32, y: i32) -> Self {
        Self { x, y }
    }
    fn add_context(&self) -> Context {
        let ctx = // ...
        ctx
    }
}

// ...

let data = Data::new(x,y);
transcript.add(label, data);