I am trying to create a ValueProxy which holds some basic information about a search a user is performing. For some reason GWT wants it to be an EntityProxy but I dont see why (nor does it make sense for this class to be an EntityProxy).
// FilterProxy extends ValueProxy
#ProxyFor(DayFilter.class)
public interface DayFilterProxy extends FilterProxy {
void setFilterValue(Date day);
Date getFilterValue();
}
public class DayFilter extends Filter {
public DayFilter() {
setOperator(FilterOperator.GREATER_THAN_OR_EQUAL);
setField("dateRequested");
}
public void setFilterValue(Date date) {
this.value = date;
}
public Date getFilterValue() {
return value;
}
}
public interface PaginationRequest<T> extends RequestContext {
Request<List<T>> paginate(int offset, int limit, String sortColumn,
boolean isSortAscending, List<FilterProxy> filters);
Request<Integer> count(List<FilterProxy> filters);
}
#Service(value=TripService.class, locator=SchedgyServiceLocator.class)
public interface TripRequest extends PaginationRequest<TripProxy> {
Request<TripProxy> save(TripProxy trip);
}
Within the activity that is sending this back to the server:
// Request is a TripRequest
DayFilterProxy filter = request.create(DayFilterProxy.class);
This results in:
java.lang.AssertionError: com.schedgy.trip.dao.filter.trip.proxy.DayFilterProxy is not an EntityProxy type
at com.google.web.bindery.requestfactory.shared.impl.IdFactory.asEntityProxy(IdFactory.java:66)
at com.google.web.bindery.requestfactory.shared.impl.IdFactory.createId(IdFactory.java:229)
at com.google.web.bindery.requestfactory.shared.impl.IdFactory.allocateId(IdFactory.java:41)
at com.google.web.bindery.requestfactory.shared.impl.AbstractRequestContext.create(AbstractRequestContext.java:478)
at com.schedgy.trip.client.activity.TripsActivity.getFilters(TripsActivity.java:56)
Any ideas? Its got to be something obvious that I am just overlooking as I have ValueProxies working elsewhere in the code.
Could it be that your DayFilterProxy is not referenced at all from the RequestContext?
Related
I am using two database oracle and sql server. I am making system generic using dependency injection.Code is following
public interface IDatabases
{
string GetEmployeeFullName();
}
public class OracleDB:IDatabases
{
public string GetEmployeeFullName()
{
return "Name oracle";
}
}
public class SqlServerDB : IDatabases
{
public string GetEmployeeFullName()
{
return "Name sql server";
}
}
public class RegistrationStaff
{
private IDatabases objDatabase;
public RegistrationStaff(IDatabases vobjDataBase)
{
this.objDatabase = vobjDataBase;
}
}
I need another function GetEmployeeId in sql server class which will return employee id which is available in sql server database.I do not want this function implementation in oracle.How can I use interface segregation with dependency injection and implement in RegistrationStaff class .
public interface IsqlServer:IDatabases
{
void GetEmployeeId();
}
I want only dependency injection using constructor
I think you are already on the right way. Make a new interface for the segregation, add the new method and let the SQL server class inherit from it. So after all you have to cast the database object in the Registration class to call the method. But I can't see a way without casting if you don't put the method at top level.
public interface IDatabases
{
string GetEmployeeFullName();
}
public interface ISQLDatabase : IDatabases
{
int GetEmployeeId();
}
public class OracleDB : IDatabases
{
public string GetEmployeeFullName()
{
return "Name oracle";
}
}
public class SqlServerDB : ISQLDatabase
{
public string GetEmployeeFullName()
{
return "Name sql server";
}
public int GetEmployeeId()
{
return 1;
}
}
public class RegistrationStaff
{
private IDatabases objDatabase;
public RegistrationStaff(IDatabases vobjDataBase)
{
this.objDatabase = vobjDataBase;
if (this.objDatabase is ISQLDatabase)
{
Console.WriteLine(((ISQLDatabase)this.objDatabase).GetEmployeeId());
}
}
}
Hi after countless hours, I figured out what I really what my problem is but still cannot find an answer.
#Override
public void onStampResult(StampResult result) {
}
onStampResult returns a Class StampResult with the follwowing parameters:
public class StampResult implements Serializable {
public SnowShoeError error;
public SnowShoeStamp stamp;
public boolean secure;
public String receipt;
public Date created;
}
SnowShoeStamp Class is:
public class SnowShoeStamp implements Serializable {
public String serial;
}
And SnowShoeError Class is:
public class SnowShoeError implements Serializable {
public int code;
public String message;
}
In onStampResult I can write down logic depending on the output of result.
