Expressions

Chapel provides the following expressions:

expression:
  literal-expression
  variable-expression
  enum-constant-expression
  call-expression
  type-expression
  iteratable-call-expression
  member-access-expression
  new-expression
  query-expression
  cast-expression
  lvalue-expression
  parenthesized-expression
  unary-expression
  binary-expression
  let-expression
  if-expression
  for-expression
  forall-expression
  reduce-expression
  scan-expression
  module-access-expression
  tuple-expression
  tuple-expand-expression
  locale-query-expression
  type-query-expression
  mapped-domain-expression

Individual expressions are defined in the remainder of this chapter and additionally as follows:

Literal Expressions

A literal value for any of the predefined types is a literal expression.

Literal expressions are given by the following syntax:

literal-expression:
  bool-literal
  integer-literal
  real-literal
  imaginary-literal
  string-literal
  bytes-literal
  range-literal
  domain-literal
  array-literal

Literal values for primitive types are described in Literals. Literal range values are described in Range Literals. Literal tuple values are described in Tuple Values. Literal values for domains are described in Rectangular Domain Values and Associative Domain Values. Literal values for arrays are described in Rectangular Array Literals and Associative Array Literals.

Variable Expressions

A use of a variable, constant, parameter, or formal argument, is itself an expression. The syntax of a variable expression is given by:

variable-expression:
  identifier

Enumeration Constant Expression

A use of an enumeration constant is itself an expression. Such a constant must be preceded by the enumeration type name. The syntax of an enumeration constant expression is given by:

enum-constant-expression:
  enum-type . identifier

For an example of using enumeration constants, see Enumerated Types.

Parenthesized Expressions

A parenthesized-expression is an expression that is delimited by parentheses as given by:

parenthesized-expression:
  ( expression )

Such an expression evaluates to the expression. The parentheses are ignored and have only a syntactical effect.

Call Expressions

Functions and function calls are defined in Procedures.

Indexing Expressions

Indexing, for example into arrays, tuples, and domains, has the same syntax as a call expression.

Indexing is performed by an implicit invocation of the this method on the value being indexed, passing the indices as the actual arguments.

Member Access Expressions

Member access expressions provide access to a field or invoke a method of an instance of a class, record, or union. They are defined in Field Accesses and Class Method Calls, respectively.

member-access-expression:
  field-access-expression
  method-call-expression

The Query Expression

A query expression is used to query a type or value within a formal argument type expression. The syntax of a query expression is given by:

query-expression:
  ? identifier[OPT]

Querying is restricted to querying the type of a formal argument, the element type of a formal argument that is an array, the domain of a formal argument that is an array, the size of a primitive type, or a type or parameter field of a formal argument type.

The identifier can be omitted. This is useful for ensuring the genericity of a generic type that defines default values for all of its generic fields when specifying a formal argument as discussed in Formal Arguments of Generic Type.

Example (query.chpl).

The following code defines a generic function where the type of the first argument is queried and stored in the type alias t and the domain of the second argument is queried and stored in the variable D:

proc foo(x: ?t, ref y: [?D] t) {
  for i in D do
    y[i] = x;
}

This allows a generic specification of assigning a particular value to all elements of an array. The value and the elements of the array are constrained to be the same type. This function can be rewritten without query expression as follows:

proc foo(x, ref y: [] x.type) {
  for i in y.domain do
    y[i] = x;
}

There is an expectation that query expressions will be allowed in more places in the future.

Casts

A cast is specified with the following syntax:

cast-expression:
  expression : type-expression

The expression is converted to the specified type. A cast expression invokes the corresponding explicit conversion (Explicit Conversions). A resolution error occurs if no such conversion exists.

LValue Expressions

An lvalue is an expression that can be used on the left-hand side of an assignment statement or on either side of a swap statement, that can be passed to a formal argument of a function that has out, inout or ref intent, or that can be returned by a function with a ref return intent (The Ref Return Intent). Valid lvalue expressions include the following:

  • Variable expressions.

  • Member access expressions.

  • Call expressions of functions with a ref return intent.

  • Indexing expressions.

LValue expressions are given by the following syntax:

lvalue-expression:
  variable-expression
  member-access-expression
  call-expression
  parenthesized-expression

The syntax is less restrictive than the definition above. For example, not all call-expressions are lvalues.

