JEP 433: Pattern Matching for switch (Fourth Preview)
Owner | Gavin Bierman |
Type | Feature |
Scope | SE |
Status | Closed / Delivered |
Release | 20 |
Component | specification / language |
Discussion | amber dash dev at openjdk dot org |
Relates to | JEP 427: Pattern Matching for switch (Third Preview) |
JEP 432: Record Patterns (Second Preview) | |
JEP 441: Pattern Matching for switch | |
Reviewed by | Alex Buckley, Brian Goetz |
Endorsed by | Brian Goetz |
Created | 2022/09/23 14:22 |
Updated | 2023/05/12 15:34 |
Issue | 8294285 |
Summary
Enhance the Java programming language with pattern matching for switch
expressions and statements. Extending pattern matching to switch
allows an
expression to be tested against a number of patterns, each with a specific
action, so that complex data-oriented queries can be expressed concisely and
safely. This is a preview language feature.
History
Pattern Matching for switch
was proposed as a preview feature by
JEP 406 and delivered in
JDK 17, proposed for a second
preview by JEP 420 and delivered in
JDK 18, and proposed for a
third preview by JEP 427 and delivered
in JDK 19. This JEP proposes a
fourth preview to enable its continued co-evolution with the
Record Patterns preview feature (JEP 432) and to
allow other refinements based upon continued experience and feedback.
The main changes since the third preview are:
-
An exhaustive switch (i.e., a
switch
expression or a patternswitch
statement) over anenum
class now throwsMatchException
rather thanIncompatibleClassChangeError
if no switch label applies at run time. -
The grammar for switch labels is simpler.
-
Inference of type arguments for generic record patterns is now supported in
switch
expressions and statements, along with the other constructs that support patterns.
Goals
-
Expand the expressiveness and applicability of
switch
expressions and statements by allowing patterns to appear incase
labels. -
Allow the historical null-hostility of
switch
to be relaxed when desired. -
Increase the safety of
switch
statements by requiring that patternswitch
statements cover all possible input values. -
Ensure that all existing
switch
expressions and statements continue to compile with no changes and execute with identical semantics.
Motivation
In Java 16, JEP 394 extended the instanceof
operator to take a
type pattern and perform pattern matching. This modest extension allows the
familiar instanceof-and-cast idiom to be simplified, making it both more concise
and less error-prone:
// Old code
if (obj instanceof String) {
String s = (String)obj;
... use s ...
}
// New code
if (obj instanceof String s) {
... use s ...
}
We often want to compare a variable such as obj
against multiple alternatives.
Java supports multi-way comparisons with switch
statements and, since
Java 14, switch
expressions (JEP 361), but unfortunately
switch
is very limited. You can only switch on values of a few types —
integral primitive types (excluding long
), their corresponding boxed forms,
enum
types, and String
— and you can only test for exact equality against
constants. We might like to use patterns to test the same variable against a
number of possibilities, taking a specific action on each, but since the
existing switch
does not support that we end up with a chain of if...else
tests such as:
static String formatter(Object obj) {
String formatted = "unknown";
if (obj instanceof Integer i) {
formatted = String.format("int %d", i);
} else if (obj instanceof Long l) {
formatted = String.format("long %d", l);
} else if (obj instanceof Double d) {
formatted = String.format("double %f", d);
} else if (obj instanceof String s) {
formatted = String.format("String %s", s);
}
return formatted;
}
This code benefits from using pattern instanceof
expressions, but it is far
from perfect. First and foremost, this approach allows coding errors to remain
hidden because we have used an overly general control construct. The intent is
to assign something to formatted
in each arm of the if...else
chain, but
there is nothing that enables the compiler to identify and verify this
invariant. If some block — perhaps one that is executed rarely — does not assign
to formatted
, we have a bug. (Declaring formatted
as a blank local would at
least enlist the compiler’s definite-assignment analysis in this effort, but
developers do not always write such declarations.) In addition, the above code
is not optimizable; absent compiler heroics it will have O(n) time
complexity, even though the underlying problem is often O(1).
