Table of Contents
Programs are organized as sets of packages. The members of a package (§7.1) are class and interface types, which are declared in compilation units of the package, and subpackages, which may contain compilation units and subpackages of their own.
Each package has its own set of names for types, which helps to prevent name conflicts. The naming structure for packages is hierarchical.
If a set of packages is sufficiently cohesive, then the packages may be grouped into a module. A module categorizes some or all of its packages as exported, which means their types may be accessed from code outside the module. If a package is not exported by a module, then only code inside the module may access its types. Furthermore, if code in a module wishes to access the packages exported by another module, then the first module must explicitly depend on the second module. Thus, a module controls how its packages use other modules (by specifying dependences) and controls how other modules use its packages (by specifying which of its packages are exported).
Modules and packages may be stored in a file system or in a database (§7.2). Modules and packages that are stored in a file system may have certain constraints on the organization of their compilation units to allow a simple implementation to find module and type declarations easily.
Code in a
compilation unit automatically has access to all types declared in its
package and also automatically imports all of the public types
declared in the predefined package java.lang.
A top level type is accessible (§6.6) outside the
package that declares it only if the type is declared public. A top
level type is accessible outside the module that declares it only if
the type is declared public and is a member of an exported package. A
type that is declared public but is not a member of an exported
package is accessible only to code inside the module.
For small programs and casual development, a package can be unnamed (§7.4.2) or have a simple name, but if code is to be widely distributed, unique package names should be chosen using qualified names. This can prevent the conflicts that would otherwise occur if two development groups happened to pick the same package name and these packages were later to be used in a single program.
The members of a package are its subpackages and all the top level class types (§7.6, §8 (Classes)) and top level interface types (§9 (Interfaces)) declared in all the compilation units (§7.3) of the package.
For example, in the Java SE Platform API:
The package java has
subpackages awt, applet,
io, lang, net,
and util, but no compilation units.
The package java.awt has a
subpackage named image, as well as a number
of compilation units containing declarations of class and
interface types.
If the fully qualified name (§6.7) of a package
is P, and Q is a subpackage
of P, then P.Q is the fully
qualified name of the subpackage, and furthermore denotes a
package.
A package may not contain two members of the same name, or a compile-time error results.
Here are some examples:
Because the package java.awt
has a subpackage image, it cannot (and does
not) contain a declaration of a class or interface type
named image.
If there is a package
named mouse and a member
type Button in that package (which then might
be referred to as mouse.Button), then there
cannot be any package with the fully qualified
name mouse.Button
or mouse.Button.Click.
If com.nighthacks.java.jag is
the fully qualified name of a type, then there cannot be any
package whose fully qualified name is
either com.nighthacks.java.jag
or com.nighthacks.java.jag.scrabble.
It is however possible for members of different packages to have the same simple name. For example, it is possible to declare a package:
package vector;
public class Vector { Object[] vec; }
that has as a member a public class
named Vector, even though the package java.util
also declares a class named Vector. These two class
types are different, reflected by the fact that they have different
fully qualified names (§6.7). The fully qualified
name of this example Vector
is vector.Vector,
whereas java.util.Vector is the fully qualified
name of the Vector class included in the
Java SE Platform. Because the package vector contains a
class named Vector, it cannot also have a
subpackage named Vector.
The hierarchical naming structure for packages is intended to be convenient for organizing related packages in a conventional manner, but has no significance in itself other than the prohibition against a package having a subpackage with the same simple name as a top level type (§7.6) declared in that package.
For example, there is no special access relationship
between a package named oliver and another package
named oliver.twist, or between packages
named evelyn.wood
and evelyn.waugh. That is, the code in a package
named oliver.twist has no better access to the
types declared within package oliver than code in
any other package.
Each host system determines how modules, packages, and compilation units are created and stored.
Each host system determines which compilation units are observable in a particular compilation (§7.3). Each host system also determines which observable compilation units are associated with a module. The observability of compilation units associated with a module determines which modules are observable (§7.7.3) and which packages are visible within those modules (§7.4.3).
The host system is free to determine that a compilation unit which
contains a module declaration is not, in fact, observable, and thus is
not associated with the module declared therein. This enables a
compiler to choose which directory on
a modulesourcepath is "really" the embodiment of a
given module. However, if the host system determines that a
compilation unit which contains a module declaration
is observable, then §7.4.3
mandates that the compilation unit must be associated with the module
declared therein, and not with any other module.
The host system is free to determine that a compilation unit which contains a type declaration is (first) observable and (second) associated with an unnamed module or an automatic module - despite no declaration of an unnamed or automatic module existing in any compilation unit, observable or otherwise.
In simple implementations of the Java SE Platform, packages and compilation units may be stored in a local file system. Other implementations may store them using a distributed file system or some form of database.
If a host system stores packages and compilation units in a database, then the database must not impose the optional restrictions (§7.6) on compilation units permissible in file-based implementations.
For example, a system that uses a database to store packages may not enforce a maximum of one public class or interface per compilation unit.