Precedence and Associativity

Operator

Associativity

Use

.
()
[]

left

member access
function call or access
function call or access

new

right

initializer call

owned
shared
borrowed
unmanaged

right

apply management strategy to a class

postfix ?
postfix !

left

compute a nilable class type
assert non-nilable and borrow

:

left

cast

**

right

exponentiation

reduce
scan
dmapped

left

reduction
scan
application of a distribution
prefix !
~

right

logical negation
bitwise negation
*
/
%

left

multiplication
division
modulus
unary +
unary -

right

positive identity
negation
<<
>>

left

left shift
right shift

&

left

bitwise/logical and

^

left

bitwise/logical xor

|

left

bitwise/logical or

+
-

left

addition
subtraction
..
..<
left
left
range initialization
open-interval range initialization
<=
>=
<
>

left

less-than-or-equal-to comparison
greater-than-or-equal-to comparison
less-than comparison
greater-than comparison

== !=

left

equal-to comparison not-equal-to comparison

&&

left

short-circuiting logical and

||

left

short-circuiting logical or

by
#
align

left

range/domain stride application
range count application
range alignment

in

left

forall expression

if then else
forall do
[ ]
for do
sync
atomic

left

conditional expression
forall expression
forall expression
for expression
sync type modifier
atomic type modifier

,

left

comma separated expressions

The above table summarizes operator and expression precedence and associativity. Operators and expressions listed earlier have higher precedence than those listed later.

Rationale.

In general, our operator precedence is based on that of the C family of languages including C++, Java, Perl, and C#. We comment on a few of the differences and unique factors here.

We find that there is tension between the relative precedence of exponentiation, unary minus/plus, and casts. The following three expressions show our intuition for how these expressions should be parenthesized.

-2**4

wants

-(2**4)

-2:uint

wants

(-2):uint

2:uint**4:uint

wants

(2:uint)**(4:uint)

Trying to support all three of these cases results in a circularity—exponentiation wants precedence over unary minus, unary minus wants precedence over casts, and casts want precedence over exponentiation. We chose to break the circularity by making unary minus have a lower precedence. This means that for the second case above:

-2:uint

requires

(-2):uint

We also chose to depart from the C family of languages by making unary plus/minus have lower precedence than binary multiplication, division, and modulus as in Fortran. We have found very few cases that distinguish between these cases. An interesting one is:

const minint = min(int(32));
...-minint/2...

Intuitively, this should result in a positive value, yet C’s precedence rules results in a negative value due to asymmetry in modern integer representations. If we learn of cases that argue in favor of the C approach, we would likely reverse this decision in order to more closely match C.

We were tempted to diverge from the C precedence rules for the binary bitwise operators to make them bind less tightly than comparisons. This would allow us to interpret:

a | b == 0

as

(a | b) == 0

However, given that no other popular modern language has made this change, we felt it unwise to stray from the pack. The typical rationale for the C ordering is to allow these operators to be used as non-short-circuiting logical operations.

In contrast to C, we give bitwise operations a higher precedence than binary addition/subtraction and comparison operators. This enables using the shift operators as shorthand for multiplication/division by powers of 2, and also makes it easier to extract and test a bitmapped field:

(x & MASK) == MASK

as

x & MASK == MASK

a + b * pow(2,y)

as

a * b << y

One final area of note is the precedence of reductions. Two common cases tend to argue for making reductions very low or very high in the precedence table:

max reduce A - min reduce A

wants

(max reduce A) - (min reduce A)

max reduce A * B

wants

max reduce (A * B)

The first statement would require reductions to have a higher precedence than the arithmetic operators while the second would require them to be lower. We opted to make reductions have high precedence due to the argument that they tend to resemble unary operators. Thus, to support our intuition:

max reduce A * B

requires

max reduce (A * B)

This choice also has the (arguably positive) effect of making the unparenthesized version of this statement result in an aggregate value if A and B are both aggregates—the reduction of A results in a scalar which promotes when being multiplied by B, resulting in an aggregate. Our intuition is that users who forget the parentheses will learn of their error at compilation time because the resulting expression is not a scalar as expected.

Operator Expressions

The application of operators to expressions is itself an expression. The syntax of a unary expression is given by:

unary-expression:
  unary-operator expression

unary-operator: one of
  + - ~ !

The syntax of a binary expression is given by:

binary-expression:
  expression binary-operator expression

binary-operator: one of
  + - * / % ** & | ^ << >> && || == != <= >= < > 'by' #

The operators are defined in subsequent sections.

Arithmetic Operators

This section describes the predefined arithmetic operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

For each operator, implicit conversions are applied to the operands of an operator such that they are compatible with one of the function forms listed, those listed earlier in the list being given preference. If no compatible implicit conversions exist, then a compile-time error occurs. In these cases, an explicit cast is required.