On Success ´stamp´ gets initialized and ´error´ does not exist.
On Error, stamp does not exist and error gets initialized.
The result gets parsed to from JSON to the Class in the following way:
try {
stampResult = gson.fromJson(result, StampResult.class);
} catch (JsonSyntaxException jsonException) {
stampResult = new StampResult();
stampResult.error = new SnowShoeError();
stampResult.error.message = result;
}
mOnStampListener.onStampResult(stampResult);
mStampBeingChecked = false;
}
How do I test if either error or stamp exists without getting a NullPointerExeption?
Unless I misunderstood your question, you simply need to check for null.
In order to handle the different cases, you could do the following:
#Override
public void onStampResult(StampResult result) {
if (result.error == null){
SnowShoeStamp stamp = result.stamp;
// Process stamp
} else {
SnowShoeError error = stampResult.error;
// Process error
}
}
I have two classes:
DocumentState and ElectronicDocumentState.
They look like this:
public class DocumentState extends LabeledEnum {
public static final DocumentState CREATED = new DocumentState("created");
// ..... - 15 other statuses
}
ElectronicDocumentState also extends LabeledEnum and has its own statuses, some are common like created, other are unique.
Then I have plenty of methods in code that take DocumentState as a parameter or are returning DocumentState as result. Now they should work also with ElectronicDocumentState.
Also I have plenty places that do:
if (DocumentSate.CREATED.equals(doc.getState()) || DocumentState.DELETED.equals(doc.getState())) {
//do something with document
}
I want to avoid 'if' and avoid creating new methods for ElectronicDocumentState as more state can appear in future.
How would you do that ?
So using the below example, how would you refactor it so it could work with DocumentState and ElectronicDocumentState? I have plenty of such methods that now should also work with ElectronicDocumentState. Someone asks us to mix electronicDocuments with documents in business logic:
private DocumentState getDocumentStateForDetails(Document doc, DocumentState sourceState) {
if (DocumentState.CREATED.equals(doc.getDocumentState())) {
if (sourceState.equals(DocumentState.CREATED)) {
return DocumentState.CREATED;
} else {
return DocumentState.BLOCKED;
}
} else {
return sourceState.getDocumentState();
}
}
If you are worrying about further extension of your state model, I'd suggest you to think about using inheritance instead of if/switch and instanceof by splitting the document class to Value and State, for instance. If you have the same set of available actions for any state of document, just use classical State pattern, otherwise, each State may have its own set of available actions:
public class Document {
public static CreatedDocument<Document> create(String author) {
return new CreatedDocument<>(new Document(author));
}
private String author;
//...
private Document(String author) {
//...
}
}
public class ElectronicDocument extends Document {
public static CreatedElectronicDocument create(String author, String url) {
return new CreatedElectronicDocument(author, url);
}
private String url;
//...
public ElectronicDocument(String author, String url) {
//...
}
//...
}
public interface DocumentState<T extends Document> {
T getDocument();
char getCode(); // might be needed for something like logging?
}
public abstract class AbstractDocumentState<T extends Document> implements DocumentState<T> {
protected final T document;
protected AbstractDocumentState(T document) {
this.document = document;
}
#Override
public T getDocument() {
return document;
}
}
public class CreatedDocument<T extends Document> extends AbstractDocumentState<T> {
public CreatedDocument(T document) {
super(document);
}
#Override
public char getCode() {
return 'C';
}
public DocumentState<T> delete() {
return new DeletedDocument<>(document);
}
}
public class CreatedElectronicDocument extends CreatedDocument<ElectronicDocument> {
public CreatedElectronicDocument(String author, String url) {
super(new ElectronicDocument(author, url));
}
public DownloadElectronicDocument download() {
return new DownloadElectronicDocument(document);
}
}
public class DownloadElectronicDocument extends AbstractDocumentState<ElectronicDocument> {
public DownloadElectronicDocument(ElectronicDocument document) {
super(document);
// DO DOWNLOAD HERE
}
#Override
public char getCode() {
return 'L';
}
public DocumentState<ElectronicDocument> delete() {
return new DeletedDocument<>(document);
}
}
public class DeletedDocument<T extends Document> extends AbstractDocumentState<T> {
public DeletedDocument(T document) {
super(document);
// DO DELETE HERE
}
#Override
public char getCode() {
return 'D';
}
}
Not sure you need getCode() now, when you use inheritance. BTW, switch works faster than a set of if/else if.
