In Dex versions 037+, what is the difference in behavior between invoke-super mid and invoke-interface mid where mid is an interface method?
I am quoting the following from the Dalvik docs [1], but I am unable to make much sense out of it:
In Dex files version 037 or later, if the method_id refers to an
interface method, invoke-super is used to invoke the most specific,
non-overridden version of that method defined on that interface. The
same method restrictions hold as for invoke-virtual. In Dex files
prior to version 037, having an interface method_id is illegal and
undefined.
invoke-interface is used to invoke an interface method, that is, on an
object whose concrete class isn't known, using a method_id that refers
to an interface.
[1] https://source.android.com/devices/tech/dalvik/dalvik-bytecode
Dex version 37 adds support for default interface methods. So if you call invoke-super with something like Lmy/interface;->someMethod()V, it will search through the interface hierarchy to find the first interface that provides an implementation of that interface.
So, for a fuller example, let's say we have a few classes.
Lmy/superinterface;
Lmy/interface; which extends Lmy/superinterface;
Lmy/superclass; which implements Lmy/interface;
Lmy/subclass; which extends Lmy/superclass;
Let's say that Lmy/superinterface; defines a default implementation of someMethod()V, and both Lmy/superclass; and Lmy/subclass; provide their own implementation
If you have an object of Lmy/subclass; in v0, then
invoke-super {v0}, Lmy/interface;->someMethod()V invokes the implementation from Lmy/superinterface;
invoke-super {v0}, Lmy/subclass;->someMethod()V invokes the implementation from Lmy/superclass;
invoke-virtual {v0}, Lmy/subclass;->someMethod()V invokes the implementation from Lmy/subclass;
Related
In our framework we have an interface with this method in the public API:
JaxbConfiguration newJaxbConfiguration(Options xjcOpts);
In the implementation, we do something like this:
import com.sun.tools.xjc.ModelLoader;
import com.sun.tools.xjc.Options;
import com.sun.tools.xjc.model.Model;
...
public JaxbConfiguration newJaxbConfiguration(Options xjcOpts) {
Model model = ModelLoader.load(xjcOpts, ...);
...
}
However, both OSGi and Java 9's jigsaw don't like that we use com.sun.tools.xjc.Options, not in our implementation and especially not in our public API interface.
How can we get rid of it?
The JDeps website lists some of the JDK internal APIs and the recommended way to replace their usage. However, the use of ModelLoader.load() is not mentioned. My guess is that this use case has not come up enough to get the attention of the JDeps team.
My recommendation would be to refactor this method so that
you pass in the data you're using to construct the Options argument, instead of passing in the Options argument
use that data to construct your JaxbConfiguration object instead of converting from the internal Model.
You don't mention what JaxbConfiguration is or what library it's from so it's hard for me to say exactly how to construct it. Anyway, this answer is about how to remove the use of the internal API. How to construct a JaxbConfiguration is probably a different question.
It is a Sequence Diagram HowTo question, not a HowTo code.
I am using Visio 2010 and developing >> reverse engineering from Microsoft Dynamics AX 2012 / X++. Yes people its all about how to map static on UML.
My class is instantiated from FORM using at its void static main(). This calls another static method, say construct() which returns an instance of the same class.
I want to show the class (in static methods) and the resulting object separately some like the meta class runs (self msgs) and finally produces the class object which finally takes over. But how will a self msg call return a value ? How do I connect it with the resulting object of the class ? I hope I make enough sense to make you guys understand.
Note, the class is not a static class, but it has a static constructor.
If you want to depict a call to constructor (i.e. static operation that is responsible for creating an object and (usually) returning it as a reply) then you have to use a createMessage construct i.e. a dashed line with an open arrow and the word create on it. While this is not directly stated in specification, usually in such case the arrow points on the lifeline box (rectangle) rather than a line itself (however I've seen information that both notations are correct).
Note that in this case the logic of constructor is hidden (encapsulated) which is a good idea in general.
You can find more details in UML specification in section 17.4, especially 17.4.4.1 and an example in section 17.6.5 on Figure 17.14.