But switch
is a perfect match for pattern matching! If we extend switch
statements and expressions to work on any type, and allow case
labels with
patterns rather than just constants, then we can rewrite the above code more
clearly and reliably:
static String formatterPatternSwitch(Object obj) {
return switch (obj) {
case Integer i -> String.format("int %d", i);
case Long l -> String.format("long %d", l);
case Double d -> String.format("double %f", d);
case String s -> String.format("String %s", s);
default -> obj.toString();
};
}
The semantics of this switch
are clear: A case
label with a pattern applies
if the value of the selector expression obj
matches the pattern. (We have shown
a switch
expression for brevity but could instead have shown a switch
statement; the switch block, including the case
labels, would be unchanged.)
The intent of this code is clearer because we are using the right control
construct: We are saying, "the parameter obj
matches at most one of the
following conditions, figure it out and evaluate the corresponding arm." As a
bonus, it is more optimizable; in this case we are more likely to be able to
perform the dispatch in O(1) time.
Switches and null
Traditionally, switch
statements and expressions throw NullPointerException
if the selector expression evaluates to null
, so testing for null
must be
done outside of the switch
:
static void testFooBar(String s) {
if (s == null) {
System.out.println("Oops!");
return;
}
switch (s) {
case "Foo", "Bar" -> System.out.println("Great");
default -> System.out.println("Ok");
}
}
This was reasonable when switch
supported only a few reference types.
However, if switch
allows a selector expression of any type, and case
labels
can have type patterns, then the standalone null
test feels like an arbitrary
distinction, and invites needless boilerplate and opportunity for error. It
would be better to integrate the null
test into the switch
by allowing a new
null
case label:
static void testFooBar(String s) {
switch (s) {
case null -> System.out.println("Oops");
case "Foo", "Bar" -> System.out.println("Great");
default -> System.out.println("Ok");
}
}
The behavior of the switch
when the value of the selector expression is null
is always determined by its case
labels. With a case null
, the switch
executes the code associated with that label; without a case null
, the
switch
throws NullPointerException
, just as before. (To maintain backward
compatibility with the current semantics of switch
, the default
label does
not match a null
selector.)
Case refinement
Experimentation with patterns in switch
suggests it is common to want to
refine the test embodied by a pattern label. For example, consider the following
code that switches over a Shape
value:
class Shape {}
class Rectangle extends Shape {}
class Triangle extends Shape { int calculateArea() { ... } }
static void testTriangle(Shape s) {
switch (s) {
case null:
break;
case Triangle t:
if (t.calculateArea() > 100) {
System.out.println("Large triangle");
break;
}
default:
System.out.println("A shape, possibly a small triangle");
}
}
The intent of this code is to have a special case for large triangles (those
whose area is over 100), and a default case for everything else (including small
triangles). However, we cannot express this directly with a single pattern. We
first have to write a case
label that matches all triangles, and then place
the test of the area of the triangle rather uncomfortably within the
corresponding statement group. Then we have to use fall-through to get the
correct behavior when the triangle has an area less than 100. (Note the careful
placement of the break
statement inside the if
block.)
The problem here is that using a single pattern to discriminate among
cases does not scale beyond a single condition — we need some way to
express a refinement to a pattern. We therefore allow when
clauses in
switch blocks to specify guards to pattern case
labels, e.g., case Triangle t when t.calculateArea() > 100
. We refer to such a case
label
as a guarded case
label, and to the boolean expression as the
guard.
With this approach, we can revisit the testTriangle
code to express the
special case for large triangles directly. This eliminates the use of
fall-through in the switch
statement, which in turn means we can enjoy concise
arrow-style (->
) rules:
static void testTriangle(Shape s) {
switch (s) {
case null ->
{ break; }
case Triangle t
when t.calculateArea() > 100 ->
System.out.println("Large triangle");
default ->
System.out.println("A shape, possibly a small triangle");
}
}
The second clause is taken if the value of s
matches the pattern Triangle t
and subsequently the guard t.calculateArea() > 100
evaluates to true
.
