Make execute entry point optional

[webmaster128]

It is encouraged to have an execute entry point but there are contracts that don't need one. We have some of those contracts here as development contracts. Those are exotic use case, but still they exist.

burner

#[entry_point]
pub fn execute(
    _deps: DepsMut,
    _env: Env,
    _info: MessageInfo,
    _msg: ExecuteMsg,
) -> StdResult<Response> {
    Err(StdError::generic_err(
        "You can only use this contract for migrations",
    ))
}

crypto-verify

#[entry_point]
pub fn execute(
    _deps: DepsMut,
    _env: Env,
    _info: MessageInfo,
    _msg: ExecuteMsg,
) -> StdResult<Response> {
    Ok(Response::default())
}

The two example s already show a dilemma: is an execution with an arbitrary message a no-op (success) or an error?

There is no reason execute must always exist. Just like query, it can be encouraged but optional.

[ethanfrey]

It is more of a sanity check that the contract author didn't name it wrong.
But #[entry_point] will enforce proper function signatures.

So this is only need to keep devs from uploading contracts but forgetting execute? That should be caught much earlier in the dev cycle. Fine to drop.