Memory management using directors in SWIG C++ / Java - java

I am trying to create a Java wrapper for my C++ library using SWIG.
In order to get all the features I need within a Java programming environment, I need directors.
More specifically, I need Java users to be able to inherit from my C++ classes and implement certain methods.
In particular, one of the method that needs to be implemented is some sort of clone() method.
In C++, the user implementation provides an object pointer Base*. This pointer is then managed by the library itself.
Base* Derived::clone() {
return new Derived(*this);
}
The problem I have with the Java wrapper is that the Java proxy class for Base acquires the management of the corresponding C++ director class SwigDirector_Base, by default.
This is certainly suitable in the general case, but not in this particular user-defined clone() function.
My personal constraint is that the Java wrapper for my C++ library uses no Java-specific code, so the user implementation should look like:
class Derived {
...
Base clone() {
return new Derived(this);
}
}
So far, to make it work and avoid garbage collection of the copied Java instance, I have used the trick mentioned in http://www.swig.org/Doc2.0/SWIGDocumentation.html#CSharp_memory_management_member_variables.
And to make sure that the Java Derived class never deletes the corresponding C++ director class SwigDirector_Base, I have used %typemap(directorout) to set the value of the Java cMemOwn flag of the copied instance from the C++ wrapper code, that is in method SwigDirector_Base::clone().
I am not so happy with this solution as it applies to all methods returning a pointer to the Base class, whether it is a copy method or not...
Any idea on how to do this on a function-specific way? Or in any other way?

Related

Is there an object in java similar to the Kotlin data class?

I'm currently learning java in order to make an app on android and I checked that swift has a structure that stores information in memory. I'd like to know, if in java this type of object exists, because the class storage the reference on the memory. Also I checked that Kotlin has a data class, does java have a similar object?
No, there is nothing like that, but there are tools, that try to mimic this behavior, for example lombok. Using #Data annotation we're getting default constructor, getters, setters, toString, equals, hashCode. We can fine-tune it by using annotations like #Getter, #NoArgsConstructor etc.
Neither Java nor Kotlin have anything similar to those Swift types you are talking about. Assignment always copies references to an object, rather than the object itself. What Kotlin's data classes do is that they create a copy method (among other things) that allows you to explicitly make a copy of an object, but you still have to actually call the method.
val b = a // b and a point to the same object, even if it is a data class
val b = a.copy() // this is what you need to do to create a copy of a data class
Java assignment copies references, not objects, and the same is true for Kotlin. There is no way around this, because it is a feature of the language itself. Copy constructors and methods (like what Kotlin's data class gives you) are the closest thing you have to such a feature. To get something like this in Java without having to manually write the code everytime, you could look into Project Lombok.
Starting with Java 14 you will have access to Record immutable class. It is similar in concept to data class in Kotlin.