(The guard is able to use any pattern variables that are declared by the pattern
in the case
label.)
Using switch
makes it easy to understand and change case labels when
application requirements change. For example, we might want to split triangles
out of the default path; we can do that by using two case clauses, one with a
guard and one without:
static void testTriangle(Shape s) {
switch (s) {
case null ->
{ break; }
case Triangle t
when t.calculateArea() > 100 ->
System.out.println("Large triangle");
case Triangle t ->
System.out.println("Small triangle");
default ->
System.out.println("Non-triangle");
}
}
Description
We enhance switch
statements and expressions in three ways:
-
Extend
case
labels to include patterns andnull
in addition to constants, -
Broaden the range of types permitted for the selector expressions of both
switch
statements andswitch
expressions, and -
Allow optional
when
clauses to followcase
labels.
For convenience, we also introduce parenthesized patterns.
Patterns in switch labels
We revise the grammar for switch labels in a switch block to read (compare JLS §14.11.1):
SwitchLabel:
case CaseConstant { , CaseConstant }
case null [, default]
case Pattern
default
The main enhancement is to introduce a new case
label, case p
, where p
is a
pattern. The essence of switch
is unchanged: The value of the selector
expression is compared to the switch labels, one of the labels is selected, and
the code associated with that label is executed or evaluated. The difference now
is that for case
labels with patterns, the label selected is determined by the
result of pattern matching rather than by an equality test. For example, in the
following code, the value of obj
matches the pattern Long l
, and the
expression associated with the label case Long l
is evaluated:
Object obj = 123L;
String formatted = switch (obj) {
case Integer i -> String.format("int %d", i);
case Long l -> String.format("long %d", l);
case Double d -> String.format("double %f", d);
case String s -> String.format("String %s", s);
default -> obj.toString();
};
After a successful pattern match we often further test the result of the match. This can lead to cumbersome code, such as:
static void test(Object obj) {
switch (obj) {
case String s:
if (s.length() == 1) { ... }
else { ... }
break;
...
}
}
The desired test — that obj
is a String
of length 1 — is unfortunately split
between the pattern case
label and the following if
statement.
To address this we introduce guarded pattern case
labels by supporting an optional
guard after the pattern label. This allows the above code to be rewritten so
that all the conditional logic is lifted into the switch label:
static void test(Object obj) {
switch (obj) {
case String s when s.length() == 1 -> ...
case String s -> ...
...
}
}
The first clause matches if obj
is both a String
and of length 1. The second
case matches if obj
is a String
of any length.
Only pattern labels can have a guard. For example, it is not valid to write a
label with a case
constant and a guard; e.g., case "Hello" when RandomBooleanExpression()
.
Sometimes we need to parenthesize patterns to improve readability. We therefore
extend the language of patterns to support parenthesized patterns written
(p)
, where p
is a pattern. A parenthesized pattern (p)
introduces the
pattern variables that are introduced by the subpattern p
. A value matches a
parenthesized pattern (p)
if it matches the pattern p
.
There are five major language design areas to consider when supporting
patterns in switch
:
- Enhanced type checking
- Exhaustiveness of
switch
expressions and statements - Scope of pattern variable declarations
- Dealing with
null
- Errors
1. Enhanced type checking
1a. Selector expression typing
Supporting patterns in switch
means that we can relax the current restrictions
on the type of the selector expression. Currently the type of the selector
expression of a normal switch
must be either an integral primitive type
(excluding long
), the corresponding boxed form (i.e., Character
, Byte
,
Short
, or Integer
), String
, or an enum
type. We extend this and require
that the type of the selector expression be either an integral primitive type
(excluding long
) or any reference type.
