Syntax

General rules of Perl 6 syntax

Perl 6 borrows many concepts from human language. Which is not surprising, considering it was designed by a linguist.

It reuses common elements in different contexts, has the notion of nouns (terms) and verbs (operators), is context-sensitive (in the every day sense, not necessarily in the Computer Science interpretation), so a symbol can have a different meaning depending on whether a noun or a verb is expected.

It is also self-clocking, so that the parser can detect most of the common errors and give good error messages.

Lexical conventions

Perl 6 code is Unicode text. Current implementations support UTF-8 as the input encoding.

See also Unicode versus ASCII symbols.

Free form

Perl 6 code is also free-form, in the sense that you are mostly free to chose the amount of whitespace you use, though in some cases, the presence or absence of whitespace carries meaning.

So you can write

if True {
    say "Hello";
}

or

    if True {
say "Hello";
        }

or

if True { say "Hello" }

or even

if True {say "Hello"}

though you can't leave out any of the remaining whitespace.

Unspace

In many places where the compiler would not allow a space you can use any amount of whitespace, as long as it is quoted with a backslash. Unspaces in tokens are not supported. Newlines that are unspaced still count when the compiler produces line numbers. Use cases for unspace are separation of postfix operators and routine argument lists.

sub alignment(+@l{ +@l };
sub long-name-alignment(+@l{ +@l };
alignment\         (1,2,3,4).say;
long-name-alignment(3,5)\   .say;
say Inf+Inf\i;

In this case, our intention was to make the . of both statements, as well as the parentheses, align, so we precede the whitespace used for padding with a \.

Separating statements with semicolons

A Perl 6 program is a list of statements, separated by semicolons ;.

say "Hello";
say "world";

A semicolon after the final statement (or after the final statement inside a block) is optional.

say "Hello";
say "world"
if True {
    say "Hello"
}
say "world"

Implied separator rule (for statements ending in blocks)

Complete statements ending in bare blocks can omit the trailing semicolon, if no additional statements on the same line follow the block's closing curly brace }. This is called the "implied separator rule". For example, you don't need to write a semicolon after an if statement block as seen above, and below.

if True { say "Hello" }
say "world";

However, semicolons are required to separate a block from trailing statements in the same line.

if True { say "Hello" }say "world";
#                     ^^^ this ; is required 

This implied statement separator rule applies in other ways, besides control statements, that could end with a bare block. For example, in combination with the colon : syntax for method calls.

my @names = <Foo Bar Baz>;
my @upper-case-names = @names.map: { .uc }    # OUTPUT: [FOO BAR BAZ] 

For a series of blocks that are part of the same if/elsif/else (or similar) construct, the implied separator rule only applies at the end of the last block of that series. These three are equivalent:

if True { say "Hello" } else { say "Goodbye" }say "world";
#                                            ^^^ this ; is required 
if True { say "Hello" } else { say "Goodbye" } # <- implied statement separator 
say "world";
if True { say "Hello" }   # still in the middle of an if/else statement 
else    { say "Goodbye" } # <- no semicolon required because it ends in a block 
                          #    without trailing statements in the same line 
say "world";

Comments

Comments are parts of the program text which are only intended for human readers; the Perl 6 compilers do not evaluate them as program text.

Comments count as whitespace in places where the absence or presence of whitespace disambiguates possible parses.

Single-line comments

The most common form of comments in Perl 6 starts with a single hash character # and goes until the end of the line.

if $age > 250 {     # catch obvious outliers 
    # this is another comment! 
    die "That doesn't look right"
}

Multi-line / embedded comments

Multi-line and embedded comments start with a hash character, followed by a backtick, and then some opening bracketing character, and end with the matching closing bracketing character. The content can not only span multiple lines, but can also be embedded inline.

if #`( why would I ever write an inline comment here? ) True {
    say "something stupid";
}

These comments can extend multiple lines

#`[
And this is how a multi would work.
That says why we do what we do below.
]
say "No more";

Brackets inside the comment can be nested, so in #`{ a { b } c }, the comment goes until the very end of the string. You may also use more complex brackets/braces, such as #`{{ double-curly-brace }}, which might help disambiguate from nested brackets/braces. You can embed these comments in expressions, as long as you don't insert them in the middle of keywords or identifiers.

