exchange.Go
Multi-threaded asynchronous support function that can convert the operations of all supported functions into asynchronous concurrent execution.
exchange.Go(method)
exchange.Go(method, ...args)Examples
-
exchange.Go()function usage example. When checking forundefined, you must usetypeof(xx) === "undefined", becausenull == undefinedholds true in JavaScript.javascriptfunction main(){ // The following four operations execute concurrently in asynchronous multi-threaded mode; they take no time and return immediately var a = exchange.Go("GetTicker") var b = exchange.Go("GetDepth") var c = exchange.Go("Buy", 1000, 0.1) var d = exchange.Go("GetRecords", PERIOD_H1) // Call the wait method to wait for the result of the asynchronous ticker retrieval var ticker = a.wait() // Returns the depth data; it may also return null if the retrieval fails var depth = b.wait() // Returns the order ID with a 1-second timeout; returns undefined on timeout. If the previous wait timed out, this object can continue calling wait var orderId = c.wait(1000) if(typeof(orderId) == "undefined") { // Timed out, retrieve again orderId = c.wait() } var records = d.wait() }pythondef main(): a = exchange.Go("GetTicker") b = exchange.Go("GetDepth") c = exchange.Go("Buy", 1000, 0.1) d = exchange.Go("GetRecords", PERIOD_H1) ticker, ok = a.wait() depth, ok = b.wait() orderId, ok = c.wait(1000) if ok == False: orderId, ok = c.wait() records, ok = d.wait()rustfn main() { // In Rust, exchange.Go uses a typed form: use the Go:: method token to specify the concurrent function; pass () for no arguments and a tuple for arguments // The following four operations execute concurrently in asynchronous multi-threaded mode; they take no time and return immediately let a = exchange.Go(Go::GetTicker, ()); let b = exchange.Go(Go::GetDepth, ()); // There is no Buy token in Rust; it is equivalent to CreateOrder, where the first argument "" indicates the current trading pair let c = exchange.Go(Go::CreateOrder, ("", "buy", 1000, 0.1)); let d = exchange.Go(Go::GetRecords, (PERIOD_H1,)); // Call the wait method to wait for the result of the asynchronous ticker retrieval; wait(0) blocks until the concurrent thread finishes running (corresponding to the parameterless wait() in JS) let ticker = a.wait(0); // Returns the depth data; it may also return Err if the retrieval fails let depth = b.wait(0); // Returns the order ID with a 1-second timeout; returns Err on timeout. If the previous wait timed out, this object can continue calling wait // Note: Err may also indicate that the order placement itself failed (indistinguishable from a timeout); in this case, calling wait again will return Err and log the error message let mut orderId = c.wait(1000); if orderId.is_err() { // Timed out, retrieve again orderId = c.wait(0); } let records = d.wait(0); }c++void main() { auto a = exchange.Go("GetTicker"); auto b = exchange.Go("GetDepth"); auto c = exchange.Go("Buy", 1000, 0.1); auto d = exchange.Go("GetRecords", PERIOD_H1); Ticker ticker; Depth depth; Records records; TId orderId; a.wait(ticker); b.wait(depth); if(!c.wait(orderId, 300)) { c.wait(orderId); } d.wait(records); } -
Calling the
wait()method on a released concurrent object will raise an error:javascriptfunction main() { var d = exchange.Go("GetRecords", PERIOD_H1) // Wait for the K-line data results to return var records = d.wait() // Calling wait again here on an asynchronous operation that has already been waited on and finished will return null and log an error message var ret = d.wait() }pythondef main(): d = exchange.Go("GetRecords", PERIOD_H1) records, ok = d.wait() ret, ok = d.wait()rustfn main() { // In Rust, exchange.Go uses a typed syntax: specify the concurrent function via the Go:: method token let d = exchange.Go(Go::GetRecords, (PERIOD_H1,)); // Wait for the K-line data results to return; wait(0) blocks until execution completes (equivalent to JS's parameterless wait()) let records = d.wait(0); // Calling wait again here on an asynchronous operation that has already been waited on and finished will return Err and log an error message let ret = d.wait(0); }c++void main() { auto d = exchange.Go("GetRecords", PERIOD_H1); Records records; d.wait(records); Records ret; d.wait(ret); } -
