Syntax-Expressions

• Operators
• Members and dereferences
• Calls
• Indexes
• Metaexpressions

Operators

Most built-in operators follow C/C++ syntax, though sometimes alternatives are available. Many operators can be overloaded. That is, structs and classes can often freely define what happens when you use a particular operator on them.

Currently, operator precedence, from highest to lowest, are:

Precedence	Associativity	Operator	Overloadable	Description
N/A	None	(...)	No	Tuple, subexpression
		[...]	No	List
		{...}	No	Dictionary (NIY)
		|...|	Yes	Absolute value (NIY)
		||...||	Yes	Length (NIY)
1	Left	.	See note 1	Member reference
		?.	See note 1	Null-safe member reference
		??.	See note 1	Auto-allocating member reference
	Postfix	(...)	Yes (NIY)	Call
		[...]	See note 2	Index
2	Prefix	@	No	Metaexpression
		*	No	Expansion (NIY)
3	Postfix	<...>	No	Generic type argument
4	Prefix	(:...)	No	Static typecast
		ref	No	Reference of
		sizeof	No	(Static) Size of
		typeid	Yes	(Dynamic) Type ID of
5	Left	as	No	Dynamic typecast
6	Postfix	!	Yes	Factorial (NIY)
		~	Yes	Inverse (NIY)
		'	Yes	Transpose (NIY)
		++	Yes	Post-increment
		--	Yes	Post-decrement
7	Right	^	Yes	Power (NIY)
8	Prefix	+	Yes	Copy
		-	Yes	Negation
		~	Yes	Bitwise NOT
		++	Yes	Pre-increment
		--	Yes	Pre-decrement
9	Left	*	Yes	Multiplication (matrix product)
		**	Yes	Multiplication (dot product) (NIY)
		><	Yes	Multiplication (cross product) (NIY)
		/	Yes	Right division
		\	Yes	Left division (NIY)
		%	Yes	Remainder
		*.	Yes	Elementwise multiplication (NIY)
		/.	Yes	Elementwise right division (NIY)
		\.	Yes	Elementwise left division (NIY)
		div	Yes	Euclidean division (NIY)
		mod	Yes	Modulo (NIY)
10	Left	+	Yes	Addition
		-	Yes	Subtraction
11	Left	<<	Yes	Bit shift left
		>>	Yes	Bit shift right
12	Left	&	Yes	Bitwise AND
13	Left	<>	Yes	Bitwise XOR
14	Left	|	Yes	Bitwise OR
15	Left	??	No	Null coalescing
16	None	::	No	Slice
		..	No	Range
17	None	==	Yes	Equal to
		!=	Yes	Not equal to
		===	No	Alias of
		!==	No	Not alias of
		<	Yes	Less than
		>	Yes	Greater than
		<=	Yes	Less than or equal to
		>=	Yes	Greater than or equal to
		<=>	Yes	Three-way comparison
		~=	Yes	Pattern match (NIY)
		in	Yes	Contained in
		not in	Yes	Not contained in
		is	See note 3	(Dynamic) Derived type of
		is not	See note 3	(Dynamic) Not derived type of
18	Prefix	!, not	Yes	Logical NOT
19	Left	&&, and	Yes	Logical AND
20	Left	||, or	Yes	Logical OR
21	Right	? :	No	Conditional
22	Right	->	No	Anonymous function (NIY)
23	Right	<<<	Yes	Receive (NIY)
		>>>	Yes	Send (NIY)
		:=	See note 4	Inline assignment
24	None	throw	No	Throw exception

Notes:

1. The member reference operator itself cannot be overloaded, but references to certain member names can be converted to method calls by defining the members as properties. It's under consideration whether to allow redirecting references to unknown member names to some fallback method (where compilation error would otherwise result).
2. References to indexes can be converted to method calls by defining indexers. Indexers can be overloaded.
3. While the is operator can be overloaded, is null and is not null is exempt from any overloading, and can thus be used as a reliable way of checking if a reference is null.
4. Overloads of ordinary assignments also apply to inline assignments.

Members and dereferences

The period, ., is used to refer to named members of objects, classes, namespaces, etc. For example, if universe is a reference to an object with an instance variable called answer, then you can say

universe.answer = 42

Following a reference in order to access the actual object is called dereferencing. Referencing and dereferencing is automatic, but depends on the types used (thus, typecasts affect the way this is done).

However, if an object reference is null (doesn't refer to anything), and the member is not static, then doing the above unconditionally will probably cause problems (most likely crash). If you're dealing with nullable references, then you'll usually have to explicitly check for null, but some operators are available to make the job easier.

• To explicitly check for null, the recommended way is to use either is null or === null, since these constructs cannot be overloaded.
• The null coalescing operator ?? (as seen in C#) allows you to substitute another object when the one you want is null. In the expression a ?? b, a is used if it's non-null, otherwise b is used.
• The null-safe member reference operator ?. (as seen in Groovy) only follows the reference if it is not null. If it is null, then the expression (data access or method call) evaluates to the default value for the inferred expression type, usually null or zero. In the expression car?.drive(), the drive method is only called if car isn't a null reference.
• As a special case, if the null-safe member reference operator is used on an allocator, then it is invoked if the reference is null. In the expression car?.new(), a new car object is allocated and assigned to car if it was null before this.
• The auto-allocating member reference operator ??. will, if the reference is null, allocate a new object of the required type and assign it to the reference. Then it follows the reference as normal. In the expression car??.drive(), a new car object is allocated (using the default allocator) and assigned to car if it was null, and then the drive method is called, regardless of whether car was previously null or not.

Calls

To call a function, method, or delegate, enclose its arguments in parentheses after its name (or other reference to it). Even if there are no arguments, the parentheses are necessary. For example,

exit()

calls the exit function with no arguments.

f = fac(2)

calls the fac function with one integer argument, and assigns its return value to the variable called f.

Indexes

Many objects that can contain other objects, such as arrays, support indexed references as a means of referring to a particular contained object. In this case, enclose the index in square brackets after the container reference. For example, if prime_list were an array of prime numbers, then

prime = prime_list[3]

would give you the 3rd prime in this list.

Metaexpressions

Metaexpressions are evaluated at compile time, and can refer to information available to the compiler that wouldn't normally be available at runtime. A simple example is the source code filename (similar to the __FILE__ macro in C/C++). Metaexpressions are useful when used in metamethods, or in implicit parameters to ordinary methods.

The form @expression results in a regular parser token for the compiler. If the expression creates a string, then it is converted to an identifier (a reference to a named object).

The form @#expression results in a string. If the expression refers to an object, then the name of that object is converted to a string.

To manipulate an object's name as a string inside the metaexpression itself, refer to the object's name property, e.g., owner.name. This is useful for constructing new identifiers, e.g., @(owner.name + "Child").

Special information that can be used in metaexpressions include:

Name	Type	Description
function	Function object	Enclosing function or method
owner	Class object	Enclosing class or struct
super	Class object	Base class or struct
module	String	Module name
file	String	Source file name
line	Integer	Source line number