Systems that use a database must, however, provide an option to convert a program to a form that obeys the restrictions, for purposes of export to file-based implementations.
As an extremely simple example of storing packages in a file system, all the packages and source and binary code in a project might be stored in a single directory and its subdirectories. Each immediate subdirectory of this directory would represent a top level package, that is, one whose fully qualified name consists of a single simple name. Each further level of subdirectory would represent a subpackage of the package represented by the containing directory, and so on.
The directory might contain the following immediate subdirectories:
com gls jag java wnj
where directory java would
contain the Java SE Platform packages; the
directories jag, gls,
and wnj might contain packages that three of the
authors of this specification created for their personal use and to
share with each other within this small group; and the
directory com would contain packages procured from
companies that used the conventions described in
§6.1 to generate unique names for their
packages.
Continuing the example, the directory
java would contain, among others, the following
subdirectories:
applet awt io lang net util
corresponding to the
packages java.applet, java.awt,
java.io, java.lang, java.net, and java.util that
are defined as part of the Java SE Platform API.
Still continuing the example, if we were to look
inside the directory util, we might see the
following files:
BitSet.java Observable.java BitSet.class Observable.class Date.java Observer.java Date.class Observer.class ...
where each of the .java files
contains the source for a compilation unit (§7.3)
that contains the definition of a class or interface whose binary
compiled form is contained in the
corresponding .class file.
Under this simple organization of packages, an implementation of the Java SE Platform would transform a package name into a pathname by concatenating the components of the package name, placing a file name separator (directory indicator) between adjacent components.
For example, if this simple organization were used
on an operating system where the file name separator
is /, the package name:
jag.scrabble.board
would be transformed into the directory name:
jag/scrabble/board
A package name component or class name might contain
a character that cannot correctly appear in a host file system's
ordinary directory name, such as a Unicode character on a system that
allows only ASCII characters in file names. As a convention, the
character can be escaped by using, say, the @
character followed by four hexadecimal digits giving the numeric value
of the character, as in
the \uxxxx escape
(§3.3).
Under this convention, the package name:
children.activities.crafts.papierM\u00e2ch\u00e9
which can also be written using full Unicode as:
children.activities.crafts.papierMâché
might be mapped to the directory name:
children/activities/crafts/papierM@00e2ch@00e9
If the @ character is not a valid
character in a file name for some given host file system, then some
other character that is not valid in a identifier could be used
instead.
CompilationUnit is the goal symbol (§2.1) for the syntactic grammar (§2.3) of Java programs. It is defined by the following production:
An ordinary compilation unit consists of three parts, each of which is optional:
A package declaration (§7.4), giving the
fully qualified name (§6.7) of the package
to which the compilation unit belongs.
A compilation unit that has no package declaration is part of
an unnamed package (§7.4.2).
import declarations (§7.5) that allow
types from other packages and static members of types to be
referred to using their simple names.
Top level type declarations (§7.6) of class and interface types.
A modular compilation unit consists of a module
declaration (§7.7), optionally preceded by
import declarations. The import declarations allow types from
packages in this module and other modules, as well as static members
of types, to be referred to using their simple names within the
module declaration.
Every compilation unit implicitly imports every public type name
declared in the predefined package java.lang, as if the
declaration import java.lang.*; appeared at the
beginning of each compilation unit immediately after any package
declaration. As a result, the names of all those types are available as
simple names in every compilation unit.
The host system determines which compilation units are
observable, except for the compilation units
in the predefined package java and its
subpackages lang and io,
which are all always observable.
Each observable compilation unit may be associated with a module, as follows:
The host system may determine that an observable ordinary
compilation unit is associated with a module chosen by the host
system, except for (i) the ordinary compilation units in the
predefined package java and its subpackages
lang and io, which are all
associated with the java.base module, and (ii) any ordinary
compilation unit in an unnamed package, which is associated with
a module as specified in §7.4.2.
The host system must determine that an observable modular compilation unit is associated with the module declared by the modular compilation unit.
The observability of a compilation unit influences the observability of its package (§7.4.3), while the association of an observable compilation unit with a module influences the observability of that module (§7.7.6).
When compiling the modular and ordinary compilation units associated
with a module M, the host system must respect the dependences
specified in M's declaration. Specifically, the host system must
limit the ordinary compilation units that would otherwise be
observable, to only those that are visible to
M. The ordinary compilation units that are visible to M
are the observable ordinary compilation units associated with the
modules that are read by M. The modules read by
M are given by the result of resolution, as
described in the java.lang.module package specification, with M as
the only root module. The host system must perform resolution to
determine the modules read by M; it is a compile-time error if
resolution fails for any of the reasons described in the
java.lang.module package specification.
The readability relation is reflexive, so M reads itself, and thus all
of the modular and ordinary compilation units associated with M are
visible to M.
The modules read by M drive the packages that are uniquely visible
to M (§7.4.3), which in turn drives both the
top level packages in scope and the meaning of package names for code
in the modular and ordinary compilation units associated with M
(§6.3, §6.5.3,
§6.5.5).