Unary Plus Operators

The unary plus operators are predefined as follows:

operator +(a: int(8)): int(8)
operator +(a: int(16)): int(16)
operator +(a: int(32)): int(32)
operator +(a: int(64)): int(64)

operator +(a: uint(8)): uint(8)
operator +(a: uint(16)): uint(16)
operator +(a: uint(32)): uint(32)
operator +(a: uint(64)): uint(64)

operator +(a: real(32)): real(32)
operator +(a: real(64)): real(64)

operator +(a: imag(32)): imag(32)
operator +(a: imag(64)): imag(64)

operator +(a: complex(64)): complex(64)
operator +(a: complex(128)): complex(128)

For each of these definitions, the result is the value of the operand.

Unary Minus Operators

The unary minus operators are predefined as follows:

operator -(a: int(8)): int(8)
operator -(a: int(16)): int(16)
operator -(a: int(32)): int(32)
operator -(a: int(64)): int(64)

operator -(a: real(32)): real(32)
operator -(a: real(64)): real(64)

operator -(a: imag(32)): imag(32)
operator -(a: imag(64)): imag(64)

operator -(a: complex(64)): complex(64)
operator -(a: complex(128)): complex(128)

For each of these definitions that return a value, the result is the negation of the value of the operand. For integral types, this corresponds to subtracting the value from zero. For real and imaginary types, this corresponds to inverting the sign. For complex types, this corresponds to inverting the signs of both the real and imaginary parts.

Negating a value of type uint or uint(w) for any width will result in a compilation error.

Addition Operators

The addition operators are predefined as follows:

operator +(a: int(8), b: int(8)): int(8)
operator +(a: int(16), b: int(16)): int(16)
operator +(a: int(32), b: int(32)): int(32)
operator +(a: int(64), b: int(64)): int(64)

operator +(a: uint(8), b: uint(8)): uint(8)
operator +(a: uint(16), b: uint(16)): uint(16)
operator +(a: uint(32), b: uint(32)): uint(32)
operator +(a: uint(64), b: uint(64)): uint(64)

operator +(a: real(32), b: real(32)): real(32)
operator +(a: real(64), b: real(64)): real(64)

operator +(a: imag(32), b: imag(32)): imag(32)
operator +(a: imag(64), b: imag(64)): imag(64)

operator +(a: complex(64), b: complex(64)): complex(64)
operator +(a: complex(128), b: complex(128)): complex(128)

operator +(a: real(32), b: imag(32)): complex(64)
operator +(a: imag(32), b: real(32)): complex(64)
operator +(a: real(64), b: imag(64)): complex(128)
operator +(a: imag(64), b: real(64)): complex(128)

operator +(a: real(32), b: complex(64)): complex(64)
operator +(a: complex(64), b: real(32)): complex(64)
operator +(a: real(64), b: complex(128)): complex(128)
operator +(a: complex(128), b: real(64)): complex(128)

operator +(a: imag(32), b: complex(64)): complex(64)
operator +(a: complex(64), b: imag(32)): complex(64)
operator +(a: imag(64), b: complex(128)): complex(128)
operator +(a: complex(128), b: imag(64)): complex(128)

For each of these definitions that return a value, the result is the sum of the two operands.

When adding signed and unsigned types of the same width (e.g. myInt32 + myUint32), the addition will be done with the unsigned type (e.g. uint(32)).

Addition over a value of real type and a value of imaginary type produces a value of complex type. Addition of values of complex type and either real or imaginary types also produces a value of complex type.

Subtraction Operators

The subtraction operators are predefined as follows:

operator -(a: int(8), b: int(8)): int(8)
operator -(a: int(16), b: int(16)): int(16)
operator -(a: int(32), b: int(32)): int(32)
operator -(a: int(64), b: int(64)): int(64)

operator -(a: uint(8), b: uint(8)): uint(8)
operator -(a: uint(16), b: uint(16)): uint(16)
operator -(a: uint(32), b: uint(32)): uint(32)
operator -(a: uint(64), b: uint(64)): uint(64)

operator -(a: real(32), b: real(32)): real(32)
operator -(a: real(64), b: real(64)): real(64)

operator -(a: imag(32), b: imag(32)): imag(32)
operator -(a: imag(64), b: imag(64)): imag(64)

operator -(a: complex(64), b: complex(64)): complex(64)
operator -(a: complex(128), b: complex(128)): complex(128)

