Inconsistent APIs between Api::patch_status and Api::replace_status (unergonomic)

[tgrushka]

Current and expected behavior

Current

Calling replace_status, one must use this unergonomic (and I argue, unnecessary) extra serde_json::to_vec wrapper:

instances
    .replace_status(
        &name,
        &PostParams::default(),
        serde_json::to_vec(&serde_json::json!({ "status": initial_status }))?,
    )
    .await?;

Expected

But patch_status is much more idiomatic and should be the way that replace_status also works:

instances
    .patch_status(
        &name,
        &PatchParams::default(),
        &Patch::Merge(&serde_json::json!({ "status": ready_status })),
    )
    .await?;

Nice to have in the future (but would be a feature request and a major API change)

Why do we need the serde_json::json!({ "status": ready_status }) wrapper at all, when it should just be type-friendly (since Api<K> knows the type) and it could just be: .patch_status(... &Patch::Merge(ready_status)) or similar?

Possible solution

If it must use serde_json::to_vec internally, just make that internal. But I thought the kubernetes API just takes JSON anyway, so why does the API user need to convert it to the u8 vec first?

Additional context

No response

Environment

k3s / k3d -- but not really relevant, this is an API semantics issue

Configuration and features

No response

Affected crates

No response

Would you like to work on fixing this bug?

maybe