Concurrently retrieve market data from multiple exchanges:
javascriptfunction main() { while(true) { var beginTS = new Date().getTime() var arrRoutine = [] var arrTicker = [] var arrName = [] for(var i = 0; i < exchanges.length; i++) { arrRoutine.push(exchanges[i].Go("GetTicker")) arrName.push(exchanges[i].GetName()) } for(var i = 0; i < arrRoutine.length; i++) { arrTicker.push(arrRoutine[i].wait()) } var endTS = new Date().getTime() var tbl = { type: "table", title: "Market Data", cols: ["Index", "Name", "Last Price"], rows: [] } for(var i = 0; i < arrTicker.length; i++) { tbl.rows.push([i, arrName[i], arrTicker[i].Last]) } LogStatus(_D(), "Total time for concurrent ticker retrieval:", endTS - beginTS, "ms", "\n", "`" + JSON.stringify(tbl) + "`") Sleep(500) } }pythonimport time import json def main(): while True: beginTS = time.time() arrRoutine = [] arrTicker = [] arrName = [] for i in range(len(exchanges)): arrRoutine.append(exchanges[i].Go("GetTicker")) arrName.append(exchanges[i].GetName()) for i in range(len(exchanges)): ticker, ok = arrRoutine[i].wait() arrTicker.append(ticker) endTS = time.time() tbl = { "type": "table", "title": "Market Data", "cols": ["Index", "Name", "Last Price"], "rows": [] } for i in range(len(arrTicker)): tbl["rows"].append([i, arrName[i], arrTicker[i]["Last"]]) LogStatus(_D(), "Total time for concurrent ticker retrieval:", endTS - beginTS, "seconds", "\n", "`" + json.dumps(tbl) + "`") Sleep(500)rustfn main() { loop { let beginTS = UnixNano() / 1000000; let mut arrRoutine = Vec::new(); let mut arrTicker = Vec::new(); let mut arrName = Vec::new(); for e in exchanges.iter() { // In Rust, exchange.Go is a typed form; the token is Go::GetTicker arrRoutine.push(e.Go(Go::GetTicker, ())); arrName.push(e.GetName()); } // On failure, record None as a placeholder to stay index-aligned with arrName for r in arrRoutine.iter() { arrTicker.push(r.wait(0).ok()); } let endTS = UnixNano() / 1000000; // Rust has no built-in JSON serialization; use format! to assemble the table's JSON text let mut rows = String::new(); for i in 0..arrTicker.len() { if let Some(ticker) = &arrTicker[i] { if !rows.is_empty() { rows.push(','); } rows += &format!(r#"[{}, "{}", {}]"#, i, arrName[i], ticker.Last); } } let tbl = format!(r#"{{"type": "table", "title": "Market Data", "cols": ["Index", "Name", "Last Price"], "rows": [{}]}}"#, rows); LogStatus!(_D(None), "Total time for concurrent ticker retrieval:", endTS - beginTS, "ms", "\n", format!("`{}`", tbl)); Sleep(500); } }c++void main() { while(true) { int length = exchanges.size(); auto beginTS = UnixNano() / 1000000; vector<Ticker> arrTicker(length); vector<string> arrName(length); // Note: run the exchanges[n].Go function once for each exchange object you add. This example requires adding four exchange objects; adjust as needed auto r0 = exchanges[0].Go("GetTicker"); auto r1 = exchanges[1].Go("GetTicker"); auto r2 = exchanges[2].Go("GetTicker"); auto r3 = exchanges[3].Go("GetTicker"); vector<GoObj*> arrRoutine = {&r0, &r1, &r2, &r3}; for(int i = 0; i < length; i++) { arrName[i] = exchanges[i].GetName(); } for(int i = 0; i < length; i++) { Ticker ticker; arrRoutine[i]->wait(ticker); arrTicker[i] = ticker; } auto endTS = UnixNano() / 1000000; json tbl = R"({ "type": "table", "title": "Market Data", "cols": ["Index", "Name", "Last Price"], "rows": [] })"_json; for(int i = 0; i < length; i++) { json arr = R"(["", "", ""])"_json; arr[0] = str_format("%d", i); arr[1] = arrName[i]; arr[2] = str_format("%f", arrTicker[i].Last); tbl["rows"].push_back(arr); } LogStatus(_D(), "Total time for concurrent ticker retrieval:", str_format("%d", endTS - beginTS), "ms", "\n", "`" + tbl.dump() + "`"); Sleep(500); } } -