operator -(a: real(32), b: imag(32)): complex(64)
operator -(a: imag(32), b: real(32)): complex(64)
operator -(a: real(64), b: imag(64)): complex(128)
operator -(a: imag(64), b: real(64)): complex(128)

operator -(a: real(32), b: complex(64)): complex(64)
operator -(a: complex(64), b: real(32)): complex(64)
operator -(a: real(64), b: complex(128)): complex(128)
operator -(a: complex(128), b: real(64)): complex(128)

operator -(a: imag(32), b: complex(64)): complex(64)
operator -(a: complex(64), b: imag(32)): complex(64)
operator -(a: imag(64), b: complex(128)): complex(128)
operator -(a: complex(128), b: imag(64)): complex(128)

For each of these definitions that return a value, the result is the value obtained by subtracting the second operand from the first operand.

Subtraction of a value of real type from a value of imaginary type, and vice versa, produces a value of complex type. Subtraction of values of complex type from either real or imaginary types, and vice versa, also produces a value of complex type.

Multiplication Operators

The multiplication operators are predefined as follows:

operator *(a: int(8), b: int(8)): int(8)
operator *(a: int(16), b: int(16)): int(16)
operator *(a: int(32), b: int(32)): int(32)
operator *(a: int(64), b: int(64)): int(64)

operator *(a: uint(8), b: uint(8)): uint(8)
operator *(a: uint(16), b: uint(16)): uint(16)
operator *(a: uint(32), b: uint(32)): uint(32)
operator *(a: uint(64), b: uint(64)): uint(64)

operator *(a: real(32), b: real(32)): real(32)
operator *(a: real(64), b: real(64)): real(64)

operator *(a: imag(32), b: imag(32)): real(32)
operator *(a: imag(64), b: imag(64)): real(64)

operator *(a: complex(64), b: complex(64)): complex(64)
operator *(a: complex(128), b: complex(128)): complex(128)

operator *(a: real(32), b: imag(32)): imag(32)
operator *(a: imag(32), b: real(32)): imag(32)
operator *(a: real(64), b: imag(64)): imag(64)
operator *(a: imag(64), b: real(64)): imag(64)

operator *(a: real(32), b: complex(64)): complex(64)
operator *(a: complex(64), b: real(32)): complex(64)
operator *(a: real(64), b: complex(128)): complex(128)
operator *(a: complex(128), b: real(64)): complex(128)

operator *(a: imag(32), b: complex(64)): complex(64)
operator *(a: complex(64), b: imag(32)): complex(64)
operator *(a: imag(64), b: complex(128)): complex(128)
operator *(a: complex(128), b: imag(64)): complex(128)

For each of these definitions that return a value, the result is the product of the two operands.

Multiplication of values of imaginary type produces a value of real type. Multiplication over a value of real type and a value of imaginary type produces a value of imaginary type. Multiplication of values of complex type and either real or imaginary types produces a value of complex type.

Division Operators

The division operators are predefined as follows:

operator /(a: int(8), b: int(8)): int(8)
operator /(a: int(16), b: int(16)): int(16)
operator /(a: int(32), b: int(32)): int(32)
operator /(a: int(64), b: int(64)): int(64)

operator /(a: uint(8), b: uint(8)): uint(8)
operator /(a: uint(16), b: uint(16)): uint(16)
operator /(a: uint(32), b: uint(32)): uint(32)
operator /(a: uint(64), b: uint(64)): uint(64)

operator /(a: real(32), b: real(32)): real(32)
operator /(a: real(64), b: real(64)): real(64)

operator /(a: imag(32), b: imag(32)): real(32)
operator /(a: imag(64), b: imag(64)): real(64)

operator /(a: complex(64), b: complex(64)): complex(64)
operator /(a: complex(128), b: complex(128)): complex(128)

operator /(a: real(32), b: imag(32)): imag(32)
operator /(a: imag(32), b: real(32)): imag(32)
operator /(a: real(64), b: imag(64)): imag(64)
operator /(a: imag(64), b: real(64)): imag(64)

operator /(a: real(32), b: complex(64)): complex(64)
operator /(a: complex(64), b: real(32)): complex(64)
operator /(a: real(64), b: complex(128)): complex(128)
operator /(a: complex(128), b: real(64)): complex(128)

operator /(a: imag(32), b: complex(64)): complex(64)
operator /(a: complex(64), b: imag(32)): complex(64)
operator /(a: imag(64), b: complex(128)): complex(128)
operator /(a: complex(128), b: imag(64)): complex(128)

For each of these definitions that return a value, the result is the quotient of the two operands.