The rules above ensure that package/type names used in annotations in a modular compilation unit (in particular, annotations applied to the module declaration) are interpreted as if they appeared in an ordinary compilation unit associated with the module.
Types declared in different ordinary compilation units can refer to each other, circularly. A Java compiler must arrange to compile all such types at the same time.
A package declaration appears within
an ordinary compilation unit to indicate the package to which the
compilation unit belongs.
A package declaration in an ordinary compilation unit specifies the name (§6.2) of the package to which the compilation unit belongs.
The package name mentioned in a package declaration must be the
fully qualified name of the package (§6.7).
The scope and shadowing of a package declaration is specified in §6.3 and §6.4.
The rules for annotation modifiers on a package declaration are specified in §9.7.4 and §9.7.5.
At most one annotated package declaration is permitted for a given
package.
The manner in which this restriction is enforced
must, of necessity, vary from implementation to implementation. The
following scheme is strongly recommended for file-system-based
implementations: The sole annotated package declaration, if it
exists, is placed in a source file called
package-info.java in the directory
containing the source files for the package. This file does not
contain the source for a class called package-info;
indeed it would be illegal for it to do so, as
package-info is not a legal identifier.
Typically package-info.java contains only a
package declaration, preceded immediately by the annotations on the
package. While the file could technically contain the source code for
one or more classes with package access, it would be very bad
form.
It is recommended that package-info.java,
if it is present, take the place of package.html
for javadoc and other similar documentation
generation systems. If this file is present, the documentation
generation tool should look for the package documentation comment
immediately preceding the (possibly annotated) package declaration
in package-info.java. In this way,
package-info.java becomes the sole repository
for package-level annotations and documentation. If, in future, it
becomes desirable to add any other package-level information, this
file should prove a convenient home for this information.
An ordinary compilation unit that has no package declaration, but
has at least one other kind of declaration, is part of
an unnamed package.
Unnamed packages are provided by the Java SE Platform principally for convenience when developing small or temporary applications or when just beginning development.
An unnamed package cannot have subpackages, since the syntax of a
package declaration always includes a reference to a named top level
package.
An implementation of the Java SE Platform must support at least one unnamed package. An implementation may support more than one unnamed package, but is not required to do so. Which ordinary compilation units are in each unnamed package is determined by the host system.
The host system must associate ordinary compilation units in an unnamed package with an unnamed module (§7.7.5), not a named module.
Example 7.4.2-1. Unnamed Package
The compilation unit:
class FirstCall {
public static void main(String[] args) {
System.out.println("Mr. Watson, come here. "
+ "I want you.");
}
}
defines a very simple compilation unit as part of an unnamed package.
In implementations of the Java SE Platform that use a hierarchical file system for storing packages, one typical strategy is to associate an unnamed package with each directory; only one unnamed package is observable at a time, namely the one that is associated with the "current working directory". The precise meaning of "current working directory" depends on the host system.
A package is observable if and only if at least one of the following is true:
An ordinary compilation unit containing a declaration of the package is observable (§7.3).
The packages java, java.lang, and java.io are
always observable.
One can conclude this from the rule above and from
the rules of observable compilation units, as follows. The predefined
package java.lang declares the class Object, so the compilation
unit for Object is always observable (§7.3).
Hence, the java.lang package is observable, and the
java package also. Furthermore, since Object is
observable, the array type Object[] implicitly exists. Its
superinterface java.io.Serializable (§10.1) also
exists, hence the java.io package is observable.
A package is visible to a module M if and only if
an ordinary compilation unit containing a declaration of the
package is visible to M.
Package visibility is meant to imply that a package is observable in a useful way to a given module. It is generally not useful to know that package P is observable merely because a subpackage P.Q is observable. For example, suppose P.Q is observable (in module M1) and P.R is observable (in module M2); then, P is observable, but where? In M1, or M2, or both? The question is redundant; during compilation of module N that requires only M1, it matters that P.Q is observable, but it does not matter that P is observable.
A package is uniquely visible to a module M
if and only if one of the following holds:
An ordinary compilation unit associated with M contains a
declaration of the package, and M does not read any other
module that exports the package to M.
No ordinary compilation unit associated with M contains a
declaration of the package, and M reads exactly one other
module that exports the package to M.
An import declaration allows a named type or a
static member to be referred to by a simple name
(§6.2) that consists of a single identifier.
Without the use of an appropriate import declaration, the only way to
refer to a type declared in another package, or a static member of
another type, is to use a fully qualified name (§6.7).
A single-type-import declaration (§7.5.1) imports a single named type, by mentioning its canonical name (§6.7).
A type-import-on-demand declaration (§7.5.2) imports all the accessible types of a named type or named package as needed, by mentioning the canonical name of a type or package.
A single-static-import declaration (§7.5.3)
imports all accessible static members with a given name from a
type, by giving its canonical name.
A static-import-on-demand declaration (§7.5.4)
imports all accessible static members of a named type as
needed, by mentioning the canonical name of a type.
The scope and shadowing of a type or member imported by these declarations is specified in §6.3 and §6.4.