Java 8 add extension/default method to class

I am looking for a java equivalent to the C# extension methods feature. Now I have been reading about Java 8's default methods, but as far as I can see, I can only add these to interfaces...
...is there any language feature that will allow me to write an extension method for a final class that doesn't implement an interface? (I'd rather not have to wrap it...)
Java doesn't have extension methods. Default methods are not extension methods. Let's look at each feature.
Default methods
Both Java and C# support this feature
Problems solved:
Many objects may implement the same interface and all of them may use the same implementation for a method. A base class could solve this issue but only if the interface implementors don't already have a base class as neither java nor C# support multiple inheritance.
An API would like to add a method to an interface without breaking the API consumers. Adding a method with a default implementation solves this.
Java's or C#'s default methods are a feature to add a default implementation to an interface. So objects that extend an interface don't have to implement the method, they could just use the default method.
interface IA { default public int AddOne(int i) { return i + 1; } }
Any object that implements IA doesn't have to implement AddOne because there is a default method that would be used.
public class MyClass implements IA { /* No AddOne implementation needed */ }
C# now has this feature in C# 8 (or .Net 5)
C#'s Extension Method
Problems solved:
Ability to add methods to sealed classes.
Ability to add methods to classes from third-party libraries without forcing inheritance.
Ability to add methods to model classes in environments where methods in model classes are not allowed for convention reasons.
The ability for IntelliSense to present these methods to you.
Example: The type string is a sealed class in C#. You cannot inherit from string as it is sealed. But you can add methods you can call from a string.
var a = "mystring";
a.MyExtensionMethed()
Java lacks this feature and would be greatly improved by adding this feature.
Conclusion
There is nothing even similar about Java's default methods and C#'s extension method features. They are completely different and solve completely different problems.
C# extension methods are static and use-site, whereas Java's default methods are virtual and declaration-site.
What I believe you are hoping for is the ability to "monkey-patch" a method into a class you do not control, but Java does not give you that (by design; it was considered and rejected.)
Another benefit of default methods over the C# approach is that they are reflectively discoverable, and in fact from the outside, don't look any different from "regular" interface methods.
One advantage of C#'s extension methods over Java's default methods is that with C#'s reified generics, extension methods are injected into types, not classes, so you can inject a sum() method into List<int>.
Above all, the main philosophical difference between Java's default methods and C#'s extension methods is that C# lets you inject methods into types you do not control (which is surely convenient for developers), whereas Java's extension methods are a first-class part of the API in which they appear (they are declared in the interface, they are reflectively discoverable, etc.) This reflects several design principles; library developers should be able to maintain control of their APIs, and library use should be transparent -- calling method x() on type Y should mean the same thing everywhere.
C# extension methods are just syntactic sugar for static methods that take the extended type as first argument. Java default methods are something completely different. To mimic C# extension methods, just write usual static methods. You will not have the syntatic sugar, however; Java does not have this feature.
Java default methods are real virtual methods. For example, they can be overridden. Consider a class X inheriting from an interface I that declares a default foo() method. If X or any of its super classes declares no own foo() method, then X will get the foo() implementation of I. Now, a subclass Y of X can override X.foo() like a usual method. Thus, default methods are not only syntactic sugar. They are real extensions of the method overriding and inheritance mechanism that cannot be mimicked by other language features.
Default methods even require special VM support, so they are not even a compiler only feature: During class loading, the hierarchy of a class has to be checked to determine which default methods it will inherit. Thus, this decision is made at runtime, not at compile time. The cool thing about it is that you do not have to recompile a class when an interface it inherits gets a new default method: The VM will, at class load time, assign that new method to it.
It is possible to have extension methods with some tricks.
You may give a try to Lombok or XTend. Although extension methods don't come with the out of the box Java implementation, both Lombok and XTend offers a fully working solution.
Lombok is a simple standalone code processing framework, which makes most of the criticized Java specific hassle less painful, including extension methods:
https://projectlombok.org/features/experimental/ExtensionMethod.html
Xtend http://www.eclipse.org/xtend/ goes a few lightyears forward, and implements a language which is a combination of the best parts of modern languages such as Scala on top of Java and Java type system. This allows implementing some classes in Xtend and others in Java within the same project. The Xtend code complies to valid Java code, so no JVM magic happens under the hood. On the other hand, it is a little too much if you have only extension methods missing.
JPropel https://github.com/nicholas22/jpropel-light implements LINQ style extension methods in Java using Lombok. It may worth of a peek :)

Interface with no methods

Why do Java introduces some interface which has no methods defined in it? For example Cloneable, Serializable, Type and many more.
Second thing : In Class.class package there is one method defined registerNatives() without body and is called from static block but Class.class is not abstract but is final. Why so?
and Why Java need some method without body to be called from static block.?
Why do Java introduces some interface which has no methods defined in it?
This are called Tagged or Marker interface. These are not used for any use or operation. These methods are used to tag or marking a class. So that you can determine whether someclass is a child of those classes.
about the second question
If you look closely you can see the declaration is
private static native void registerNatives();
So registerNatives is a native methods.
So what is native methods. If you see this so question
The method is implemented in "native" code. That is, code that does
not run in the JVM. It's typically written in C or C++.
Native methods are usually used to interface with system calls or
libraries written in other programming languages.
So these methods are loaded from native codes. So you don't need to declare the body of the methods but still they are not abstract as they have their implementation from native codes.
Marker interface is used as a tag to inform a message to the java compiler so that it can add special behavior to the class implementing it. Java marker interface has no members in it.
The purpose of Marker interfaces is to force some kind of functionality in the classes by providing some functionality to a class if it implements the marker interface.
Read Java Marker Interface also see What is the use of marker interfaces in Java?
For the first one you are actually asking for a Marker Interface. Marker Interfaces are by design not supposed to add anything to behavior but support only polymorphic transformation of the object. e.g. Serializable makes an object capable of streaming across JVM boundaries. Marker interfaces follow the 'universal type substitution' philosophy.
For second one, you are actually asking for JNI. Java doesnot implement all its code in Java form. I mean in classes and code that follow Java syntax. Some time or the other you need to drill down to the native platform API to implement something for that API. e.g. sockets and TCP communication. It is this feature of Java that actually makes it platform independent. The JVM runtime is platform dependent as it uses platform based native methods and dll or .so libraries to implement and integrate with the platform. We as programmers call the high level Java SDK API calls.
One of the "clean" features of the Java programming language is that it mandates a separation between interfaces (pure behavior) and classes (state and behavior). Interfaces are used in Java to specify the behavior of derived classes.
Often you will come across interfaces in Java that have no behavior. In other words, they are just empty interface definitions. These are known as marker interfaces. Some examples of marker interfaces in the Java API include:
java.lang.Cloneable
java.io.Serializable
java.util.EventListener
Marker interfaces are also called "tag" interfaces since they tag all the derived classes into a category based on their purpose. For example, all classes that implement the Cloneable interface can be cloned (i.e., the clone() method can be called on them). The Java compiler checks to make sure that if the clone() method is called on a class and the class implements the Cloneable interface. For example, consider the following call to the clone() method on an object o:
SomeObject o = new SomeObject();
SomeObject ref = (SomeObject)(o.clone());
If the class SomeObject does not implement the interface Cloneable (and Cloneable is not implemented by any of the superclasses that SomeObject inherits from), the compiler will mark this line as an error. This is because the clone() method may only be called by objects of type "Cloneable." Hence, even though Cloneable is an empty interface, it serves an important purpose.
registerNatives()
native method are implemented in JVM itself.
What does the registerNatives() method do?
Why Java need some method without body to be called from static block.?
This is called from static block because we need to call this method when classes are loaded and not when it's instance is created.

