How to handle keys pressed almost in the same time? - codenameone

Question

I'm trying to resolve a problem with the search bar. It works but the problem is that if I press two keys almost at the same time, the app will only search the words with the first key pressed.
Here are the logs:
In this one, it works when I press the P then R:
[EDT] 0:4:9,283 - p
[EDT] 0:4:9,348 - 10
[EDT] 0:4:9,660 - pr
[EDT] 0:4:9,722 - 3
The second one doesn't because I press P and R nearly at the same time:
[EDT] 0:4:35,237 - p
[EDT] 0:4:35,269 - pr
[EDT] 0:4:35,347 - 0
[EDT] 0:4:35,347 - 10
The logs here are generated to show the String searched and the result size. As you can see, the first case get results before typing the next char and the second case got all results when the two chars are typed.
The main problem is that in the second case, results from the 'p' String are shown instead of those of 'pr'.
I'm using the searchbar from the Toolbar API with addSearchCommand and an InfiniteContainer to show result data.
Could it be a problem in the order of the events from the addSearchCommand are treated ?
EDIT: Here is the client side code. Server side it's just a simple rest service call which fetch the data from the database.
public static ArrayList<Patient>getSearchedPatient(int index,int amount, String word)
{
ArrayList<Patient> listPatient = null;
Response reponse;
try {
reponse = RestManager.executeRequest(
Rest.get(server + "/patients/search")
.queryParam("index", String.valueOf(index))
.queryParam("amount", String.valueOf(amount))
.queryParam("word", word),
RequestResult.ENTITIES_LIST,
Patient.class);
listPatient = (ArrayList<Patient>)reponse.getResponseData();
Log.p(""+listPatient.size());
} catch (RestManagerException e) {
LogError("", e);
}
return listPatient;
}
private static Response executeRequest(RequestBuilder req, RequestResult type, Class objectClass) throws RestManagerException
{
Response response = null;
try {
switch (type) {
case BYTES:
response = req.getAsBytes();
break;
case JSON_MAP:
response = req.acceptJson().getAsJsonMap();
break;
case ENTITY:
response = req.acceptJson().getAsProperties(objectClass);
break;
case ENTITIES_LIST:
response = req.acceptJson().getAsPropertyList(objectClass);
break;
default:
case STRING:
response = req.getAsString();
break;
}
} catch (Exception e) {
log().error("Erreur à l'exécution de la requête", e);
response = null;
}
if(response == null)
return null;
return response;
}

Answer 1

So the trick here is a simple one. Don't make a request... Most users type fast enough to saturate your network connection speed so you will see completion suggestions referring to things that are no longer relevant.
This is a non-trivial implementation which I discuss in-depth in the Uber book where such a feature is implemented.
The solution is to send a request after a delay while caching responses to avoid double requests and ideally canceling request in progress when applicable. The solution in the Uber book does all 3 I'll try to cover just the basics in this mockup code. First you need a field for the timer and current request. Ideally you would also have a Map containing cached data:
private UITimer delayedRequest;
private String currentSearch;
private Map<String, String> searchCache = new HashMap<>();
Then you need to bind a listener like this:
tb.addSearchCommand(e -> {
String s = (String)e.getSource();
if(s == null) {
if(delayedRequest != null) {
delayedRequest.cancel();
delayedRequest = null;
}
return;
}
if(currentSearch != null && s.equals(currentSearch)) {
return;
}
if(delayedRequest != null) {
delayedRequest.cancel();
delayedRequest = null;
}
currenSearch = s;
delayedRequest = UITimer.timer(100, false, () -> {
doSearchCode();
});
});
I didn't include here usage of the cache which you need to check within the search method and fill up in the result code. I also didn't implement canceling requests already in progress.