For example, in the following pattern switch
the selector expression obj
is
matched with type patterns involving a class type, an enum
type, a record type,
and an array type, along with a null
case
label and a default
:
record Point(int i, int j) {}
enum Color { RED, GREEN, BLUE; }
static void typeTester(Object obj) {
switch (obj) {
case null -> System.out.println("null");
case String s -> System.out.println("String");
case Color c -> System.out.println("Color: " + c.toString());
case Point p -> System.out.println("Record class: " + p.toString());
case int[] ia -> System.out.println("Array of ints of length" + ia.length);
default -> System.out.println("Something else");
}
}
Every case
label in the switch block must be compatible with the selector
expression. For a case
label with a pattern, known as a pattern label, we
use the existing notion of compatibility of an expression with a pattern
(JLS §14.30.1).
1b. Dominance of case
labels
Supporting pattern case
labels means that for a given value of the selector
expression it is now possible for more than one case
label to potentially apply
(previously, at most only one case
label could apply). For example, if the
selector expression evaluates to a String
then both the case
labels case String s
and case CharSequence cs
would apply.
The first issue to resolve is deciding exactly which label should apply in this
circumstance. Rather than attempt a complicated best-fit approach, we adopt a
simpler semantics: The first case
label appearing in a switch block that
applies to a value is chosen.
static void first(Object obj) {
switch (obj) {
case String s ->
System.out.println("A string: " + s);
case CharSequence cs ->
System.out.println("A sequence of length " + cs.length());
default -> {
break;
}
}
}
In this example, if the value of obj
is of type String
then the first case
label will apply; if it is of type CharSequence
but not of type String
then
the second pattern label will apply.
But what happens if we swap the order of these two case
labels?
static void error(Object obj) {
switch (obj) {
case CharSequence cs ->
System.out.println("A sequence of length " + cs.length());
case String s -> // Error - pattern is dominated by previous pattern
System.out.println("A string: " + s);
default -> {
break;
}
}
}
Now if the value of obj
is of type String
the CharSequence
case
label
applies, since it appears first in the switch block. The String
case
label
is unreachable in the sense that there is no value of the selector expression
that would cause it to be chosen. By analogy to unreachable code, this is
treated as a programmer error and results in a compile-time error.
More precisely, we say that the first case
label case CharSequence cs
dominates the second case
label case String s
because every value that
matches the pattern String s
also matches the pattern CharSequence cs
, but
not vice versa. This is because the type of the second pattern, String
, is a
subtype of the type of the first pattern, CharSequence
.
An unguarded pattern case
label dominates a guarded pattern case
label that
has the same pattern. For example, the (unguarded) pattern case
label case String s
dominates the guarded pattern case
label case String s when s.length() > 0
, since every value that matches the case
label case String s when s.length() > 0
must match the case
label case String s
.
A guarded pattern case
label dominates another pattern case
label (guarded
or unguarded) only when both the former's pattern dominates the latter's pattern
and when its guard is a constant expression of value true
. For example, the
guarded pattern case
label case String s when true
dominates the pattern
case
label case String s
. We do not analyze the guarding expression any
further in order to determine more precisely which values match the pattern
label (a problem which is undecidable in general).
A pattern case
label can dominate a constant case
label. For example, the
pattern case
label case Integer i
dominates the constant case
label case 42
, and the pattern case
label case E e
dominates the constant case
label
case A
when A
is a member of enum
class type E
. A guarded pattern
case
label dominates a constant case
label if the same pattern case
label
without the guard does. In other words, we do not check the guard, since this is
undecidable in general. For example, the pattern case
label case String s when s.length() > 1
dominates the constant case
label case "hello"
, as
expected; but case Integer i when i != 0
dominates the case
label case 0
.
All of this suggests a simple, predictable, and readable ordering of case
labels
in which the constant case
labels should appear before the guarded pattern
case
labels, and those should appear before the unguarded pattern case
labels:
Integer i = ...
switch (i) {
case -1, 1 -> ... // Special cases
case Integer i when i > 0 -> ... // Positive integer cases
case Integer i -> ... // All the remaining integers
}
The compiler checks all case
labels. It is a compile-time error for a case
label in a switch block to be dominated by a preceding case
label in that
switch block. This dominance requirement ensures that if a switch block
contains only type pattern case
labels, they will appear in subtype order.
