Cancelling Market Depth requests

[dkimot]

Using the MPE below, when calling subscription.cancel() TWS returns "4\02\09000\0300\0Can't find EId with tickerId:9000\0". This is true if cancel is called directly or if the subscription is passed to drop.

use std::time::Duration;
use std::thread::sleep;

use ibapi::Client;

fn main() {
    env_logger::init();

    let connection_url = "127.0.0.1:4001";

    let client = Client::connect(connection_url, 105).expect("connection to TWS failed!");
    println!("Successfully connected to TWS at {connection_url}");

    let contract = ibapi::contracts::Contract::stock("TSLA");
    let subscription = client.market_depth(&contract, 5, true).expect("error requesting market depth");
    if let Some(row) = subscription.try_next() {
        println!("row: {row:?}");
    }

    println!("sleeping...");
    sleep(Duration::from_secs(5));

    if let Some(error) = subscription.error() {
        println!("error: {:?}", error);
    }

    subscription.cancel();
    sleep(Duration::from_secs(1));
}

I'm not sure where things are going wrong and the IBKR docs on this are sparse to say the least.

[dkimot]

Figured it out. The impl for MarketDepths calls encoders::encode_cancel_tick_by_tick(request_id). This sends the message ID 98 which is wrong. Changing this to a new method cancel_market_depth(request_id) with the provided implementation works. This isn't a perfect solution since I'm hard coding the final 1 -- I believe this is to denote that I'm requesting a "smart" market depth.

cancel_market_depth implementation:

pub(super) fn encode_cancel_market_depth(request_id: i32) -> Result<RequestMessage, Error> {
    let mut message = RequestMessage::new();

    const VERSION: i32 = 1;

    message.push_field(&OutgoingMessages::CancelMarketDepth);
    message.push_field(&VERSION);
    message.push_field(&request_id);
    message.push_field(&1);

    Ok(message)
}

I validated the expected message by using the python ib_sync library. I don't have access to the TWS api repo so I'm guessing on the message structure.

[wboayue]

Thanks for the detailed writeup. I'll verify the arguments with the TWS api source and fix this in the next patch.