Division of values of imaginary type produces a value of real type. Division over a value of real type and a value of imaginary type produces a value of imaginary type. Division of values of complex type and either real or imaginary types produces a value of complex type.

When the operands are integers, the result (quotient) is also an integer. If b does not divide a exactly, then there are two candidate quotients \(q1\) and \(q2\) such that \(b * q1\) and \(b * q2\) are the two multiples of b closest to a. The integer result \(q\) is the candidate quotient which lies closest to zero.

Modulus Operators

The modulus operators are predefined as follows:

operator %(a: int(8), b: int(8)): int(8)
operator %(a: int(16), b: int(16)): int(16)
operator %(a: int(32), b: int(32)): int(32)
operator %(a: int(64), b: int(64)): int(64)

operator %(a: uint(8), b: uint(8)): uint(8)
operator %(a: uint(16), b: uint(16)): uint(16)
operator %(a: uint(32), b: uint(32)): uint(32)
operator %(a: uint(64), b: uint(64)): uint(64)

For each of these definitions that return a value, the result is the remainder when the first operand is divided by the second operand.

The sign of the result is the same as the sign of the dividend a, and the magnitude of the result is always smaller than that of the divisor b. For integer operands, the % and / operators are related by the following identity:

var q = a / b;
var r = a % b;
writeln(q * b + r == a);    // true

Note

This does not have the same behavior as mod when b is negative.

There is an expectation that the predefined modulus operators will be extended to handle real, imaginary, and complex types in the future.

Exponentiation Operators

The exponentiation operators are predefined as follows:

operator **(a: int(8), b: int(8)): int(8)
operator **(a: int(16), b: int(16)): int(16)
operator **(a: int(32), b: int(32)): int(32)
operator **(a: int(64), b: int(64)): int(64)

operator **(a: uint(8), b: uint(8)): uint(8)
operator **(a: uint(16), b: uint(16)): uint(16)
operator **(a: uint(32), b: uint(32)): uint(32)
operator **(a: uint(64), b: uint(64)): uint(64)

operator **(a: real(32), b: real(32)): real(32)
operator **(a: real(64), b: real(64)): real(64)

For each of these definitions that return a value, the result is the value of the first operand raised to the power of the second operand.

There is an expectation that the predefined exponentiation operators will be extended to handle imaginary and complex types in the future.

Bitwise Operators

This section describes the predefined bitwise operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

Bitwise Complement Operators

The bitwise complement operators are predefined as follows:

operator ~(a: int(8)): int(8)
operator ~(a: int(16)): int(16)
operator ~(a: int(32)): int(32)
operator ~(a: int(64)): int(64)

operator ~(a: uint(8)): uint(8)
operator ~(a: uint(16)): uint(16)
operator ~(a: uint(32)): uint(32)
operator ~(a: uint(64)): uint(64)

For each of these definitions, the result is the bitwise complement of the operand.

Bitwise And Operators

The bitwise and operators are predefined as follows:

operator &(a: bool, b: bool): bool

operator &(a: int(?w), b: int(w)): int(w)
operator &(a: uint(?w), b: uint(w)): uint(w)

operator &(a: int(?w), b: uint(w)): uint(w)
operator &(a: uint(?w), b: int(w)): uint(w)

For each of these definitions, the result is computed by applying the logical and operation to the bits of the operands.

Chapel allows mixing signed and unsigned integers of the same size when passing them as arguments to bitwise and. In the mixed case the result is of the same size as the arguments and is unsigned. No run-time error is issued, even if the apparent sign changes as the required conversions are performed.

Rationale.

The mathematical meaning of integer arguments is discarded when they are passed to bitwise operators. Instead the arguments are treated simply as bit vectors. The bit-vector meaning is preserved when converting between signed and unsigned of the same size. The choice of unsigned over signed as the result type in the mixed case reflects the semantics of standard C.

Bitwise Or Operators

The bitwise or operators are predefined as follows:

operator |(a: bool, b: bool): bool

operator |(a: int(?w), b: int(w)): int(w)
operator |(a: uint(?w), b: uint(w)): uint(w)

operator |(a: int(?w), b: uint(w)): uint(w)
operator |(a: uint(?w), b: int(w)): uint(w)

For each of these definitions, the result is computed by applying the logical or operation to the bits of the operands. Chapel allows mixing signed and unsigned integers of the same size when passing them as arguments to bitwise or. No run-time error is issued, even if the apparent sign changes as the required conversions are performed.