(The notion of dominance is analogous to conditions on the catch
clauses of a
try
statement, where it is an error if a catch
clause that catches an
exception class E
is preceded by a catch
clause that can catch E
or a
superclass of E
(JLS §11.2.3). Logically, the preceding catch
clause dominates the subsequent catch
clause.)
It is also a compile-time error for a switch block of a switch
expression or
switch
statement to have more than one match-all switch label. The match-all
labels are default
and pattern case
labels where the pattern unconditionally
matches the selector expression. For example, the type pattern String s
unconditionally matches a selector expression of type String
, and the type
pattern Object o
unconditionally matches a selector expression of any
reference type.
1c. Inference of type arguments in record patterns
If a record pattern names a generic record class but gives no type arguments (i.e., the record pattern uses a raw type) then the type arguments are always inferred. For example:
record MyPair<S,T>(S fst, T snd){};
static void recordInference(MyPair<String, Integer> pair){
switch (pair) {
case MyPair(var f, var s) ->
... // Inferred record Pattern MyPair<String,Integer>(var f, var s)
...
}
}
Inference of type arguments for record patterns is supported in all constructs
that support patterns: switch
statements and expressions, instanceof
expressions, and enhanced for
statements.
2. Exhaustiveness of switch
expressions and statements
A switch
expression requires that all possible values of the selector
expression be handled in the switch block; in other words, it must be exhaustive.
This maintains the property that successful evaluation of a switch
expression
will always yield a value. For normal switch
expressions, this is enforced by
a fairly straightforward set of extra conditions on the switch block.
For pattern switch
expressions and statements, we achieve this by defining a
notion of type coverage of switch labels in a switch block. The type coverage
of all the switch labels in the switch block is then combined to determine if
the switch block exhausts all the possibilities of the selector expression.
Consider this (erroneous) pattern switch
expression:
static int coverage(Object obj) {
return switch (obj) { // Error - not exhaustive
case String s -> s.length();
};
}
The switch block has only one switch label, case String s
. This matches any
value of obj
whose type is a subtype of String
. We
therefore say that the type coverage of this switch label is every subtype of
String
. This pattern switch
expression is not exhaustive because the type
coverage of its switch block (all subtypes of String
) does not include the
type of the selector expression (Object
).
Consider this (still erroneous) example:
static int coverage(Object obj) {
return switch (obj) { // Error - still not exhaustive
case String s -> s.length();
case Integer i -> i;
};
}
The type coverage of this switch block is the union of the coverage of its two
switch labels. In other words, the type coverage is the set of all subtypes of
String
and the set of all subtypes of Integer
. But, again, the type coverage
still does not include the type of the selector expression, so this pattern
switch
expression is also not exhaustive and causes a compile-time error.
The type coverage of a default
label is all types, so this example is (at
last!) legal:
static int coverage(Object obj) {
return switch (obj) {
case String s -> s.length();
case Integer i -> i;
default -> 0;
};
}
If the type of the selector expression is a sealed class
(JEP 409) then the type coverage check can take into account the
permits
clause of the sealed class to determine whether a switch block is
exhaustive. This can sometimes remove the need for a default
clause. Consider
the following example of a sealed
interface S
with three permitted
subclasses A
, B
, and C
:
sealed interface S permits A, B, C {}
final class A implements S {}
final class B implements S {}
record C(int i) implements S {} // Implicitly final
static int testSealedExhaustive(S s) {
return switch (s) {
case A a -> 1;
case B b -> 2;
case C c -> 3;
};
}
The compiler can determine that the type coverage of the switch block is the
types A
, B
, and C
. Since the type of the selector expression, S
, is a
sealed interface whose permitted subclasses are exactly A
, B
, and C
, this
switch block is exhaustive. As a result, no default
label is needed.
