changeset 2096:4ec3a2683655

8172183: Provide a javadoc description for jdk.dynalink module Reviewed-by: attila, jlaskey
author sundar
date Tue, 03 Jan 2017 22:14:41 +0530
parents 2a0437036a64
children d45caa3cd523
files README RELEASE_README src/jdk.dynalink/share/classes/jdk/dynalink/package-info.java src/jdk.dynalink/share/classes/module-info.java
diffstat 4 files changed, 202 insertions(+), 222 deletions(-) [+]
line wrap: on
line diff
--- a/README	Thu Dec 22 10:36:34 2016 -0800
+++ b/README	Tue Jan 03 22:14:41 2017 +0530
@@ -24,33 +24,33 @@
 
 You can clone Nashorn Mercurial forest using this command:
 
-    hg fclone http://hg.openjdk.java.net/nashorn/jdk8 nashorn~jdk8
+    hg fclone http://hg.openjdk.java.net/nashorn/jdk9 nashorn~jdk9
     
 To update your copy of the forest (fwith the latest code:
 
-    (cd nashorn~jdk8 ; hg fpull)
+    (cd nashorn~jdk9 ; hg fpull)
     
 Or just the nashorn subdirectory with
 
-    (cd nashorn~jdk8/nashorn ; hg pull -u)
+    (cd nashorn~jdk9/nashorn ; hg pull -u)
     
 To learn about Mercurial in detail, please visit http://hgbook.red-bean.com.
 
 - How to build?
 
-To build Nashorn, you need to install JDK 8. You may use the Nashorn
+To build Nashorn, you need to install JDK 9. You may use the Nashorn
 forest build (recommended) or down load from java.net.  You will need to
 set JAVA_HOME environmental variable to point to your JDK installation
 directory.
 
-    cd nashorn~jdk8/nashorn/make
+    cd nashorn~jdk9/nashorn/make
     ant clean; ant
 
 - How to run?
 
 Use the jjs script (see RELESE_README):
 
-    cd nashorn~jdk8/nashorn
+    cd nashorn~jdk9/nashorn
     sh bin/jjs <your .js file>
 
 Nashorn supports javax.script API. It is possible to drop nashorn.jar in
@@ -64,7 +64,7 @@
 Comprehensive development documentation is found in the Nashorn JavaDoc. You can
 build it using:
 
-    cd nashorn~jdk8/nashorn/make
+    cd nashorn~jdk9/nashorn/make
     ant javadoc
     
 after which you can view the generated documentation at dist/javadoc/index.html.
@@ -90,7 +90,7 @@
 test/script/external/test262 a symbolic link to that directory. After
 you've done this, you can run the ECMA-262 tests using:
 
-    cd nashorn~jdk8/nashorn/make
+    cd nashorn~jdk9/nashorn/make
     ant test262
 
 Ant target to get/update external test suites:
@@ -101,7 +101,7 @@
 These tests take time, so we have a parallelized runner for them that
 takes advantage of all processor cores on the computer:
 
-    cd nashorn~jdk8/nashorn/make
+    cd nashorn~jdk9/nashorn/make
     ant test262parallel
     
 - How to write your own test?
--- a/RELEASE_README	Thu Dec 22 10:36:34 2016 -0800
+++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
@@ -1,20 +0,0 @@
-The Nashorn repo is in the process of being migrated to OpenJDK and as such is
-incomplete in several areas.
-
-- The build system is not fully integrated.  When complete, Nashorn will be
-installed in its proper location in the JRE.
-
-- Once integrated, the correct version of the JDK will be wrapped around 
-Nashorn.  In the meantime, ensure you use JDK8 b68 or later.
-
-- The jjs tool has not been implemented in binary form yet.  Use "sh bin/jjs" 
-(or bin/jjs.bat on windows) in the interm.
-
-- The Dynalink component is not fully integrated into Nashorn as yet, but will
-be when details are finalized.
-
-- And, finally Nashorn is still in development.  To stay up to date, subscribe
-to nashorn-dev@openjdk.java.net at
-
-    http://mail.openjdk.java.net/mailman/listinfo/nashorn-dev.
-
--- a/src/jdk.dynalink/share/classes/jdk/dynalink/package-info.java	Thu Dec 22 10:36:34 2016 -0800
+++ b/src/jdk.dynalink/share/classes/jdk/dynalink/package-info.java	Tue Jan 03 22:14:41 2017 +0530
@@ -82,197 +82,6 @@
 */
 