Rationale.

The same as for bitwise and (Bitwise And Operators).

Bitwise Xor Operators

The bitwise xor operators are predefined as follows:

operator ^(a: bool, b: bool): bool

operator ^(a: int(?w), b: int(w)): int(w)
operator ^(a: uint(?w), b: uint(w)): uint(w)

operator ^(a: int(?w), b: uint(w)): uint(w)
operator ^(a: uint(?w), b: int(w)): uint(w)

For each of these definitions, the result is computed by applying the XOR operation to the bits of the operands. Chapel allows mixing signed and unsigned integers of the same size when passing them as arguments to bitwise xor. No run-time error is issued, even if the apparent sign changes as the required conversions are performed.

Rationale.

The same as for bitwise and (Bitwise And Operators).

Shift Operators

This section describes the predefined shift operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

The shift operators are predefined as follows:

operator <<(a: int(8), b): int(8)
operator <<(a: int(16), b): int(16)
operator <<(a: int(32), b): int(32)
operator <<(a: int(64), b): int(64)

operator <<(a: uint(8), b): uint(8)
operator <<(a: uint(16), b): uint(16)
operator <<(a: uint(32), b): uint(32)
operator <<(a: uint(64), b): uint(64)

operator >>(a: int(8), b): int(8)
operator >>(a: int(16), b): int(16)
operator >>(a: int(32), b): int(32)
operator >>(a: int(64), b): int(64)

operator >>(a: uint(8), b): uint(8)
operator >>(a: uint(16), b): uint(16)
operator >>(a: uint(32), b): uint(32)
operator >>(a: uint(64), b): uint(64)

The type of the second actual argument must be any integral type.

The << operator shifts the bits of a left by the integer b. The new low-order bits are set to zero.

The >> operator shifts the bits of a right by the integer b. When a is negative, the new high-order bits are set to one; otherwise the new high-order bits are set to zero.

The value of b must be non-negative.

The value of b must be less than the number of bits in a.

Logical Operators

This section describes the predefined logical operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

The Logical Negation Operator

The logical negation operator is predefined for booleans and integers as follows:

operator !(a: bool): bool
operator !(a: int(?w)): bool
operator !(a: uint(?w)): bool

For the boolean form, the result is the logical negation of the operand. For the integer forms, the result is true if the operand is zero and false otherwise.

The Logical And Operator

The logical and operator is predefined over bool type. It returns true if both operands evaluate to true; otherwise it returns false. If the first operand evaluates to false, the second operand is not evaluated and the result is false.

The logical and operator over expressions a and b given by

a && b

is evaluated as the expression

if isTrue(a) then isTrue(b) else false

The function isTrue is predefined over bool type as follows:

proc isTrue(a:bool) do return a;

Overloading the logical and operator over other types is accomplished by overloading the isTrue function over other types.

The Logical Or Operator

The logical or operator is predefined over bool type. It returns true if either operand evaluate to true; otherwise it returns false. If the first operand evaluates to true, the second operand is not evaluated and the result is true.

The logical or operator over expressions a and b given by

a || b

is evaluated as the expression

if isTrue(a) then true else isTrue(b)

The function isTrue is predefined over bool type as described in The Logical And Operator. Overloading the logical or operator over other types is accomplished by overloading the isTrue function over other types.

Relational Operators

This section describes the predefined relational operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

Ordered Comparison Operators

The “less than” comparison operators are predefined over numeric types as follows:

operator <(a: int(8), b: int(8)): bool
operator <(a: int(16), b: int(16)): bool
operator <(a: int(32), b: int(32)): bool
operator <(a: int(64), b: int(64)): bool

operator <(a: uint(8), b: uint(8)): bool
operator <(a: uint(16), b: uint(16)): bool
operator <(a: uint(32), b: uint(32)): bool
operator <(a: uint(64), b: uint(64)): bool

operator <(a: int(64), b: uint(64)): bool
operator <(a: uint(64), b: int(64)): bool

operator <(a: real(32), b: real(32)): bool
operator <(a: real(64), b: real(64)): bool

The result of a < b is true if a is less than b; otherwise the result is false.