Some extra care is needed when a permitted direct subclass only implements a
specific parameterization of a (generic) sealed
superclass. For example:
sealed interface I<T> permits A, B {}
final class A<X> implements I<String> {}
final class B<Y> implements I<Y> {}
static int testGenericSealedExhaustive(I<Integer> i) {
return switch (i) {
// Exhaustive as no A case possible!
case B<Integer> bi -> 42;
}
}
The only permitted subclasses of I
are A
and B
, but the compiler can
detect that the switch block need only cover the class B
to be exhaustive
since the selector expression is of type I<Integer>
.
This condition of exhaustiveness applies to both pattern switch
expressions
and pattern switch
statements. To ensure backward compatibility, all
existing switch
statements will compile unchanged. But if a switch
statement
uses any of the switch
enhancements described in this JEP then the compiler
will check that it is exhaustive. (Future compilers of the Java language may
emit warnings for legacy switch
statements that are not exhaustive.)
More precisely, exhaustiveness is required of any switch
statement that uses
pattern or null
labels or whose selector expression is not one of the legacy
types (char
, byte
, short
, int
, Character
, Byte
, Short
, Integer
,
String
, or an enum
type). For example:
sealed interface S permits A, B, C {}
final class A implements S {}
final class B implements S {}
record C(int i) implements S {} // Implicitly final
static void switchStatementExhaustive(S s) {
switch (s) { // Error - not exhaustive;
// missing clause for permitted class B!
case A a :
System.out.println("A");
break;
case C c :
System.out.println("C");
break;
};
}
Making most switches exhaustive is just a matter of adding a simple default
clause at the end of the switch block. This leads to clearer and easier to
verify code. For example, the following pattern switch
statement is not
exhaustive and is erroneous:
Object obj = ...
switch (obj) { // Error - not exhaustive!
case String s:
System.out.println(s);
break;
case Integer i:
System.out.println("Integer");
break;
}
It can be made exhaustive trivially:
Object obj = ...
switch (obj) {
case String s:
System.out.println(s);
break;
case Integer i:
System.out.println("Integer");
break;
default: // Now exhaustive!
break;
}
The notion of exhaustiveness is made more complicated by record patterns (JEP 432) since these patterns support the nesting of other patterns inside them. Accordingly, the notion of exhaustiveness has to reflect this potentially recursive structure.
3. Scope of pattern variable declarations
Pattern variables (JEP 394) are local variables that are
declared by patterns. Pattern variable declarations are unusual in that their
scope is flow-sensitive. As a recap consider the following example, where the
type pattern String s
declares the pattern variable s
:
static void test(Object obj) {
if ((obj instanceof String s) && s.length() > 3) {
System.out.println(s);
} else {
System.out.println("Not a string");
}
}
The declaration of s
is in scope in the right-hand operand of the &&
expression, as well as in the "then" block. However, it is not in scope in the
"else" block: In order for control to transfer to the "else" block the pattern
match must fail, in which case the pattern variable will not have been
initialized.
We extend this flow-sensitive notion of scope for pattern variable declarations
to encompass pattern declarations occurring in case
labels with three new
rules:
-
The scope of a pattern variable declaration which occurs in a switch label includes any
when
clause of that label. -
The scope of a pattern variable declaration which occurs in a
case
label of aswitch
rule includes the expression, block, orthrow
statement that appears to the right of the arrow. -
The scope of a pattern variable declaration which occurs in a
case
label of aswitch
labeled statement group includes the block statements of the statement group. Falling through acase
label that declares a pattern variable is forbidden.
This example shows the first rule in action:
static void test(Object obj) {
switch (obj) {
case Character c
when c.charValue() == 7:
System.out.println("Ding!");
break;
default:
break;
}
}
}
The scope of the declaration of the pattern variable c
includes the when
expression of the switch label.
This variant shows the second rule in action:
static void test(Object obj) {
switch (obj) {
case Character c -> {
if (c.charValue() == 7) {
System.out.println("Ding!");
}
System.out.println("Character");
}
case Integer i ->
throw new IllegalStateException("Invalid Integer argument: "
+ i.intValue());
default -> {
break;
}
}
}
Here the scope of the declaration of the pattern variable c
is the block to
the right of the first arrow. The scope of the declaration of the pattern
variable i
is the throw
statement to the right of the second arrow.