If you'd like to stay with your enum classes, why wouldn't to extract common states to a super class?
public class CommonDocumentState extends LabeledEnum {
public static final CommonDocumentState CREATED = new CommonDocumentState ("created");
..... - n other statuses
}
public class DocumentState extends CommonDocumentState {
..... - m other statuses
}
public class ElectronicDocumentState extends CommonDocumentState {
..... - k other statuses
}
That's the only way to have such generic rules like
if (DocumentSate.CREATED.equals(doc.getState()) || DocumentState.DELETED.equals(doc.getState())) {
//do something with document
}
be working for both DocumentState and ElectronicDocumentState.
There are not enough information about your domain to provide a final answer, but I have some suggestions:
it seems that both DocumentState and ElectronicDocumentState inherit from LabeledEnum; if you want to manage both of them in your methods you can make ElectronicDocumentState inherit from DocumentState of it's feasibile merge the two classes. This will allow to pass ElectronicDocumentState or DocumentState in your methods and perhaps solve your second question.
if you want to avoid the if you can build a list of allowed method and check against the list, something like:
L
public class YourClass {
List<DocumentSate> allowedStates=//init here or in constructor
....
public void yourMethod(....) {
if (allowedStates.contains(doc.getState())) {
//do something
}
}
allowedStates may be factored out in separate class if it is a common case. If you find the refactoring feasible, may be you can check if you are dealing with a finite state machine and implement it (with the help of some exisiting libraries).
I want to use java interface in a way that i will make a call defining interface in my other class like 'private SoapURL soapURL;' and than i can access any class's method for example : i want to use login:-
private SoapURL soapURL;
SoapUrl = LoginSoap ();
String nameSpace = soapURL.getMethodName();
String url = soapURL.getUrl();
Is there any way to do something like this. I am sorry i am not very good with Object Oriented principles but if there is a solution for my problem i would like to know it. Thanks in advance.
public interface SoapURL {
public String getNameSpace();
public String getUrl();
public String getSoapAction();
public String getMethodName();
public String getTag();
}
LoginSoap class
public class LoginSoap implements SoapURL {
#Override
public String getNameSpace() {
return "https://host.com/MobileWFC/";
}
#Override
public String getUrl() {
return "https://host.com/MobileWFC/MobileWS.asmx";
}
#Override
public String getSoapAction() {
return "https://host.com/MobileWFC/UserControl";
}
#Override
public String getMethodName() {
return "UserControl";
}
#Override
public String getTag() {
return "Login Activity";
}
}
SignUpSoap class
public class SignUpSoap implements SoapURL {
#Override
public String getNameSpace() {
return "https://host.com/MobileWFC/";
}
#Override
public String getUrl() {
return "https://host.com/MobileWFC/MobileWS.asmx";
}
#Override
public String getSoapAction() {
return "https://host.com/MobileWFC/UserRegister";
}
#Override
public String getMethodName() {
return "UserRegister";
}
#Override
public String getTag() {
return "SignUp Activity";
}
}
ResetPasswordSoap class
public class ResetPasswordSoap implements SoapURL {
#Override
public String getNameSpace() {
return "https://host.com/MobileWFC/";
}
#Override
public String getUrl() {
return "https://host.com/MobileWFC/MobileWS.asmx";
}
#Override
public String getSoapAction() {
return "https://host.com/MobileWFC/UserPasswordReset";
}
#Override
public String getMethodName() {
return "UserPasswordReset";
}
#Override
public String getTag() {
return "Forget Password Activity";
}
}
Your implementation looks correct. To make use of it, you can do this in main:
SoapURL reset = new ResetPasswordSoap();
System.out.println(reset.getUrl());
This is a method of minimizing coupling in large systems. And reduces dependency between objects by making use of a common interface for groups of objects that work together. You might be new at Object oriented principles, but you are one step ahead of the game already
To pass it to a function, you do:
public JPanel resetPass(SoapURL reset) {
...