How to declare and call a native method containing pointers in java

How to declare and call a native method in c containing pointers using java?
please it is urgent.Because there is no concept of pointers in java i am getting error.
Use JNI(Java Native Interfaces) to call native methods in java. Use this JNI specification pdf as a reference "java.sun.com/docs/books/jni/download/jni.pdf".
There is even a easier approach to the subject, if you or your boss are willing to pay 50$!
Is called eXcelsior xFunction. Try the evaluation version.
Works very well and is simple to use.
The only thing xFunction library does not do have is implementation of address arithmetic in Java, which i assume you will possibly never need.
From their site:
With xFunction, you no longer need to implement those ugly native methods. Instead, you extend and instantiate xFunction classes to create conventional Java objects representing external functions, data structures, pointers, and callbacks. All necessary data conversions and external function calls are done seamlessly by the xFunction library:
import com.excelsior.xFunction.*;
...
/* Call Beep() from KERNEL32.DLL */
xFunction f =
new xFunction( "kernel32",
"int Beep(int,int)" );
f.invoke( new Argument(1770),
new Argument(100) );

Matlab / Java API callback

I'm designing an API (in Java) and expect to have users accessing the API from Matlab. The problem is that I want the API to provide a piece of functionality like:
javaApi.waitUntilPredicateIsTrue(Predicate<JavaObj> test);
My API (in the background) gets hold of instances of Java Obj (via some mechanism, e.g. polling). I want this API method to block until one of these instances, when passed to the Predicate evaluates to true. If I was calling this API from Java, I'd do:
javaApi.waitUntilPredicateIsTrue(new Predicate<JavaObj>() {
public boolean evaluate(JavaObj jo) {
return "READY".equals(jo.getState());
}
});
You get the idea.
How can this be called from within Matlab? Can I use anonymous inner classes from Matlab? Can I declare a Matlab classdef which extends the interface Predicate (can this cope with the Java generic version)?
That sounds like a tough question. I'm still running R2006b so this may have changed, but it looks like MATLAB will not translate function handles (incl. anonymous functions) and structures into Java objects. I don't know about MATLAB custom classes, since the syntax has changed. Strings, arrays, and cell arrays will translate properly. They don't comment at all on implementing interfaces. (:p :p :p BOO HISS)
edit: just found this page on Matlab Central, it talks about some undocumented interfaces.
Matlab has a much nicer solution than forcing users to create a whole class just to provide a single method. Take a look at their anonymous functions.
Note that anonymous functions in Matlab have odd scoping rules. Make sure you read the "Variables Used in the Expression" section of the linked help page. If you want more traditional lexical scoping, take a look at nested functions.
EDIT:
I am assuming that you will be doing the polling from Matlab, not passing the predicate function to Java. Example:
function waitForPredicate(pred)
while pred
end
end
waitForPredicate(#()javaApi.isMyConditionMet());

Categories

Resources