The third rule is more complicated. Let us first consider an example where
there is only one case
label for a switch
labeled statement group:
static void test(Object obj) {
switch (obj) {
case Character c:
if (c.charValue() == 7) {
System.out.print("Ding ");
}
if (c.charValue() == 9) {
System.out.print("Tab ");
}
System.out.println("Character");
default:
System.out.println();
}
}
The scope of the declaration of the pattern variable c
includes all the
statements of the statement group, namely the two if
statements and
the println
statement. The scope does not include the statements of the
default
statement group, even though the execution of the first statement
group can fall through the default
switch label and execute these statements.
We forbid the possibility of falling through a case
label that declares a
pattern variable. Consider this erroneous example:
static void test(Object obj) {
switch (obj) {
case Character c:
if (c.charValue() == 7) {
System.out.print("Ding ");
}
if (c.charValue() == 9) {
System.out.print("Tab ");
}
System.out.println("character");
case Integer i: // Compile-time error
System.out.println("An integer " + i);
default:
break;
}
}
If this were allowed and the value of obj
were a
Character
then execution of the switch block could fall through the second
statement group (after case Integer i:
) where the pattern variable i
would
not have been initialized. Allowing execution to fall through a case
label that
declares a pattern variable is therefore a compile-time error.
This is why a switch label consisting of multiple pattern labels, e.g. case Character c: case Integer i: ...
, is not permitted. Similar reasoning applies
to the prohibition of multiple patterns within a single case
label: Neither
case Character c, Integer i: ...
nor case Character c, Integer i -> ...
is
allowed. If such case
labels were allowed then both c
and i
would be in
scope after the colon or arrow, yet only one of them would have been initialized
depending on whether the value of obj
was a Character
or an Integer
.
On the other hand, falling through a label that does not declare a pattern variable is safe, as this example shows:
void test(Object obj) {
switch (obj) {
case String s:
System.out.println("A string");
default:
System.out.println("Done");
}
}
4. Dealing with null
Traditionally, a switch
throws NullPointerException
if the selector
expression evaluates to null
. This is well-understood behavior and we do not
propose to change it for any existing switch
code.
However, given that there is a reasonable and non-exception-bearing semantics
for pattern matching and null
values, we have the opportunity to make pattern
switch
more null
-friendly while remaining compatible with existing switch
semantics.
First, we introduce a new null
case
label . We then lift the blanket
rule that a switch
immediately throws NullPointerException
if the value of
the selector expression is null
. Instead, we inspect the case
labels to
determine the behavior of a switch
:
-
If the selector expression evaluates to
null
then anynull
case label is said to match. If there is no such label associated with the switch block then theswitch
throwsNullPointerException
, as before. -
If the selector expression evaluates to a non-
null
value then we select a matchingcase
label, as normal. If nocase
label matches then anydefault
label is considered to match.
For example, given the declaration below, evaluating test(null)
will print
null!
rather than throw NullPointerException
:
static void test(Object obj) {
switch (obj) {
case null -> System.out.println("null!");
case String s -> System.out.println("String");
default -> System.out.println("Something else");
}
}
This new behavior around null
is as if the compiler automatically
enriches the switch block with a case null
whose body throws
NullPointerException
. In other words, this code:
static void test(Object obj) {
switch (obj) {
case String s -> System.out.println("String: " + s);
case Integer i -> System.out.println("Integer");
default -> System.out.println("default");
}
}
is equivalent to:
static void test(Object obj) {
switch (obj) {
case null -> throw new NullPointerException();
case String s -> System.out.println("String: "+s);
case Integer i -> System.out.println("Integer");
default -> System.out.println("default");
}
}
In both examples, evaluating test(null)
will cause NullPointerException
to
be thrown.