 /**
- * <p>
- * Dynalink is a library for dynamic linking of high-level operations on objects.
- * These operations include "read a property",
- * "write a property", "invoke a function" and so on. Dynalink is primarily
- * useful for implementing programming languages where at least some expressions
- * have dynamic types (that is, types that can not be decided statically), and
- * the operations on dynamic types are expressed as
- * {@link java.lang.invoke.CallSite call sites}. These call sites will be
- * linked to appropriate target {@link java.lang.invoke.MethodHandle method handles}
- * at run time based on actual types of the values the expressions evaluated to.
- * These can change between invocations, necessitating relinking the call site
- * multiple times to accommodate new types; Dynalink handles all that and more.
- * <p>
- * Dynalink supports implementation of programming languages with object models
- * that differ (even radically) from the JVM's class-based model and have their
- * custom type conversions.
- * <p>
- * Dynalink is closely related to, and relies on, the {@link java.lang.invoke}
- * package.
- * <p>
- *
- * While {@link java.lang.invoke} provides a low level API for dynamic linking
- * of {@code invokedynamic} call sites, it does not provide a way to express
- * higher level operations on objects, nor methods that implement them. These
- * operations are the usual ones in object-oriented environments: property
- * access, access of elements of collections, invocation of methods and
- * constructors (potentially with multiple dispatch, e.g. link- and run-time
- * equivalents of Java overloaded method resolution). These are all functions
- * that are normally desired in a language on the JVM. If a language is
- * statically typed and its type system matches that of the JVM, it can
- * accomplish this with use of the usual invocation, field access, etc.
- * instructions (e.g. {@code invokevirtual}, {@code getfield}). However, if the
- * language is dynamic (hence, types of some expressions are not known until
- * evaluated at run time), or its object model or type system don't match
- * closely that of the JVM, then it should use {@code invokedynamic} call sites
- * instead and let Dynalink manage them.
- * <h2>Example</h2>
- * Dynalink is probably best explained by an example showing its use. Let's
- * suppose you have a program in a language where you don't have to declare the
- * type of an object and you want to access a property on it:
- * <pre>
- * var color = obj.color;
- * </pre>
- * If you generated a Java class to represent the above one-line program, its
- * bytecode would look something like this:
- * <pre>
- * aload 2 // load "obj" on stack
- * invokedynamic "GET:PROPERTY:color"(Object)Object // invoke property getter on object of unknown type
- * astore 3 // store the return value into local variable "color"
- * </pre>
- * In order to link the {@code invokedynamic} instruction, we need a bootstrap
- * method. A minimalist bootstrap method with Dynalink could look like this:
- * <pre>
- * import java.lang.invoke.*;
- * import jdk.dynalink.*;
- * import jdk.dynalink.support.*;
- *
- * class MyLanguageRuntime {
- *     private static final DynamicLinker dynamicLinker = new DynamicLinkerFactory().createLinker();
- *
- *     public static CallSite bootstrap(MethodHandles.Lookup lookup, String name, MethodType type) {
- *         return dynamicLinker.link(
- *             new SimpleRelinkableCallSite(
- *                 new CallSiteDescriptor(lookup, parseOperation(name), type)));
- *     }
- *
- *     private static Operation parseOperation(String name) {
- *         ...
- *     }
- * }
- * </pre>
- * There are several objects of significance in the above code snippet:
- * <ul>
- * <li>{@link jdk.dynalink.DynamicLinker} is the main object in Dynalink, it
- * coordinates the linking of call sites to method handles that implement the
- * operations named in them. It is configured and created using a
- * {@link jdk.dynalink.DynamicLinkerFactory}.</li>
- * <li>When the bootstrap method is invoked, it needs to create a
- * {@link java.lang.invoke.CallSite} object. In Dynalink, these call sites need
- * to additionally implement the {@link jdk.dynalink.RelinkableCallSite}
- * interface. "Relinkable" here alludes to the fact that if the call site
- * encounters objects of different types at run time, its target will be changed
- * to a method handle that can perform the operation on the newly encountered
- * type. {@link jdk.dynalink.support.SimpleRelinkableCallSite} and
- * {@link jdk.dynalink.support.ChainedCallSite} (not used in the above example)
- * are two implementations already provided by the library.</li>
- * <li>Dynalink uses {@link jdk.dynalink.CallSiteDescriptor} objects to
- * preserve the parameters to the bootstrap method: the lookup and the method type,
- * as it will need them whenever it needs to relink a call site.</li>
- * <li>Dynalink uses {@link jdk.dynalink.Operation} objects to express
- * dynamic operations. It does not prescribe how would you encode the operations
- * in your call site, though. That is why in the above example the
- * {@code parseOperation} function is left empty, and you would be expected to
- * provide the code to parse the string {@code "GET:PROPERTY:color"}
- * in the call site's name into a named property getter operation object as
- * {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY).named("color")}.
- * </ul>
- * <p>What can you already do with the above setup? {@code DynamicLinkerFactory}
- * by default creates a {@code DynamicLinker} that can link Java objects with the
- * usual Java semantics. If you have these three simple classes:
- * <pre>
- * public class A {
- *     public String color;
- *     public A(String color) { this.color = color; }
- * }
- *
- * public class B {
- *     private String color;
- *     public B(String color) { this.color = color; }
- *     public String getColor() { return color; }
- * }
- *
- * public class C {
- *     private int color;
- *     public C(int color) { this.color = color; }
- *     public int getColor() { return color; }
- * }
- * </pre>
- * and you somehow create their instances and pass them to your call site in your
- * programming language:
- * <pre>
- * for each(var obj in [new A("red"), new B("green"), new C(0x0000ff)]) {
- *     print(obj.color);
- * }
- * </pre>
- * then on first invocation, Dynalink will link the {@code .color} getter
- * operation to a field getter for {@code A.color}, on second invocation it will
- * relink it to {@code B.getColor()} returning a {@code String}, and finally on
- * third invocation it will relink it to {@code C.getColor()} returning an {@code int}.
- * The {@code SimpleRelinkableCallSite} we used above only remembers the linkage
- * for the last encountered type (it implements what is known as a <i>monomorphic
- * inline cache</i>). Another already provided implementation,
- * {@link jdk.dynalink.support.ChainedCallSite} will remember linkages for
- * several different types (it is a <i>polymorphic inline cache</i>) and is
- * probably a better choice in serious applications.
- * <h2>Dynalink and bytecode creation</h2>
- * {@code CallSite} objects are usually created as part of bootstrapping
- * {@code invokedynamic} instructions in bytecode. Hence, Dynalink is typically
- * used as part of language runtimes that compile programs into Java
- * {@code .class} bytecode format. Dynalink does not address the aspects of
- * either creating bytecode classes or loading them into the JVM. That said,
- * Dynalink can also be used without bytecode compilation (e.g. in language
- * interpreters) by creating {@code CallSite} objects explicitly and associating
- * them with representations of dynamic operations in the interpreted program
- * (e.g. a typical representation would be some node objects in a syntax tree).
- * <h2>Available operations</h2>
- * Dynalink defines several standard operations in its
- * {@link jdk.dynalink.StandardOperation} class. The linker for Java
- * objects can link all of these operations, and you are encouraged to at
- * minimum support and use these operations in your language too. The
- * standard operations {@code GET} and {@code SET} need to be combined with
- * at least one {@link jdk.dynalink.Namespace} to be useful, e.g. to express a
- * property getter, you'd use {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY)}.
- * Dynalink defines three standard namespaces in the {@link jdk.dynalink.StandardNamespace} class.
- * To associate a fixed name with an operation, you can use
- * {@link jdk.dynalink.NamedOperation} as in the previous example:
- * {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY).named("color")}
- * expresses a getter for the property named "color".
- * <h2>Operations on multiple namespaces</h2>
- * Some languages might not have separate namespaces on objects for
- * properties, elements, and methods, and a source language construct might
- * address several of them at once. Dynalink supports specifying multiple
- * {@link jdk.dynalink.Namespace} objects with {@link jdk.dynalink.NamespaceOperation}.
- * <h2>Language-specific linkers</h2>
- * Languages that define their own object model different than the JVM
- * class-based model and/or use their own type conversions will need to create
- * their own language-specific linkers. See the {@link jdk.dynalink.linker}
- * package and specifically the {@link jdk.dynalink.linker.GuardingDynamicLinker}
- * interface to get started.
- * <h2>Dynalink and Java objects</h2>
- * The {@code DynamicLinker} objects created by {@code DynamicLinkerFactory} by
- * default contain an internal instance of
- * {@code BeansLinker}, which is a language-specific linker
- * that implements the usual Java semantics for all of the above operations and
- * can link any Java object that no other language-specific linker has managed
- * to link. This way, all language runtimes have built-in interoperability with
- * ordinary Java objects. See {@link jdk.dynalink.beans.BeansLinker} for details
- * on how it links the various operations.
- * <h2>Cross-language interoperability</h2>
- * A {@code DynamicLinkerFactory} can be configured with a
- * {@link jdk.dynalink.DynamicLinkerFactory#setClassLoader(ClassLoader) class
- * loader}. It will try to instantiate all
- * {@link jdk.dynalink.linker.GuardingDynamicLinkerExporter} classes visible to
- * that class loader and compose the linkers they provide into the
- * {@code DynamicLinker} it creates. This allows for interoperability between
- * languages: if you have two language runtimes A and B deployed in your JVM and
- * they export their linkers through the above mechanism, language runtime A
- * will have a language-specific linker instance from B and vice versa inside
- * their {@code DynamicLinker} objects. This means that if an object from
- * language runtime B gets passed to code from language runtime A, the linker
- * from B will get a chance to link the call site in A when it encounters the
- * object from B.
+ * Contains interfaces and classes that are used to link an {@code invokedynamic} call site.
  */
 package jdk.dynalink;
--- a/src/jdk.dynalink/share/classes/module-info.java	Thu Dec 22 10:36:34 2016 -0800
+++ b/src/jdk.dynalink/share/classes/module-info.java	Tue Jan 03 22:14:41 2017 +0530
@@ -24,7 +24,198 @@
  */
 