An import declaration makes types or members available by their
simple names only within the compilation unit that actually contains
the import declaration. The scope of the type(s) or member(s)
introduced by an import declaration specifically does not include
other compilation units in the same package, other import
declarations in the current compilation unit, or a package
declaration in the current compilation unit (except for the
annotations of a package declaration).
A single-type-import declaration imports a single type by giving its canonical name, making it available under a simple name in the module, class, and interface declarations of the compilation unit in which the single-type-import declaration appears.
The TypeName must be the canonical name of a class type, interface type, enum type, or annotation type (§6.7).
The type must be either a member of a named package, or a member of a type whose outermost lexically enclosing type declaration (§8.1.3) is a member of a named package, or a compile-time error occurs.
It is a compile-time error if the named type is not accessible (§6.6).
If two single-type-import declarations in the same compilation unit attempt to import types with the same simple name, then a compile-time error occurs, unless the two types are the same type, in which case the duplicate declaration is ignored.
If the type imported by the single-type-import declaration is declared
in the compilation unit that contains the import declaration, the
import declaration is ignored.
If a single-type-import declaration imports a type whose simple name is n, and the compilation unit also declares a top level type (§7.6) whose simple name is n, a compile-time error occurs.
If a compilation unit contains both a single-type-import declaration that imports a type whose simple name is n, and a single-static-import declaration (§7.5.3) that imports a type whose simple name is n, a compile-time error occurs, unless the two types are the same type, in which case the duplicate declaration is ignored.
Example 7.5.1-1. Single-Type-Import
import java.util.Vector;
causes the simple name Vector to
be available within the class and interface declarations in a
compilation unit. Thus, the simple name Vector
refers to the type declaration Vector in the
package java.util in all places where it is not shadowed
(§6.4.1) or obscured
(§6.4.2) by a declaration of a field, parameter,
local variable, or nested type declaration with the same name.
Note that the actual declaration
of java.util.Vector is generic
(§8.1.2). Once imported, the
name Vector can be used without qualification in a
parameterized type such as Vector<String>, or
as the raw type Vector. A related limitation of the
import declaration is that a nested type declared inside a generic
type declaration can be imported, but its outer type is always
erased.
Example 7.5.1-2. Duplicate Type Declarations
This program:
import java.util.Vector;
class Vector { Object[] vec; }
causes a compile-time error because of the duplicate
declaration of Vector, as does:
import java.util.Vector; import myVector.Vector;
where myVector is a package
containing the compilation unit:
package myVector;
public class Vector { Object[] vec; }
Example 7.5.1-3. No Import of a Subpackage
Note that an import declaration cannot import a
subpackage, only a type.
For example, it does not work to try to import
java.util and then use the name util.Random to
refer to the type java.util.Random:
import java.util;
class Test { util.Random generator; }
// incorrect: compile-time error
Example 7.5.1-4. Importing a Type Name that is also a Package Name
Package names and type names are usually different
under the naming conventions described in
§6.1. Nevertheless, in a contrived example where
there is an unconventionally-named package Vector,
which declares a public class whose name
is Mosquito:
package Vector;
public class Mosquito { int capacity; }
and then the compilation unit:
package strange;
import java.util.Vector;
import Vector.Mosquito;
class Test {
public static void main(String[] args) {
System.out.println(new Vector().getClass());
System.out.println(new Mosquito().getClass());
}
}
the single-type-import declaration importing
class Vector from package java.util does not
prevent the package name Vector from appearing and
being correctly recognized in subsequent import declarations. The
example compiles and produces the output:
class java.util.Vector class Vector.Mosquito
A type-import-on-demand declaration allows all accessible types of a named package or type to be imported as needed.
The PackageOrTypeName must be the canonical name (§6.7) of a package, a class type, an interface type, an enum type, or an annotation type.
If the PackageOrTypeName denotes a type (§6.5.4), then the type must be either a member of a named package, or a member of a type whose outermost lexically enclosing type declaration (§8.1.3) is a member of a named package, or a compile-time error occurs.
It is a compile-time error if the named package is not uniquely visible to the current module (§7.4.3), or if the named type is not accessible (§6.6).
It is not a compile-time error to name either java.lang or the named
package of the current compilation unit in a type-import-on-demand
declaration. The type-import-on-demand declaration is ignored in such
cases.
Two or more type-import-on-demand declarations in the same compilation unit may name the same type or package. All but one of these declarations are considered redundant; the effect is as if that type was imported only once.
If a compilation unit contains both a type-import-on-demand
declaration and a static-import-on-demand declaration
(§7.5.4) that name the same type, the effect is
as if the static member types of that type
(§8.5, §9.5) were imported
only once.
Example 7.5.2-1. Type-Import-on-Demand
import java.util.*;
causes the simple names of all public types
declared in the package java.util to be available within the class
and interface declarations of the compilation unit. Thus, the simple
name Vector refers to the
type Vector in the package java.util in all places
in the compilation unit where that type declaration is not shadowed
(§6.4.1) or obscured
(§6.4.2).