}
// In main:
JPanel resetPassPanel = resetPass(reset);
Just do, for example:
SoapURL example = new LoginSoap();
String a = example.getTag();
a should be equal to "Login Activity"
The main use of Interface is polymorphism, or the ability to perform the same
operation on a number of different objects,
which is exactly what you wanted in your scenario
Your approach is absolutely fine , just a modification needed
private SoapURL soapURL;
//SoapUrl = LoginSoap (); // This line should be replaced with the Below line
soapURL=new LoginSoap();
String nameSpace = soapURL.getMethodName();
String url = soapURL.getUrl();
Since LoginSoap, SignUpSoap,ResetPasswordSoap classes are implemented classes of SoapURL Interface , thus reference variable of SoapURL can store Object of any of these child classes
soapURL=new LoginSoap();//soapURL.someMethod will call method of LoginSoapClass
soapURL=new SignUpSoap();// will call method of SignUpSoap class
soapURL=new ResetPasswordSoap();
I have a Java client which wants to communicate with a device through messages over serial communication. The client should be able to use a clean API, abstracting the ugly details of the serial communication. The client can send many types of messages through that API and gets responses. I'm searching for advice which way is best to implement this API.
For simplicity, say we have only two message types: HelloMessage which triggers a HelloResponse and InitMessage which triggers an InitResponse (in reality, there are many more)
Designing the API (that is, the Java abstraction of the device) I could have:
One method per message type:
public class DeviceAPI {
public HelloResponse sendHello(HelloMessage){...}
public InitResponse sendInit(InitMessage){...}
... and many more message types ....
This is nicely type safe. (It could also be many times the same send() method, overloaded, but that's about the same). But it is very explicit, and not very flexible - we cannot add messages without modification of the API.
I could also have a single send method, which takes all message types:
class HelloMessage implements Message
class HelloResponse implements Response
...
public class DeviceAPI {
public Response send(Message msg){
if(msg instanceof HelloMessage){
// do the sending, get the response
return theHelloResponse
} else if(msg instanceof ...
This simplifies the API (only one method) and allows for additional Message types to be added later without changing the API. At the same time, it requires the Client to check the Response type and cast it to the right type.
Client code:
DeviceAPI api = new DeviceAPI();
HelloMessage msg = new HelloMessage();
Response rsp = api.send(msg);
if(rsp instanceOf HelloResponse){
HelloResponse hrsp = (HelloResponse)rsp;
... do stuff ...
This is ugly in my opinion.
What do you recommend? Are there other approaches which give cleaner results?
References welcome! How did others solve this?
Here is a way to do it in type-safe (and extensible) way using generics:
public interface MessageType {
public static final class HELLO implements MessageType {};
}
public interface Message<T extends MessageType> {
Class<T> getTypeClass();
}
public interface Response<T extends MessageType> {
}
public class HelloMessage implements Message<MessageType.HELLO> {
private final String name;
public HelloMessage(final String name) {
this.name = name;
}
#Override
public Class<MessageType.HELLO> getTypeClass() {
return MessageType.HELLO.class;
}
public String getName() {
return name;
}
}
public class HelloResponse implements Response<MessageType.HELLO> {
private final String name;
public HelloResponse(final String name) {
this.name = name;
}
public String getGreeting() {
return "hello " + name;
}
}
public interface MessageHandler<T extends MessageType, M extends Message<T>, R extends Response<T>> {
R handle(M message);
}
public class HelloMessageHandler
implements MessageHandler<MessageType.HELLO, HelloMessage, HelloResponse> {
#Override
public HelloResponse handle(final HelloMessage message) {
return new HelloResponse(message.getName());
}
}
import java.util.HashMap;
import java.util.Map;
public class Device {
#SuppressWarnings("rawtypes")
private final Map<Class<? extends MessageType>, MessageHandler> handlers =
new HashMap<Class<? extends MessageType>, MessageHandler>();
public <T extends MessageType, M extends Message<T>, R extends Response<T>>
void registerHandler(
final Class<T> messageTypeCls, final MessageHandler<T, M, R> handler) {
handlers.put(messageTypeCls, handler);
}
#SuppressWarnings("unchecked")
private <T extends MessageType, M extends Message<T>, R extends Response<T>>
MessageHandler<T, M, R> getHandler(final Class<T> messageTypeCls) {
return handlers.get(messageTypeCls);
}
public <T extends MessageType, M extends Message<T>, R extends Response<T>>
R send(final M message) {
MessageHandler<T, M, R> handler = getHandler(message.getTypeClass());
R resposnse = handler.handle(message);
return resposnse;
}
}
public class Main {
public static void main(final String[] args) {
Device device = new Device();
HelloMessageHandler helloMessageHandler = new HelloMessageHandler();
device.registerHandler(MessageType.HELLO.class, helloMessageHandler);
HelloMessage helloMessage = new HelloMessage("abhinav");
HelloResponse helloResponse = device.send(helloMessage);
System.out.println(helloResponse.getGreeting());
}
}
To add support for a new message type, implement MessageType interface to create a new message type, implement Message, Response and MessageHandler interfaces for the new MessageType class and register the handler for the new message type by calling Device.registerHandler.