Concurrently call the
exchange.IO("api", ...)function:javascriptfunction main() { /* Test the OKX futures order placement endpoint POST /api/v5/trade/order */ var beginTS = new Date().getTime() var param = {"instId":"BTC-USDT-SWAP","tdMode":"cross","side":"buy","ordType":"limit","px":"16000","sz":"1","posSide":"long"} var ret1 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", JSON.stringify(param)) var ret2 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", JSON.stringify(param)) var ret3 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", JSON.stringify(param)) var id1 = ret1.wait() var id2 = ret2.wait() var id3 = ret3.wait() var endTS = new Date().getTime() Log("id1:", id1) Log("id2:", id2) Log("id3:", id3) Log("Concurrent order time:", endTS - beginTS, "ms") }pythonimport time import json def main(): beginTS = time.time() param = {"instId":"BTC-USDT-SWAP","tdMode":"cross","side":"buy","ordType":"limit","px":"16000","sz":"1","posSide":"long"} ret1 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", json.dumps(param)) ret2 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", json.dumps(param)) ret3 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", json.dumps(param)) id1, ok1 = ret1.wait() id2, ok2 = ret2.wait() id3, ok3 = ret3.wait() endTS = time.time() Log("id1:", id1) Log("id2:", id2) Log("id3:", id3) Log("Concurrent order time:", endTS - beginTS, "seconds")rustfn main() { /* Test the OKX futures order placement endpoint POST /api/v5/trade/order */ let beginTS = UnixNano() / 1000000; // Rust does not support JSON serialization, so construct the parameters directly using a raw string let param = r#"{"instId":"BTC-USDT-SWAP","tdMode":"cross","side":"buy","ordType":"limit","px":"16000","sz":"1","posSide":"long"}"#; // In Rust, exchange.Go uses a typed form: the token is Go::IO, and the parameters are passed as a tuple let ret1 = exchange.Go(Go::IO, ("api", "POST", "/api/v5/trade/order", "", param)); let ret2 = exchange.Go(Go::IO, ("api", "POST", "/api/v5/trade/order", "", param)); let ret3 = exchange.Go(Go::IO, ("api", "POST", "/api/v5/trade/order", "", param)); let id1 = ret1.wait(0); let id2 = ret2.wait(0); let id3 = ret3.wait(0); let endTS = UnixNano() / 1000000; Log!("id1:", id1); Log!("id2:", id2); Log!("id3:", id3); Log!("Concurrent order time:", endTS - beginTS, "ms"); }c++void main() { auto beginTS = UnixNano() / 1000000; json param = R"({"instId":"BTC-USDT-SWAP","tdMode":"cross","side":"buy","ordType":"limit","px":"16000","sz":"1","posSide":"long"})"_json; auto ret1 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", param.dump()); auto ret2 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", param.dump()); auto ret3 = exchange.Go("IO", "api", "POST", "/api/v5/trade/order", "", param.dump()); json id1 = R"({})"_json; json id2 = R"({})"_json; json id3 = R"({})"_json; ret1.wait(id1); ret2.wait(id2); ret3.wait(id3); auto endTS = UnixNano() / 1000000; Log("id1:", id1); Log("id2:", id2); Log("id3:", id3); Log("Concurrent order time:", endTS - beginTS, "ms"); } -
Testing the automatic release mechanism
javascriptfunction main() { var counter = 0 var arr = [] // Variables used to test persistently referencing concurrent objects var symbols = ["BTC_USDT", "ETH_USDT", "SOL_USDT", "LTC_USDT", "EOS_USDT"] while (true) { var arrRoutine = [] for (var symbol of symbols) { var r = exchange.Go("GetTicker", symbol) arrRoutine.push(r) // Record the concurrent object, used to call the r.wait() function to get the result; cleared every loop iteration // arr.push(r) // If this line is used, the runtime will persistently reference the concurrent objects, preventing them from being automatically released; when the number of concurrent tasks exceeds 2000, it