The declaration might be shadowed by a
single-type-import declaration of a type whose simple name
is Vector; by a type
named Vector and declared in the package to which
the compilation unit belongs; or any nested classes or
interfaces.
The declaration might be obscured by a declaration
of a field, parameter, or local variable
named Vector.
(It would be unusual for any of these conditions to occur.)
A single-static-import declaration imports all
accessible static members with a given simple name from a type. This
makes these static members available under their simple name in the
module, class, and interface declarations of the
compilation unit in which the single-static-import declaration
appears.
The TypeName must be the canonical name (§6.7) of a class type, interface type, enum type, or annotation type.
The type must be either a member of a named package, or a member of a type whose outermost lexically enclosing type declaration (§8.1.3) is a member of a named package, or a compile-time error occurs.
It is a compile-time error if the named type is not accessible (§6.6).
The Identifier must name at least one static member of the named
type. It is a compile-time error if there is
no static member of that name, or if all of the
named members are not accessible.
It is permissible for one single-static-import declaration to import several fields or types with the same name, or several methods with the same name and signature. This occurs when the named type inherits multiple fields, member types, or methods, all with the same name, from its own supertypes.
If two single-static-import declarations in the same compilation unit attempt to import types with the same simple name, then a compile-time error occurs, unless the two types are the same type, in which case the duplicate declaration is ignored.
If a single-static-import declaration imports a type whose simple name is n, and the compilation unit also declares a top level type (§7.6) whose simple name is n, a compile-time error occurs.
If a compilation unit contains both a single-static-import declaration that imports a type whose simple name is n, and a single-type-import declaration (§7.5.1) that imports a type whose simple name is n, a compile-time error occurs, unless the two types are the same type, in which case the duplicate declaration is ignored.
A static-import-on-demand declaration allows all
accessible static members of a named type to be
imported as needed.
The TypeName must be the canonical name (§6.7) of a class type, interface type, enum type, or annotation type.
The type must be either a member of a named package, or a member of a type whose outermost lexically enclosing type declaration (§8.1.3) is a member of a named package, or a compile-time error occurs.
It is a compile-time error if the named type is not accessible (§6.6).
Two or more static-import-on-demand declarations in the same compilation unit may name the same type; the effect is as if there was exactly one such declaration.
Two or more static-import-on-demand declarations in the same compilation unit may name the same member; the effect is as if the member was imported exactly once.
It is permissible for one static-import-on-demand declaration to import several fields or types with the same name, or several methods with the same name and signature. This occurs when the named type inherits multiple fields, member types, or methods, all with the same name, from its own supertypes.
If a compilation unit contains both a static-import-on-demand
declaration and a type-import-on-demand declaration
(§7.5.2) that name the same type, the effect is
as if the static member types of that type
(§8.5, §9.5) were imported
only once.
A top level type declaration declares a top level class type (§8 (Classes)) or a top level interface type (§9 (Interfaces)).
Extra ";" tokens appearing at the level of type
declarations in a compilation unit have no effect on the meaning of
the compilation unit. Stray semicolons are permitted in the Java programming language
solely as a concession to C++ programmers who are used to placing
";" after a class declaration. They should not be used in new
Java code.
In the absence of an access modifier, a top level type has package
access: it is accessible only within ordinary
compilation units of the package in which it is declared
(§6.6.1). A type may be declared public to
grant access to the type from code in other packages of the
same module, and potentially from code in packages of other
modules.
It is a compile-time error if a top level type declaration contains
any one of the following access modifiers: protected, private, or
static.
It is a compile-time error if the name of a top level type appears as the name of any other top level class or interface type declared in the same package.
The scope and shadowing of a top level type is specified in §6.3 and §6.4.
The fully qualified name of a top level type is specified in §6.7.
Example 7.6-1. Conflicting Top Level Type Declarations
package test;
import java.util.Vector;
class Point {
int x, y;
}
interface Point { // compile-time error #1
int getR();
int getTheta();
}
class Vector { Point[] pts; } // compile-time error #2
Here, the first compile-time error is caused by the
duplicate declaration of the name Point as both a
class and an interface in the same package. A second compile-time
error is the attempt to declare the name
Vector both by a class type declaration and by a
single-type-import declaration.
Note, however, that it is not an error for the name of a class to also name a type that otherwise might be imported by a type-import-on-demand declaration (§7.5.2) in the compilation unit (§7.3) containing the class declaration. Thus, in this program:
package test;
import java.util.*;
class Vector {} // not a compile-time error
the declaration of the
class Vector is permitted even though there is also
a class java.util.Vector. Within this compilation
unit, the simple name Vector refers to the
class test.Vector, not
to java.util.Vector (which can still be referred to
by code within the compilation unit, but only by its fully qualified
name).
Example 7.6-2. Scope of Top Level Types
package points;
class Point {
int x, y; // coordinates
PointColor color; // color of this point
Point next; // next point with this color
static int nPoints;
}
class PointColor {
Point first; // first point with this color
PointColor(int color) { this.color = color; }
private int color; // color components
}
This program defines two classes that use each other
in the declarations of their class members. Because the class
types Point and PointColor have
all the type declarations in package points,
including all those in the current compilation unit, as their scope,
this program compiles correctly. That is, forward reference is not a
problem.