 /**
- * Dynalink
+ * <p>
+ * Dynalink is a library for dynamic linking of high-level operations on objects.
+ * These operations include "read a property",
+ * "write a property", "invoke a function" and so on. Dynalink is primarily
+ * useful for implementing programming languages where at least some expressions
+ * have dynamic types (that is, types that can not be decided statically), and
+ * the operations on dynamic types are expressed as
+ * {@link java.lang.invoke.CallSite call sites}. These call sites will be
+ * linked to appropriate target {@link java.lang.invoke.MethodHandle method handles}
+ * at run time based on actual types of the values the expressions evaluated to.
+ * These can change between invocations, necessitating relinking the call site
+ * multiple times to accommodate new types; Dynalink handles all that and more.
+ * <p>
+ * Dynalink supports implementation of programming languages with object models
+ * that differ (even radically) from the JVM's class-based model and have their
+ * custom type conversions.
+ * <p>
+ * Dynalink is closely related to, and relies on, the {@link java.lang.invoke}
+ * package.
+ * <p>
+ *
+ * While {@link java.lang.invoke} provides a low level API for dynamic linking
+ * of {@code invokedynamic} call sites, it does not provide a way to express
+ * higher level operations on objects, nor methods that implement them. These
+ * operations are the usual ones in object-oriented environments: property
+ * access, access of elements of collections, invocation of methods and
+ * constructors (potentially with multiple dispatch, e.g. link- and run-time
+ * equivalents of Java overloaded method resolution). These are all functions
+ * that are normally desired in a language on the JVM. If a language is
+ * statically typed and its type system matches that of the JVM, it can
+ * accomplish this with use of the usual invocation, field access, etc.
+ * instructions (e.g. {@code invokevirtual}, {@code getfield}). However, if the
+ * language is dynamic (hence, types of some expressions are not known until
+ * evaluated at run time), or its object model or type system don't match
+ * closely that of the JVM, then it should use {@code invokedynamic} call sites
+ * instead and let Dynalink manage them.
+ * <h2>Example</h2>
+ * Dynalink is probably best explained by an example showing its use. Let's
+ * suppose you have a program in a language where you don't have to declare the
+ * type of an object and you want to access a property on it:
+ * <pre>
+ * var color = obj.color;
+ * </pre>
+ * If you generated a Java class to represent the above one-line program, its
+ * bytecode would look something like this:
+ * <pre>
+ * aload 2 // load "obj" on stack
+ * invokedynamic "GET:PROPERTY:color"(Object)Object // invoke property getter on object of unknown type
+ * astore 3 // store the return value into local variable "color"
+ * </pre>
+ * In order to link the {@code invokedynamic} instruction, we need a bootstrap
+ * method. A minimalist bootstrap method with Dynalink could look like this:
+ * <pre>
+ * import java.lang.invoke.*;
+ * import jdk.dynalink.*;
+ * import jdk.dynalink.support.*;
+ *
+ * class MyLanguageRuntime {
+ *     private static final DynamicLinker dynamicLinker = new DynamicLinkerFactory().createLinker();
+ *
+ *     public static CallSite bootstrap(MethodHandles.Lookup lookup, String name, MethodType type) {
+ *         return dynamicLinker.link(
+ *             new SimpleRelinkableCallSite(
+ *                 new CallSiteDescriptor(lookup, parseOperation(name), type)));
+ *     }
+ *
+ *     private static Operation parseOperation(String name) {
+ *         ...
+ *     }
+ * }
+ * </pre>