I've got a fully working example now of what you want:
To define the types of messages:
public interface MessageType {
public static class INIT implements MessageType { }
public static class HELLO implements MessageType { }
}
Base Message and Response classes:
public class Message<T extends MessageType> {
}
public class Response<T extends MessageType> {
}
Create custom init messages and responses:
public class InitMessage extends Message<MessageType.INIT> {
public InitMessage() {
super();
}
public String getInit() {
return "init";
}
}
public class InitResponse extends Response<MessageType.INIT> {
public InitResponse() {
super();
}
public String getInit() {
return "init";
}
}
Create custom hello messages and responses:
public class HelloMessage extends Message<MessageType.HELLO> {
public HelloMessage() {
super();
}
public String getHello() {
return "hello";
}
}
public class HelloResponse extends Response<MessageType.HELLO> {
public HelloResponse() {
super();
}
public String getHello() {
return "hello";
}
}
The DeviceAPI:
public class DeviceAPI {
public <T extends MessageType, R extends Response<T>, M extends Message<T>> R send(M message) {
if (message instanceof InitMessage) {
InitMessage initMessage = (InitMessage)message;
System.out.println("api: " + initMessage.getInit());
return (R)(new InitResponse());
}
else if (message instanceof HelloMessage) {
HelloMessage helloMessage = (HelloMessage)message;
System.out.println("api: " + helloMessage.getHello());
return (R)(new HelloResponse());
}
else {
throw new IllegalArgumentException();
}
}
}
Note that it does require an instanceof-tree, but you need that to handle what kind of message it is.
And a working example:
public static void main(String[] args) {
DeviceAPI api = new DeviceAPI();
InitMessage initMsg = new InitMessage();
InitResponse initResponse = api.send(initMsg);
System.out.println("client: " + initResponse.getInit());
HelloMessage helloMsg = new HelloMessage();
HelloResponse helloResponse = api.send(helloMsg);
System.out.println("client: " + helloResponse.getHello());
}
Output:
api: init
client: init
api: hello
client: hello
UPDATE: Added example on how to get input from the messages the client wants to send.
You could have a system of message handlers, and your DeviceAPI could choose which handler is suitable for the incoming message; and delegate it to the appropriate message handler:
class DeviceAPI {
private List<Handler> msgHandlers = new ArrayList<Handler>();
public DeviceAPI(){
msgHandlers.add(new HelloHandler());
//Your other message handlers can be added
}
public Response send(Message msg) throws Exception{
for (Handler handler : msgHandlers) {
if (handler.canHandle(msg)){
return handler.handle(msg);
}
}
throw new Exception("No message handler defined for " + msg);
}
}
The HelloHandler would look like:
interface Handler<T extends Message, U extends Response> {
boolean canHandle(Message message);
U handle(T message);
}
class HelloHandler implements Handler<HelloMessage, HelloResponse> {
#Override
public boolean canHandle(Message message) {
return message instanceof HelloMessage;
}
#Override
public HelloResponse handle(HelloMessage message) {
//Process your message
return null;
}
}
Ditto for your other messages. I'm sure you could make it more elegant, but the idea still remains the same - donot have one monster method with ifs; instead use polymorphism.
I don't think this is ugly at all:
if(rsp instanceOf HelloResponse){
HelloResponse hrsp = (HelloResponse)rsp;
...
else if ...
as long as you don't have like 100 different responses. You can encapsulate many kind of responses in one, depending on the data they have. For example:
class GenericResponse{
private String message;
private int responseType;
...
}
I have developed some multiplayer games and this is a good way to do it.
In case you have too many different types of messages, you just can use generic java types, like the example of skiwi above.
hope it helps