Example 7.6-3. Fully Qualified Names
class Point { int x, y; }
In this code, the class Point is
declared in a compilation unit with no package declaration, and
thus Point is its fully qualified name, whereas in
the code:
package vista;
class Point { int x, y; }
the fully qualified name of the
class Point is vista.Point. (The
package name vista is suitable for local or
personal use; if the package were intended to be widely distributed,
it would be better to give it a unique package name
(§6.1).)
An implementation of the Java SE Platform must keep track of types within packages by the combination of their enclosing module names and their binary names (§13.1). Multiple ways of naming a type must be expanded to binary names to make sure that such names are understood as referring to the same type.
For example, if a compilation unit contains the single-type-import declaration (§7.5.1):
import java.util.Vector;
then within that compilation unit, the simple
name Vector and the fully qualified
name java.util.Vector refer to the same
type.
If and only if packages are stored in a file system
(§7.2), the host system may choose to enforce the
restriction that it is a compile-time error if a type is not found in
a file under a name composed of the type name plus an extension (such
as .java or .jav) if either of
the following is true:
This restriction implies that there must be at most
one such type per compilation unit. This restriction makes it easy for
a Java compiler to find a
named class within a package. In practice, many programmers choose to
put each class or interface type in its own compilation unit, whether
or not it is public or is referred to by code in other compilation
units.
For example, the source code for a public
type wet.sprocket.Toad would be found in a
file Toad.java in the
directory wet/sprocket, and the corresponding
object code would be found in the file Toad.class
in the same directory.
A module declaration specifies a new named module. A named module specifies dependences on other modules to define the universe of classes and interfaces available to its own code; and specifies which of its packages are exported or opened in order to populate the universe of classes and interfaces available to other modules which specify a dependence on it.
A "dependence" is what is expressed by a requires
directive, independent of whether a module exists with the name
specified by the directive. A "dependency" is the observable module
enumerated by resolution (as described in the java.lang.module package
specification) for a given requires directive. Generally, the rules
of the Java programming language are more interested in dependences than dependencies.
A module declaration introduces a module name that can be used in
other module declarations to express relationships between modules. A
module name consists of one or more Java identifiers
(§3.8) separated by "." tokens.
There are two kinds of modules: normal modules and open modules. The kind of a module determines the nature of access to the module's types, and the members of those types, for code outside the module.
A normal module, without the open modifier, grants access at compile
time and run time to types in only those packages which are explicitly
exported.
An open module, with the open modifier, grants access at compile
time to types in only those packages which are explicitly exported,
but grants access at run time to types in all its packages, as if all
packages had been exported.
For code outside a module (whether the module is normal or open), the
access granted at compile time or run time to types in the module's
exported packages is specifically to the public and protected
types in those packages, and the public and protected members of
those types (§6.6). No access is granted at
compile time or run time to types, or their members, in packages which
are not exported. Code inside the module may access public and
protected types, and the public and protected members of those
types, in all packages in the module at both compile time and run
time.
Distinct from access at compile time and access at run time, the Java SE Platform provides reflective access via the Core Reflection API (§1.4). A normal module grants reflective access to types in only those packages which are explicitly exported or explicitly opened (or both). An open module grants reflective access to types in all its packages, as if all packages had been opened.
For code outside a normal module, the reflective access granted to
types in the module's exported (and not opened) packages is
specifically to the public and protected types in those packages,
and the public and protected members of those types. The
reflective access granted to types in the module's opened packages
(whether exported or not) is to all types in those packages, and all
members of those types. No reflective access is granted to types, or
their members, in packages which are not exported or opened. Code
inside the module enjoys reflective access to all types, and all their
members, in all packages in the module.
For code outside an open module, the reflective access granted to types in the module's opened packages (that is, all packages in the module) is to all types in those packages, and all members of those types. Code inside the module enjoys reflective access to all types, and all their members, in all packages in the module.
The directives of a module declaration specify
the module's dependences on other modules (via requires,
§7.7.1), the packages it makes available to other
modules (via exports and opens, §7.7.2), the
services it consumes (via uses, §7.7.3), and
the services it provides (via provides, §7.7.4).
requires
{RequiresModifier}
ModuleName ; exports PackageName
[to ModuleName
{, ModuleName}] ; opens PackageName
[to ModuleName
{, ModuleName}] ; uses TypeName ; provides TypeName
with TypeName
{, TypeName} ;
If and only if packages are stored in a file system
(§7.2), the host system may choose to enforce the
restriction that it is a compile-time error if a module declaration is
not found in a file under a name composed
of module-info plus an extension (such
as .java or .jav).