Pod comments

Pod syntax can be used for multi-line comments

say "this is code";
 
=begin comment
 
Here are several
lines
of comment
 
=end comment
 
say 'code again';

Identifiers

Identifiers are grammatical building blocks that may be used to give a name to entities/objects such as constants, variables (e.g. Scalars) and routines (e.g. Subs and Methods). In a variable name, any sigil (and twigil) precedes the identifier and does not form a part thereof.

constant c = 299792458;     # identifier "c" names an Int 
my $a = 123;                # identifier "a" in the name "$a" of a Scalar 
sub hello { say "Hello!" }# identifier "hello" names a Sub 

Identifiers come in different forms: ordinary identifiers, extended identifiers, and compound identifiers.

Ordinary identifiers

An ordinary identifier is composed of a leading alphabetic character which may be followed by one or more alphanumeric characters. It may also contain isolated, embedded apostrophes ' and/or hyphens -, provided that the next character is each time alphabetic.

The definitions of "alphabetic" and "alphanumeric" include appropriate Unicode characters. Which characters are "appropriate" depends on the implementation. In the Rakudo/MoarVM Perl 6 implementation alphabetic characters include characters with the Unicode General Category value Letter (L), and the underscore _. Alphanumeric characters additionally include characters with the Unicode General Category value Number, Decimal Digit (Nd).

# valid ordinary identifiers: 
x
_snake_oil
something-longer
with-numbers1234
don't-do-that
piece_of_π
駱駝道                 # "Rakuda-dō", Japanese for "Way of the camel" 
# invalid ordinary identifiers: 
42                    # identifier does not start with alphabetic character 
with-numbers1234-5    # embedded hyphen not followed by alphabetic character 
is-prime?             # question mark is not alphanumeric 
x²                    # superscript 2 is not alphanumeric (explained above) 

Extended identifiers

It is often convenient to have names that contain characters that are not allowed in ordinary identifiers. Use cases include situations where a set of entities shares a common "short" name, but still needs for each of its elements to be identifiable individually. For example, you might use a module whose short name is Dog, while its long name includes its naming authority and version:

Dog:auth<Somebody>:ver<1.0>  # long module names including author and version 
Dog:auth<Somebody>:ver<2.0>
 
use Dog:auth<Somebody>:ver<2.0>;
# Selection of second module causes its full name to be aliased to the 
# short name for the rest of # the lexical scope, allowing a declaration 
# like this. 
my Dog $spot .= new("woof");

Similarly, sets of operators work together in various syntactic categories with names like prefix, infix and postfix. The official names of these operators often contain characters that are excluded from ordinary identifiers. The long name is what constitutes the extended identifier, and includes this syntactic category; the short name will be included in quotes in the definition:

infix:<+>                 # the official name of the operator in $a + $b 
infix:<*>                 # the official name of the operator in $a * $b 
infix:«<=»                # the official name of the operator in $a <= $b 

For all such uses, you can append one or more colon-separated string to an ordinary identifier to create a so-called extended identifier. When appended to an identifier (that is, in postfix position), this colon-separated string generates unique variants of that identifier.

These strings have the form :key<value>, wherein key or value are optional; that is, after the colon that separates it from a regular identifier, there will be a key and/or a quoting bracketing construct such as < >, « » or [' '] which quotes one or more arbitrary characters value: [1]

# exemplary valid extended identifiers: 
postfix:<²>               # the official long name of the operator in $x² 
WOW:That'sAwesome
WOW:That's<<🆒>>
party:sweet<16>
 
# exemplary invalid extended identifiers: 
party:16<sweet>           # 16 is not an ordinary identifier 
party:16sweet
party:!a                  # ...and neither is !a 
party:$a                  # ...nor $a 