The “greater than” comparison operators are predefined over numeric types as follows:

operator >(a: int(8), b: int(8)): bool
operator >(a: int(16), b: int(16)): bool
operator >(a: int(32), b: int(32)): bool
operator >(a: int(64), b: int(64)): bool

operator >(a: uint(8), b: uint(8)): bool
operator >(a: uint(16), b: uint(16)): bool
operator >(a: uint(32), b: uint(32)): bool
operator >(a: uint(64), b: uint(64)): bool

operator >(a: int(64), b: uint(64)): bool
operator >(a: uint(64), b: int(64)): bool

operator >(a: real(32), b: real(32)): bool
operator >(a: real(64), b: real(64)): bool

The result of a > b is true if a is greater than b; otherwise the result is false.

The “less than or equal to” comparison operators are predefined over numeric types as follows:

operator <=(a: int(8), b: int(8)): bool
operator <=(a: int(16), b: int(16)): bool
operator <=(a: int(32), b: int(32)): bool
operator <=(a: int(64), b: int(64)): bool

operator <=(a: uint(8), b: uint(8)): bool
operator <=(a: uint(16), b: uint(16)): bool
operator <=(a: uint(32), b: uint(32)): bool
operator <=(a: uint(64), b: uint(64)): bool

operator <=(a: int(64), b: uint(64)): bool
operator <=(a: uint(64), b: int(64)): bool

operator <=(a: real(32), b: real(32)): bool
operator <=(a: real(64), b: real(64)): bool

The result of a <= b is true if a is less than or equal to b; otherwise the result is false.

The “greater than or equal to” comparison operators are predefined over numeric types as follows:

operator >=(a: int(8), b: int(8)): bool
operator >=(a: int(16), b: int(16)): bool
operator >=(a: int(32), b: int(32)): bool
operator >=(a: int(64), b: int(64)): bool

operator >=(a: uint(8), b: uint(8)): bool
operator >=(a: uint(16), b: uint(16)): bool
operator >=(a: uint(32), b: uint(32)): bool
operator >=(a: uint(64), b: uint(64)): bool

operator >=(a: int(64), b: uint(64)): bool
operator >=(a: uint(64), b: int(64)): bool

operator >=(a: real(32), b: real(32)): bool
operator >=(a: real(64), b: real(64)): bool

The result of a >= b is true if a is greater than or equal to b; otherwise the result is false.

The ordered comparison operators are predefined over strings as follows:

operator <(a: string, b: string): bool
operator >(a: string, b: string): bool
operator <=(a: string, b: string): bool
operator >=(a: string, b: string): bool

Comparisons between strings are defined based on the ordering of the character set used to represent the string, which is applied elementwise to the string’s characters in order.

Equality Comparison Operators

The equality comparison operators == and != are predefined over bool and the numeric types as follows:

operator ==(a: int(8), b: int(8)): bool
operator ==(a: int(16), b: int(16)): bool
operator ==(a: int(32), b: int(32)): bool
operator ==(a: int(64), b: int(64)): bool

operator ==(a: uint(8), b: uint(8)): bool
operator ==(a: uint(16), b: uint(16)): bool
operator ==(a: uint(32), b: uint(32)): bool
operator ==(a: uint(64), b: uint(64)): bool

operator ==(a: int(64), b: uint(64)): bool
operator ==(a: uint(64), b: int(64)): bool

operator ==(a: real(32), b: real(32)): bool
operator ==(a: real(64), b: real(64)): bool

operator ==(a: imag(32), b: imag(32)): bool
operator ==(a: imag(64), b: imag(64)): bool

operator ==(a: complex(64), b: complex(64)): bool
operator ==(a: complex(128), b: complex(128)): bool

operator !=(a: int(8), b: int(8)): bool
operator !=(a: int(16), b: int(16)): bool
operator !=(a: int(32), b: int(32)): bool
operator !=(a: int(64), b: int(64)): bool

operator !=(a: uint(8), b: uint(8)): bool
operator !=(a: uint(16), b: uint(16)): bool
operator !=(a: uint(32), b: uint(32)): bool
operator !=(a: uint(64), b: uint(64)): bool

operator !=(a: int(64), b: uint(64)): bool
operator !=(a: uint(64), b: int(64)): bool

operator !=(a: real(32), b: real(32)): bool
operator !=(a: real(64), b: real(64)): bool

operator !=(a: imag(32), b: imag(32)): bool
operator !=(a: imag(64), b: imag(64)): bool

operator !=(a: complex(64), b: complex(64)): bool
operator !=(a: complex(128), b: complex(128)): bool

The result of a == b is true if a and b contain the same value; otherwise the result is false. The result of a != b is equivalent to !(a == b).