To aid comprehension, it is customary, though not required, for a
module declaration to group its directives, so that the requires
directives which pertain to modules are visually distinct from the
exports and opens directives which pertain to packages, and from
the uses and provides directives which pertain to services. For
example:
module com.example.foo {
requires com.example.foo.http;
requires java.logging;
requires transitive com.example.foo.network;
exports com.example.foo.bar;
exports com.example.foo.internal to com.example.foo.probe;
opens com.example.foo.quux;
opens com.example.foo.internal to com.example.foo.network,
com.example.foo.probe;
uses com.example.foo.spi.Intf;
provides com.example.foo.spi.Intf with com.example.foo.Impl;
}
The opens directives can be avoided if the module
is open:
open module com.example.foo {
requires com.example.foo.http;
requires java.logging;
requires transitive com.example.foo.network;
exports com.example.foo.bar;
exports com.example.foo.internal to com.example.foo.probe;
uses com.example.foo.spi.Intf;
provides com.example.foo.spi.Intf with com.example.foo.Impl;
}
Development tools for the Java programming language are encouraged
to highlight requires transitive directives and unqualified
exports directives, as these form the primary API of a module.
The requires directive specifies the name of a module on which the
current module has a dependence.
A requires directive must not appear in the declaration of the
java.base module, or a compile-time error occurs, because it is the
primordial module and has no dependences (§8.1.4).
If the declaration of a module does not express a dependence on the
java.base module, and the module is not itself java.base, then the
module has an implicitly declared dependence on the java.base module.
The requires keyword may be followed by the modifier transitive.
This causes any module which requires the current module to have an
implicitly declared dependence on the module specified by the
requires transitive directive.
The requires keyword may be followed by the modifier static. This
specifies that the dependence, while mandatory at compile time, is
optional at run time.
If the declaration of a module expresses a dependence on the
java.base module, and the module is not itself java.base, then it
is a compile-time error if a modifier appears after the requires
keyword.
It is a compile-time error if more than one requires directive in a
module declaration specifies the same module name.
It is a compile-time error if resolution, as described in the
java.lang.module package specification, with the current module as the
only root module, fails for any of the reasons described in the
java.lang.module package specification.
For example, if a requires directive specifies a
module that is not observable, or if the current module directly or
indirectly expresses a dependence on itself.
If resolution succeeds, then its result specifies the modules that are read by the current module. The modules read by the current module determine which ordinary compilation units are visible to the current module (§7.3). The types declared in those ordinary compilation units (and only those ordinary compilation units) may be accessible to code in the current module (§6.6).
The Java SE Platform distinguishes between named modules
that are explicitly declared (that is, with a module declaration) and
named modules that are implicitly declared (that is, automatic
modules). However, the Java programming language does not surface the distinction:
requires directives refer to named modules without regard for
whether they are explicitly declared or implicitly declared.
While automatic modules are convenient for
migration, they are unreliable in the sense that their names and
exported packages may change when their authors convert them to
explicitly declared modules. A Java compiler is encouraged to issue a
warning if a requires directive refers to an automatic module. An
especially strong warning is recommended if the transitive modifier
appears in the directive.
Example 7.1.1-1. Resolution of requires transitive directives
Suppose there are four module declarations as follows:
module m.A {
requires m.B;
}
module m.B {
requires transitive m.C;
}
module m.C {
requires transitive m.D;
}
module m.D {
exports p;
}
where the package p exported
by m.D is declared as follows:
package p;
public class Point {}
and where a package client in
module m.A refers to the public type
Point in the exported package p:
package client;
import p.Point;
public class Test {
public static void main(String[] args) {
System.out.println(new Point());
}
}
The modules may be compiled as follows, assuming that the current directory has one subdirectory per module, named after the module it contains:
javac --module-source-path . -d . --module m.D javac --module-source-path . -d . --module m.C javac --module-source-path . -d . --module m.B javac --module-source-path . -d . --module m.A
The program client.Test may be
run as follows:
java --module-path . --module m.A/client.Test
The reference from code in m.A to
the exported public type Point
in m.D is legal because m.A
reads m.D, and m.D exports the
package containing Point. Resolution determines
that m.A reads m.D as
follows:
m.A requires m.B and
therefore reads m.B.
Since m.A reads m.B, and
since m.B requires
transitive m.C, resolution determines that
m.A reads m.C.
Then, since m.A reads m.C,
and since m.C requires
transitive m.D, resolution determines that
m.A reads m.D.
In effect, a module may read another module through
multiple levels of dependence, in order to support arbitrary amounts
of refactoring. Once a module is released for someone to reuse (via
requires), the module's author has committed to its name and API but
is free to refactor its content into other modules which the original
module reuses (via requires transitive) for the benefit of
consumers. In the example above, package p may have
been exported originally by m.B
(thus, m.A requires m.B) but
refactoring has caused some of m.B's content to
move into m.C and m.D. By using
a chain of requires transitive directives, the family
of m.B, m.C,
and m.D can preserve access to
package p for code in m.A
without forcing any changes to the requires directives
of m.A. Note that package p
in m.D is not "re-exported"
by m.C and m.B;
rather, m.A is made to read m.D
directly.
The exports directive specifies the name of a package to be exported
by the current module. For code in other modules, this grants access
at compile time and run time to the public and protected types in
the package, and the public and protected members of those types
(§6.6). It also grants reflective access to those
types and members for code in other modules.