will report an error: ```InternalError: too many routine wait, max is 2000```. counter++ } // Iterate over arrRoutine and call r.wait() to get the results LogStatus(_D(), "routine number:", counter) Sleep(50) } }rustfn main() { let mut counter = 0; let mut arr: Vec<TypedRoutine<Go::GetTicker>> = Vec::new(); // Variables used to test persistently referencing concurrent objects let symbols = ["BTC_USDT", "ETH_USDT", "SOL_USDT", "LTC_USDT", "EOS_USDT"]; loop { let mut arrRoutine = Vec::new(); for symbol in symbols { // In Rust, exchange.Go uses a typed form, with the token being Go::GetTicker let r = exchange.Go(Go::GetTicker, (symbol,)); arrRoutine.push(r); // Record the concurrent object, used to call the r.wait(0) function to get the result; cleared every loop iteration // arr.push(r); // If this line is used, the runtime will persistently reference the concurrent objects, preventing them from being automatically released; when the number of concurrent tasks exceeds 2000, it will report an error: InternalError: too many routine wait, max is 2000. counter += 1; } // Iterate over arrRoutine and call r.wait(0) to get the results LogStatus!(_D(None), "routine number:", counter); Sleep(50); } }
Returns
| Type | Description |
object | The |
Arguments
| Name | Type | Required | Description |
method | string | Yes | The |
arg | string / number / bool / object / array / function / any (any type supported by the platform) | No | The parameters of the concurrent execution function. The |
See Also
Mail_Go HttpQuery_Go EventLoop exchange.IO (API rate limiting control)
Remarks
This function only creates multi-threaded execution tasks when running in live trading. Backtesting does not support multi-threaded concurrent execution of tasks (it can be used in backtesting, but is still executed sequentially).
After the exchange.Go() function returns an object, you can call its wait() function through that object to obtain the data returned by the thread. When the concurrent multi-threaded tasks have finished executing and the related variables are no longer referenced, the underlying system will automatically handle resource reclamation.
The wait() method supports a timeout parameter:
-
Do not set the timeout parameter, i.e.
wait(), or set the timeout parameter to 0, i.e.wait(0). In this case, thewait()function will block and wait until the concurrent thread finishes running, and return the execution result of the concurrent thread. -
Set the timeout parameter to -1, i.e.
wait(-1). In this case, thewait()function will return immediately. The return value differs across programming languages; for details, please refer to the call examples in this section. -
Set a specific timeout parameter, i.e.
wait(300). In this case, thewait()function will wait at most 300 milliseconds before returning.
Although the underlying system has an automatic reclamation mechanism, if the related variables are continuously referenced, the concurrent threads will not be released. When the number of concurrent threads exceeds 2000, an error will be reported: "too many routine wait, max is 2000".
Supported functions: GetTicker, GetDepth, GetTrades, GetRecords, GetAccount, GetOrders, GetOrder, CancelOrder, Buy, Sell, GetPositions, IO, etc. When these functions are called concurrently, they are all executed based on the current exchange exchange object.
The difference between the Python language and the JavaScript language is that in Python, the wait() function of a concurrent object returns two values: the first is the result returned by the asynchronous API call, and the second indicates whether the asynchronous call is completed.
python
def main():
d = exchange.Go("GetRecords", PERIOD_D1)
# ok is guaranteed to return True, unless the strategy is stopped
ret, ok = d.wait()
# If the wait times out, or you wait on an instance that has already finished, ok returns False
ret, ok = d.wait(100)