The equality comparison operators are predefined over classes as follows:

operator ==(a: RootClass, b: RootClass): bool
operator !=(a: RootClass, b: RootClass): bool

The result of a == b is true if a and b reference the same storage location; otherwise the result is false. The result of a != b is equivalent to !(a == b).

Default equality comparison operators are generated for records if the user does not define them. These operators are described in Default Comparison Operators.

The equality comparison operators are predefined over strings as follows:

operator ==(a: string, b: string): bool
operator !=(a: string, b: string): bool

The result of a == b is true if the sequence of characters in a matches exactly the sequence of characters in b; otherwise the result is false. The result of a != b is equivalent to !(a == b).

Class Operators

The keywords owned, shared, borrowed, and unmanaged act as a prefix unary operator when specifying the management strategy for a class type. See Class Types.

The unary postfix operator ? results in the nilable variant of a class type. See Nilable Class Types.

The unary postfix operator ! asserts that the receiver is not storing nil and borrows from it. See Nilable Class Types.

Miscellaneous Operators

This section describes several miscellaneous operators. These operators can be redefined over different types using operator overloading (Function and Operator Overloading).

The String Concatenation Operator

The string concatenation operator + is predefined for string arguments and returns a new string that is the concatenation of its arguments:

operator +(s0: string, s1: string): string

Example (string-concat.chpl).

The code:

var x: string = "hi";
var y: string = " there";
var z = x + y;

will cause z to be a new string containing the value "hi there".

The By Operator

The operator by is predefined on ranges and rectangular domains. It is described in By Operator for ranges and Domain Striding for domains.

The Align Operator

The operator align is predefined on ranges and rectangular domains. It is described in Align Operator for ranges and Domain Alignment for domains.

The Range Count Operator

The operator # is predefined on ranges. It is described in  Count Operator.

Let Expressions

A let expression allows variables to be declared at the expression level and used within that expression. The syntax of a let expression is given by:

let-expression:
  'let' variable-declaration-list 'in' expression

The scope of the variables is the let-expression.

Example (let.chpl).

Let expressions are useful for defining variables in the context of an expression. In the code

let x: real = a*b, y = x*x in 1/y

the value determined by a*b is computed and converted to type real if it is not already a real. The square of the real is then stored in y and the result of the expression is the reciprocal of that value.

Conditional Expressions

A conditional expression is given by the following syntax:

if-expression:
  'if' expression 'then' expression 'else' expression
  'if' expression 'then' expression

The conditional expression is evaluated in two steps. First, the expression following the if keyword is evaluated. Then, if the expression evaluated to true, the expression following the then keyword is evaluated and taken to be the value of this expression. Otherwise, the expression following the else keyword is evaluated and taken to be the value of this expression. In both cases, the unselected expression is not evaluated.

The ‘else’ clause can be omitted only when the conditional expression is nested immediately inside a for or forall expression. Such an expression is used to filter predicates as described in Filtering Predicates in For Expressions and Filtering Predicates in Forall Expressions, respectively.

Example (condexp.chpl).

This example shows how if-then-else can be used in a context in which an expression is expected. The code

writehalf(8);
writehalf(21);
writehalf(1000);

proc writehalf(i: int) {
  var half = if (i % 2) then i/2 +1 else i/2;
  writeln("Half of ",i," is ",half);
}

produces the output

Half of 8 is 4
Half of 21 is 11
Half of 1000 is 500

For Expressions

A for expression is given by the following syntax:

for-expression:
  'for' index-var-declaration 'in' iteratable-expression 'do' expression
  'for' iteratable-expression 'do' expression

A for expression is an iterator that executes a for loop (The For Loop), evaluates the body expression on each iteration of the loop, and yields each resulting value.

When a for expression is used to initialize a variable, such as

var X = for iterableExpression() do computeValue();

the variable will be inferred to have an array type. The array’s domain is defined by the iterable-expression following the same rules as for promotion, both in the regular case Promotion and in the zippered case Zippered Promotion.

Filtering Predicates in For Expressions

A conditional expression that is immediately enclosed in a for expression and does not require an else clause filters the iterations of the for expression. The iterations for which the condition does not hold are not reflected in the result of the for expression.

When a for expression with a filtering predicate is captured into a variable, the resulting array has a 0-based one-dimensional domain.

Example (yieldPredicates.chpl).

The code

var A = for i in 1..10 do if i % 3 != 0 then i;

declares an array A that is initialized to the integers between 1 and 10 that are not divisible by 3.