The opens directive specifies the name of a package to be opened by
the current module. For code in other modules, this grants access at
run time, but not compile time, to the public and protected types
in the package, and the public and protected members of those
types. It also grants reflective access to all types in the package,
and all their members, for code in other modules.
It is a compile-time error if the package specified by exports is
not declared by a compilation unit associated with the current module
(§7.3).
It is permitted for opens to specify a package which is not declared
by a compilation unit associated with the current module. (If the
package should happen to be declared by an observable compilation unit
associated with another module, the opens directive has no effect on
that other module.)
It is a compile-time error if more than one exports directive in a
module declaration specifies the same package name.
It is a compile-time error if more than one opens directive in a
module declaration specifies the same package name.
It is a compile-time error if an opens directive appears in the
declaration of an open module.
If an exports or opens directive has a to clause, then the
directive is qualified; otherwise, it
is unqualified. For a qualified directive, the
public and protected types in the package, and their public and
protected members, are accessible solely to code in the modules
specified in the to clause. The modules specified in the to clause
are referred to as friends of the current
module. For an unqualified directive, these types and their members
are accessible to code in any module.
It is permitted for the to clause of an exports or opens
directive to specify a module which is not observable
(§7.7.6).
It is a compile-time error if the to clause of a given exports
directive specifies the same module name more than once.
It is a compile-time error if the to clause of a given opens
directive specifies the same module name more than once.
The uses directive specifies a service for
which the current module may discover providers via java.util.ServiceLoader.
The service must be a class type, an interface type, or an annotation
type. It is a compile-time error if a uses directive specifies an
enum type (§8.9) as the service.
The service may be declared in the current module or in another module. If the service is not declared in the current module, then the service must be accessible to code in the current module (§6.6), or a compile-time error occurs.
It is a compile-time error if more than one uses directive in a
module declaration specifies the same service.
The provides directive specifies a service for which the with
clause specifies one or more service providers to
java.util.ServiceLoader.
The service must be a class type, an interface type, or an annotation
type. It is a compile-time error if a provides directive specifies
an enum type (§8.9) as the service.
The service may be declared in the current module or in another module. If the service is not declared in the current module, then the service must be accessible to code in the current module (§6.6), or a compile-time error occurs.
Every service provider must be a class type or an interface type, that
is public, and that is top level or nested static, or a
compile-time error occurs.
Every service provider must be declared in the current module, or a compile-time error occurs.
If a service provider explicitly declares a public constructor with
no formal parameters, or implicitly declares a public default
constructor (§8.8.9), then that constructor is
called the provider constructor.
If a service provider explicitly declares a public static method
called provider with no formal parameters, then
that method is called the provider method.
If a service provider has a provider method, then its return type must
(i) either be declared in the current module, or be declared in another
module and be accessible to code in the current module; and (ii) be a
subtype of the service specified in the provides directive; or a
compile-time error occurs.
While a service provider that is specified by a
provides directive must be declared in the current module, its
provider method may have a return type that is declared in
another module. Also, note that when a service
provider declares a provider method, the service provider itself need
not be a subtype of the service.
If a service provider does not have a provider method, then that
service provider must have a provider constructor and must be a
subtype of the service specified in the provides directive, or a
compile-time error occurs.
It is a compile-time error if more than one provides directive in a
module declaration specifies the same service.
It is a compile-time error if the with clause of a given provides
directive specifies the same service provider more than once.
An observable ordinary compilation unit that the host system does not associate with a named module (§7.3) is associated with an unnamed module.
Unnamed modules are provided by the Java SE Platform in recognition of the fact that programs developed prior to Java SE 9 could not declare named modules. In addition, the reasons for the Java SE Platform providing unnamed packages (§7.4.2) are largely applicable to unnamed modules.
An implementation of the Java SE Platform must support at least one unnamed module. An implementation may support more than one unnamed module, but is not required to do so. Which ordinary compilation units are associated with each unnamed module is determined by the host system.
The host system may associate ordinary compilation units in a named package with an unnamed module.
The rules for unnamed modules are designed to maximize their interoperation with named modules, as follows:
An unnamed module reads every observable module (§7.7.6).
By virtue of the fact that an ordinary compilation unit associated with an unnamed module is observable, the associated unnamed module is observable. Thus, if the implementation of the Java SE Platform supports more than one unnamed module, every unnamed module is observable; and each unnamed module reads every unnamed module including itself.
However, it is important to realize that the
ordinary compilation units of an unnamed module are
never visible to a named module
(§7.3) because no requires directive can
arrange for a named module to read an unnamed module. The Core
Reflection API of the Java SE Platform may be used to arrange for a
named module to read an unnamed module at run time.
An unnamed module exports every package whose ordinary compilation units are associated with that unnamed module.
An unnamed module opens every package whose ordinary compilation units are associated with that unnamed module.
A module is observable if at least one of the following is true:
A modular compilation unit containing the declaration of the module is observable (§7.3).
An ordinary compilation unit associated with the module is observable.