We preserve the intuition from the existing switch
construct that performing a
switch over null
is an exceptional thing to do. The difference in a pattern
switch
is that you have a mechanism to directly handle this case inside the
switch
rather than outside. If you see a null
label in a switch block then
this label will match a null
value. If you do not see a null
label in a
switch block then switching over a null
value will throw
NullPointerException
, as before.
It is also meaningful, and not uncommon, to want to combine a null
case with a
default
. To that end we allow a null
case label to have an optional
default
; for example:
Object obj = ...
switch (obj) {
...
case null, default ->
System.out.println("The rest (including null)");
}
The value of obj
matches this label if either it is the null reference value,
or none of the other case
labels match.
It is a compile-time error for a switch block to have both a null
case
label with a default
and a default
label.
5. Errors
Pattern matching can complete abruptly. For example, when matching a
value against a record pattern, the record’s accessor method can complete
abruptly. In this case, pattern matching is defined to complete abruptly
by throwing a MatchException
. If such a pattern appears as a label in a
switch
then the switch
will also complete abruptly by throwing a
MatchException
.
If a pattern is guarded with a when
expression, and evaluating the when
expression completes abruptly, then the switch
completes abruptly for the same
reason.
If no label in a pattern switch
matches the value of the selector
expression then the switch
completes abruptly by throwing a
MatchException
, since pattern switches must be exhaustive.
For example:
record R(int i){
public int i(){ // accessor method for i
return i / 0;
}
}
static void exampleAnR(R r) {
switch(r) {
case R(var i): System.out.println(i);
}
}
The invocation exampleAnR(new R(42))
causes a MatchException
to be
thrown.
By contrast:
static void example(Object obj) {
switch (obj) {
case R r when (r.i / 0 == 1): System.out.println("It's an R!");
default: break;
}
}
The invocation example(new R(42))
causes an ArithmeticException
to be thrown.
To align with pattern switch
semantics, switch
expressions over an
enum
class now throw MatchException
rather than
IncompatibleClassChangeError
when no switch label applies at run
time. This is a minor incompatible change to the language.
Future work
-
At the moment, a pattern
switch
does not support the primitive typesboolean
,long
,float
, anddouble
. Their utility seems minimal, but support for these could be added. -
We expect that, in the future, general classes will be able to declare deconstruction patterns to specify how they can be matched against. Such deconstruction patterns can be used with a pattern
switch
to yield very succinct code. For example, if we have a hierarchy ofExpr
with subtypes forIntExpr
(containing a singleint
),AddExpr
andMulExpr
(containing twoExpr
s), andNegExpr
(containing a singleExpr
), we can match against anExpr
and act on the specific subtypes all in one step:int eval(Expr n) { return switch (n) { case IntExpr(int i) -> i; case NegExpr(Expr n) -> -eval(n); case AddExpr(Expr left, Expr right) -> eval(left) + eval(right); case MulExpr(Expr left, Expr right) -> eval(left) * eval(right); default -> throw new IllegalStateException(); }; }
Without such pattern matching, expressing ad-hoc polymorphic calculations like this requires using the cumbersome visitor pattern. Pattern matching is generally more transparent and straightforward.
-
It may also be useful to add AND and OR patterns, to allow more expressivity for
case
labels with patterns.
Alternatives
-
Rather than support a pattern
switch
we could instead define a typeswitch
that just supports switching on the type of the selector expression. This feature is simpler to specify and implement but considerably less expressive. -
There are many other syntactic options for guarded pattern labels, such as
p where e
,p if e
, or evenp &&& e
. -
An alternative to guarded pattern labels is to support guarded patterns directly as a special pattern form, e.g.
p && e
. Having experimented with this in previous previews, the resulting ambiguity with boolean expressions has led us to preferwhen
clauses in pattern switches.
Dependencies
This JEP builds on pattern matching for instanceof
(JEP 394) and
also the enhancements offered by switch
expressions (JEP 361). When
the Record Patterns preview feature (JEP 432) is finalized, the resulting
implementation will likely make use of dynamic constants (JEP 309).