+ * There are several objects of significance in the above code snippet:
+ * <ul>
+ * <li>{@link jdk.dynalink.DynamicLinker} is the main object in Dynalink, it
+ * coordinates the linking of call sites to method handles that implement the
+ * operations named in them. It is configured and created using a
+ * {@link jdk.dynalink.DynamicLinkerFactory}.</li>
+ * <li>When the bootstrap method is invoked, it needs to create a
+ * {@link java.lang.invoke.CallSite} object. In Dynalink, these call sites need
+ * to additionally implement the {@link jdk.dynalink.RelinkableCallSite}
+ * interface. "Relinkable" here alludes to the fact that if the call site
+ * encounters objects of different types at run time, its target will be changed
+ * to a method handle that can perform the operation on the newly encountered
+ * type. {@link jdk.dynalink.support.SimpleRelinkableCallSite} and
+ * {@link jdk.dynalink.support.ChainedCallSite} (not used in the above example)
+ * are two implementations already provided by the library.</li>
+ * <li>Dynalink uses {@link jdk.dynalink.CallSiteDescriptor} objects to
+ * preserve the parameters to the bootstrap method: the lookup and the method type,
+ * as it will need them whenever it needs to relink a call site.</li>
+ * <li>Dynalink uses {@link jdk.dynalink.Operation} objects to express
+ * dynamic operations. It does not prescribe how would you encode the operations
+ * in your call site, though. That is why in the above example the
+ * {@code parseOperation} function is left empty, and you would be expected to
+ * provide the code to parse the string {@code "GET:PROPERTY:color"}
+ * in the call site's name into a named property getter operation object as
+ * {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY).named("color")}.
+ * </ul>
+ * <p>What can you already do with the above setup? {@code DynamicLinkerFactory}
+ * by default creates a {@code DynamicLinker} that can link Java objects with the
+ * usual Java semantics. If you have these three simple classes:
+ * <pre>
+ * public class A {
+ *     public String color;
+ *     public A(String color) { this.color = color; }
+ * }
+ *
+ * public class B {
+ *     private String color;
+ *     public B(String color) { this.color = color; }
+ *     public String getColor() { return color; }
+ * }
+ *
+ * public class C {
+ *     private int color;
+ *     public C(int color) { this.color = color; }
+ *     public int getColor() { return color; }
+ * }
+ * </pre>
+ * and you somehow create their instances and pass them to your call site in your
+ * programming language:
+ * <pre>
+ * for each(var obj in [new A("red"), new B("green"), new C(0x0000ff)]) {
+ *     print(obj.color);
+ * }
+ * </pre>
+ * then on first invocation, Dynalink will link the {@code .color} getter
+ * operation to a field getter for {@code A.color}, on second invocation it will
+ * relink it to {@code B.getColor()} returning a {@code String}, and finally on
+ * third invocation it will relink it to {@code C.getColor()} returning an {@code int}.
+ * The {@code SimpleRelinkableCallSite} we used above only remembers the linkage
+ * for the last encountered type (it implements what is known as a <i>monomorphic
+ * inline cache</i>). Another already provided implementation,
+ * {@link jdk.dynalink.support.ChainedCallSite} will remember linkages for
+ * several different types (it is a <i>polymorphic inline cache</i>) and is
+ * probably a better choice in serious applications.
+ * <h2>Dynalink and bytecode creation</h2>
+ * {@code CallSite} objects are usually created as part of bootstrapping
+ * {@code invokedynamic} instructions in bytecode. Hence, Dynalink is typically
+ * used as part of language runtimes that compile programs into Java