If you want to use a static operation other than constructor and call it without a use of class instance you have to model class as object (after all class is an object itself at least on analytical level). Note that the type of message can be either synchronous or asynchronous depending on your needs.
With this approach you can provide details on how the class handles this function (i.e. what other calls does it make).
For more details see "Applied UML and Patterns" by Craig Larman, section 15.4, Figure 15.20. Note however that Larman suggest a use of <<metaclass>> stereotype. Yet the called object is a class (metaclass is a class whose instance is class so this is not our case) so the stereotype should be <<class>>.
What I need: a class with two parents, which are ContextBoundObject and another class.
Why: I need to access the ContextBoundOject to log the method calls.
Composition works? As of now, no (types are not recognized, among other things).
Are other ways to do this? Yes, but not so automatable and without third-party components (maybe a T4 could do, but I'm no expert).
A more detailed explanation.
I need to extend System classes (some of which have already MarshalByRefObject (which is the parent of ContextBoundObject) for parent, for example ServiceBase and FileSystemWatcher, and some not, for example Exception and Timer) to access some inner workings of the framework, so I can log method calls (for now; in future it may change).
If I use this way I only have to add a class name to the object I want to log, instead of adding the logging calls to every method, but obviously I can't do this:
public class MyService:ServiceBase,ContextBoundObject,IDisposable{
public MyService(){}
public Dispose(){}
}
so one could try the usual solution, interfaces, but then if I call Run as in:
ServiceBase.Run(new MyService());
using a hypotethical interface IServiceBase it wouldn't work, because the type ServiceBase is not castable to IServiceBase -- it doesn't inherit from any interface. The problem is even worse with exceptions: throw only accepts a type descending from Exception.
The reverse, producing a IContextBoundObject interface, doesn't seem to work either: the logging mechanism doesn't work by methods, so I don't need to implement any, just an attribute and some small internal classes (and inheriting from ContextBoundObject, not even from MarshalByRefObject, which the metadata present as practically the same).
From what I see, extending from ContextBoundObject puts the extended class in a Proxy (probably because in this way the method calls use SyncProcessMessage(IMessage) and so can be intercepted and logged), maybe there's a way to do it without inheritance, or maybe there could be pre or post compiling techniques available for surrounding methods with logging calls (like T4 Text Templates), I don't know.
If someone wants to give this a look, I used a customized version of MSTestExtentions in my program to do the logging (of the method calls).
Any ideas are appreciated. There could be the need for more explanations, just ask.
Logging method calls is usually done using attributes to annotate classes or methods for which you want to have logging enabled. This is called Aspect Oriented Programming.
For this to work, you need a software that understands those attributes and post-processes your assembly by adding the necessary code to the methods / classes that have been annotated.
For C# there exists PostSharp. See here for an introduction.
Experimenting with proxies I found a way that apparently logs explicit calls.
Essentially I create a RealProxy like in example in the msdn, then obtain the TransparentProxy and use that as the normal object.
The logging is done in the Invoke method overridden in the customized RealProxy class.
static void Main(){
...
var ServiceClassProxy=new ServiceRealProxy(typeof(AServiceBaseClass),new object[]{/*args*/});
aServiceInstance=(AServiceBaseClass)ServiceClassProxy.GetTransparentProxy();
ServiceBase.Run(aServiceInstance);
...
}
In the proxy class the Invoke will be done like this:
class ServiceRealProxy:RealProxy{
...
[SecurityPermissionAttribute(SecurityAction.LinkDemand, Flags=SecurityPermissionFlag.Infrastructure)]
public override IMessage Invoke(IMessage myIMessage){
// remember to set the "__Uri" property you get in the constructor
...
/* logging before */
myReturnMessage = ChannelServices.SyncDispatchMessage(myIMessage);
/* logging after */
...
return myReturnMessage;
// it could be useful making a switch for all the derived types from IMessage; I see 18 of them, from
// System.Runtime.Remoting.Messaging.ConstructionCall
// ... to
// System.Runtime.Remoting.Messaging.TransitionCall
}
...
}
I have still to investigate extensively, but the logging happened. This isn't an answer to my original problem because I have still to test this on classes that don't inherit from MarshalByRefObject.
