Could anybody explain what is the bytecode libraries? For example, some Hibernate 3.5 book tells me that Hibernate uses javaassit or CGLib bytecode libraries. For what this libraries exists? Thank you.
When you write a Java app, you have to compile with a compiler (e.g. javac) first. Some classfiles (*.class) are generated. There is the bytecode. The class file is a binary file that can be executed by a Java Virtual Machine.
You may want to read a classfile, modify a classfile (eg. for instrumentation) or create (generate) a completely new classfile. If you want to do this, a good library can make it easier. You don't have to care about the exact structure and many constants (e.g. opcodes).
CGLib homepage says that Hibernate "Uses cglib to generate proxies for persistent classes."
Related
ALL,
I'm trying to figure out a proper link order when I build my software on Linux with gcc.
I have an .a library, which export an interface class (class with a lot of pure virtuals).
This a library is linked to libraryA.so and libraryB.so, because both those libraries have an actual implementation for the interface.
Now libraryB is also linked to unixODBC.
Now, both libraryA and libraryB are linked to libraryC, which actually instantiates the classes from them.
If I understand correctly the linking order should be as follows:
librarya.a libraryA odbc odbcinst libraryB
for libraryC.
But after successful build and running make install, trying to load libraryC fails, because libraryB is not visible.
Is my linking correct? Or better yet - is my understanding of the linking is correct - the referenced library should be first and then follows referencing libraries and I just need to update the LD_LIBRARY_PATH variable.
Thank you for clarifying.
I'm tasked with creating an SDK that can be consumed from both Android & Java applications using ReactiveX programming. I already have an android project using RxAndroid created, but now I need to extend it with RxJava2.
The question I'm facing is whether I should create 'regular' java class library and use it for both scenarios or create 2 separate packages (which would mean a lot of duplicate code + maintenance).
Is this even possible? And if so, is it a good practice?
whether I should create 'regular' java class library and use it for both scenarios
Yes. What I would do to start is simply change your Android library project to be a standard Java library and replace RxAndroid dependency by RxJava. Most code should still compile. Code which doesn't will mostly use schedulers provided by RxAndroid and can be changed to take Scheduler parameters.
Then create an Android Library project which depends on the Java Library and put the RxAndroid-specific code there.
As an addition to #AlexeyRomanov's answer, feel free to check out this library which could be used for both Android and Java projects: https://github.com/JakeWharton/RxRelay.
Its basically an extension to RxJava, but it might give you a solid idea where to go. Good luck!
i am new to openJDK8, i want to locate the file(s) where the bytecode instruction set is present which is used to generate the bytecode of a source .java file.
Secondly i need basic knowledge of the bytecode generation and execution process.
i am a research student and working on openJDK bytecode. Can you please guide?
The question seems to be conflating/mixing a number of different things. If you are creating a new bytecode there's several things you need to do.
Specify/define the bytecode
Modify the compiler to generate that bytecode
Modify the runtime to understand that bytecode
The Java bytecodes are specified in the Java Virtual Machine Specification. If you need to define a new one, you need to specify it to a similar degree, though you don't have to publish it there.
The part of Java SDK that is responsible from taking Java source code and producing Java bytecode is the Java compiler (javac). The source code for OpenJDK's javac is available here. The source code for Eclipse's Java Compiler is also available, but I don't know where.
Once you have the java compiler generating your custom bytecode, you have to teach the JVM how to interpret it.
The OpenJDK VM (hotspot) has several components that handle executing the bytecode. There's more than one interpreter (that reads and executes bytecode) and multiple Just-In-Time optimizing compilers that read the bytecode and (possibly) compile it into machine code before executing it. The source code for all of them is part of the hotspot. You probably just want to limit yourself to the interpreter and disable the compilers for initial work. If so, this set of notes explain how the hotspot interpreter works.
Prerelease versions of JDK 7 contained a class java.dyn.InvokeDynamic that allowed creating invokedynamic instructions from source code.
See here:
http://fwierzbicki.blogspot.com/2009/08/invokedynamic-and-jython-part-i.html
In the official JDK 7 release this class seems to have disappeared. Does anyone know if this source-code option is still supported?
java.dyn package has been renamed to java.lang.invoke.
Though I don't think you can create the 'invokedynamic' instruction from Java, the instruction is there for other dynamic languages.
In Java you can use 'java.lang.invoke.MethodHandle' as a faster alternative to reflection, examples are available in the javadoc of the MethodHandle class. Note that invokedynamic instruction itself relies on MethodHandle for dynamic linking of methods. (For more details you can read Oracle's article New JDK 7 Feature: Support for Dynamically Typed Languages in the Java Virtual Machine)
I want to compile jdk itself. I mean, I want to modify String.class in jdk to see created string objects in the system. Is there any way to modify classes in the jdk? When I try to modify source by modifying rt.jar, I got the error.
java.lang.NullPointerException
at java.util.Hashtable.put(Hashtable.java:394)
at java.lang.System.initProperties(Native Method)
at java.lang.System.initializeSystemClass(Unknown Source)
Probably there is a signature problem.
That doesn't look like a signature problem. It looks like you changed something that's causing Hashtable to dereference a null pointer. Review the change you made and see why it's doing this. Recall that Java keeps internal references to String constants in some conditions. I'm guessing you broke one of those.
Is there any way to modify classes in the jdk?
Well, you can download, modify and build the OpenJDK releases of Java 6 from source. There's information on how to do this on the OpenJDK site.
But beware that changes to low-level Java classes (such as String) can have effects that are hard to a non-expert to understand. And the consequence could well be a JVM that fails in a way that makes println or printStackTrace() inoperative.
There is maybe another way: you download java.lang.String original source, doing your modifications and compile only this class.
When you start your main programm be aware when loading classes: first load your String class then the java runtime classes. Refer to the java manual and the -Xbootclasspath/p options to do it in the proper order.
When installing a JDK you may choose to install the sources as well. Do so. Or download the sources separately. Then
expand the src.zip and get the String.java file.
create a new project containing the String.java in package java/lang.
change it accordingly to your needs.
just compile it.
put the class into the bootclass path of your JDK. See More infos on Bootclassbath.
run your app.
java -Xbootclasspath/p:<changed String classpath> -cp <regular classpath> <your application main class>
But changing the JDK might not be a good idea and you are not allowed to ship a changed JDK (at least up to 1.6) due license restrictions.
And yes, your problem is most likely somewhere else. Remember select isn't broken ;-)