+ * {@code .class} bytecode format. Dynalink does not address the aspects of
+ * either creating bytecode classes or loading them into the JVM. That said,
+ * Dynalink can also be used without bytecode compilation (e.g. in language
+ * interpreters) by creating {@code CallSite} objects explicitly and associating
+ * them with representations of dynamic operations in the interpreted program
+ * (e.g. a typical representation would be some node objects in a syntax tree).
+ * <h2>Available operations</h2>
+ * Dynalink defines several standard operations in its
+ * {@link jdk.dynalink.StandardOperation} class. The linker for Java
+ * objects can link all of these operations, and you are encouraged to at
+ * minimum support and use these operations in your language too. The
+ * standard operations {@code GET} and {@code SET} need to be combined with
+ * at least one {@link jdk.dynalink.Namespace} to be useful, e.g. to express a
+ * property getter, you'd use {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY)}.
+ * Dynalink defines three standard namespaces in the {@link jdk.dynalink.StandardNamespace} class.
+ * To associate a fixed name with an operation, you can use
+ * {@link jdk.dynalink.NamedOperation} as in the previous example:
+ * {@code StandardOperation.GET.withNamespace(StandardNamespace.PROPERTY).named("color")}
+ * expresses a getter for the property named "color".
+ * <h2>Operations on multiple namespaces</h2>
+ * Some languages might not have separate namespaces on objects for
+ * properties, elements, and methods, and a source language construct might
+ * address several of them at once. Dynalink supports specifying multiple
+ * {@link jdk.dynalink.Namespace} objects with {@link jdk.dynalink.NamespaceOperation}.
+ * <h2>Language-specific linkers</h2>
+ * Languages that define their own object model different than the JVM
+ * class-based model and/or use their own type conversions will need to create
+ * their own language-specific linkers. See the {@link jdk.dynalink.linker}
+ * package and specifically the {@link jdk.dynalink.linker.GuardingDynamicLinker}
+ * interface to get started.
+ * <h2>Dynalink and Java objects</h2>
+ * The {@code DynamicLinker} objects created by {@code DynamicLinkerFactory} by
+ * default contain an internal instance of
+ * {@code BeansLinker}, which is a language-specific linker
+ * that implements the usual Java semantics for all of the above operations and
+ * can link any Java object that no other language-specific linker has managed
+ * to link. This way, all language runtimes have built-in interoperability with
+ * ordinary Java objects. See {@link jdk.dynalink.beans.BeansLinker} for details
+ * on how it links the various operations.
+ * <h2>Cross-language interoperability</h2>
+ * A {@code DynamicLinkerFactory} can be configured with a
+ * {@link jdk.dynalink.DynamicLinkerFactory#setClassLoader(ClassLoader) class
+ * loader}. It will try to instantiate all
+ * {@link jdk.dynalink.linker.GuardingDynamicLinkerExporter} classes visible to
+ * that class loader and compose the linkers they provide into the
+ * {@code DynamicLinker} it creates. This allows for interoperability between
+ * languages: if you have two language runtimes A and B deployed in your JVM and
+ * they export their linkers through the above mechanism, language runtime A
+ * will have a language-specific linker instance from B and vice versa inside
+ * their {@code DynamicLinker} objects. This means that if an object from
+ * language runtime B gets passed to code from language runtime A, the linker
+ * from B will get a chance to link the call site in A when it encounters the
+ * object from B.
  */
 module jdk.dynalink {
     requires java.logging;