I am very new to concept of IOC and I understand the fact that they help us resolve different classes in different contexts. Your calling class will just interact with Interface and Interface with decide which implementation to give you and it takes care of newing up the object.
Please do correct me if I am understanding is wrong because my question is based on that:
Now, I see this pattern very often in these projects:
private readonly IEmailService emailService;
private readonly ITemplateRenderer templateRenderer;
private readonly IHtmlToTextTransformer htmlToTextTransformer;
public TemplateEmailService(IEmailService emailService,
ITemplateRenderer templateRenderer,
IHtmlToTextTransformer htmlToTextTransformer)
{
this.emailService = emailService;
this.htmlToTextTransformer = htmlToTextTransformer;
this.templateRenderer = templateRenderer;
}
I understand that this helps using all the implementations of these classes without newing them up and also you don't have to decide WHICH implementaion to get, your IOC decides it for you, right?
but when I code like this, I do not even touch any IOC congiguration files. And again I am usin git for 2 days only but from all the tutorials that I have read, I was expecting my self to configure something which says "Resolve IParent to Child" class. But it works without me doing anything like it. Is it because there is only one implementaion of these interfaces? and If I do have more than one implementations then and then only I will have to configure resolved explicitly?
The code sample you have is a case of Constructor Injection.
In a traditional code, you would have a parameterless constructor, and in it you would "new-up" your objects like this:
IEmailService emailService = new EmailService();
So your code is explictly controlling which implementation gets assigned to the interface variable.
In IoC using constructor injection, control is inverted, meaning the container is "driving the bus" and is creating your TemplateEmailService object. When it is about to create it, the container looks at your constructor parameters (IEmailService , ITemplateRenderer , etc.) and feeds those objects to your class for use.
The IoC container can be configured so that interface A gets fulfilled by implementation B (or C) explicitly. Each one has a way to do it. Or it could do it by convention (IFoo fulfilled by Foo), or even attributes in classes, whatever.
So to answer your question-- you can explicitly define which implementations get used to fulfill certain interfaces. Got to read the IoC container docs for how to.
One more thing - "when you code like this", you technically don't have to be using an IoC container. In fact, your class should not have a direct reference to the container - it will maximize the reusability, and also allow easy testing. So you would wire-up interfaces to implementation classes elsewhere.
Hypothetic and academic question.
pseudo-code:
<pre><code>
class Book{
read(theReader)
}
class BookWithMemory extends Book {
read(theReader, aTimestamp = null)
}
</pre></code>
Assuming:
an interface (if supported) would prohibit it
default value for parameters are supported
Notes:
PHP triggers an strict standards error for this.
I'm not surprised that PHP strict mode complains about such an override. It's very easy for a similar situation to arise unintentionally in which part of a class hierarchy was edited to use a new signature and a one or a few classes have fallen out of sync.
To avoid the ambiguity, name the new method something different (for this example, maybe readAt?), and override read to call readAt in the new class. This makes the intent plain to the interpreter as well as anyone reading the code.
The actual behavior in such a case is language-dependent -- more specifically, it depends on how much of the signature makes up the method selector, and how parameters are passed.
If the name alone is the selector (as in PHP or Perl), then it's down to how the language handles mismatched method parameter lists. If default arguments are processed at the call site based on the static type of the receiver instead of at the callee's entry point, when called through a base class reference you'd end up with an undefined argument value instead of your specified default, similarly to what would happen if there was no default specified.
If the number of parameters (with or without their types) are part of the method selector (as in Erlang or E), as is common in dynamic languages that run on JVM or CLR, you have two different methods. Create a new overload taking additional arguments, and override the base method with one that calls the new overload with default argument values.
If I am reading the question correctly, this question seems very language specific (as in it is not applicable to all dynamic languages), as I know you can do this in ruby.
class Book
def read(book)
puts book
end
end
class BookWithMemory < Book
def read(book,aTimeStamp = nil)
super book
puts aTimeStamp
end
end
I am not sure about dynamic languages besides ruby. This seems like a pretty subjective question as well, as at least two languages were designed on either side of the issue (method overloading vs not: ruby vs php).