In an extended identifier, the postfix string is considered an integral part of the name, so infix:<+> and infix:<-> are two different operators. The bracketing characters used, however, do not count as part of it; only the quoted data matters. So these are all the same name:

infix:<+>
infix:<<+>>
infix:«+»
infix:['+']
infix:('+')

Similarly, all of this works:

my $foo:bar<baz> = 'quux';
say $foo:bar«baz»;                               # OUTPUT: «quux␤» 
my $take-me:<home> = 'Where the glory has no end';
say $take-me:['home'];                           # OUTPUT: «Where [...]␤» 
my $foo:bar<2> = 5;
say $foo:bar(1+1);                               # OUTPUT: «5␤» 

Where an extended identifier comprises two or more colon pairs, their order is generally significant:

my $a:b<c>:d<e> = 100;
my $a:d<e>:b<c> = 200;
say $a:b<c>:d<e>;               # OUTPUT: «100␤», NOT: «200␤» 

An exception to this rule is module versioning; so these identifiers effectively name the same module:

use ThatModule:auth<Somebody>:ver<2.7.18.28.18>
use ThatModule:ver<2.7.18.28.18>:auth<Somebody>

Furthermore, extended identifiers support compile-time interpolation; this requires the use of constants for the interpolation values:

constant $c = 42;  # Constant binds to Int; $-sigil enables interpolation 
my $a:foo<42> = "answer";
say $a:foo«$c»;    # OUTPUT: «answer␤» 

Although quoting bracketing constructs are generally interchangeable in the context of identifiers, they are not identical. In particular, angle brackets < > (which mimic single quote interpolation characteristics) cannot be used for the interpolation of constant names.

constant $what = 'are';
my @we:<are>= <the champions>;
say @we:«$what»;     # OUTPUT: «[the champions]␤» 
say @we:<$what>;
# Compilation error: Variable '@we:<$what>' is not declared 

Compound identifiers

A compound identifier is an identifier that is composed of two or more ordinary and/or extended identifiers that are separated from one another by a double colon ::.

The double colon :: is known as the namespace separator or the package delimiter, which clarifies its semantic function in a name: to force the preceding portion of the name to be considered a package/namespace through which the subsequent portion of the name is to be located:

module MyModule {               # declare a module package 
    our $var = "Hello";         # declare package-scoped variable 
}
say $MyModule::var              # OUTPUT: «Hello␤» 

In the example above, MyModule::var is a compound identifier, composed of the package name identifier MyModule and the identifier part of the variable name var. Altogether $MyModule::var is often referred to as a package-qualified name.

Separating identifiers with double colons causes the rightmost name to be inserted into existing (see above example) or automatically created packages:

my $foo::bar = 1;
say OUR::.keys;           # OUTPUT: «(foo)␤» 
say OUR::foo.HOW          # OUTPUT: «Perl6::Metamodel::PackageHOW.new␤» 

The last lines shows how the foo package was created automatically, as a deposit for variables in that namespace.

The double colon syntax enables runtime interpolation of a string into a package or variable name using ::($expr) where you'd ordinarily put a package or variable name:

my $buz = "quux";
my $bur::quux = 7;
say $bur::($buz);               # OUTPUT: «7␤» 

Term term:<>

You can use term:<> to introduce new terms, which is handy for introducing constants that defy the rules of normal identifiers:

use Testplan 1constant &term:<👍> = &ok.assuming(True);
👍
# OUTPUT: «1..1␤ok 1 - ␤» 

But terms don't have to be constant: you can also use them for functions that don't take any arguments, and force the parser to expect an operator after them. For instance:

sub term:<dice> { (1..6).pick };
say dice + dice;

can print any number between 2 and 12.

If instead we had declared dice as a regular

sub dice() {(1...6).pick }

, the expression dice + dice would be parsed as dice(+(dice())), resulting in an error since sub dice expects zero arguments.

Statements and expressions

Perl 6 programs are made of lists of statements. A special case of a statement is an expression, which returns a value. For example if True { say 42 } is syntactically a statement, but not an expression, whereas 1 + 2 is an expression (and thus also a statement).

The do prefix turns statements into expressions. So while

my $x = if True { 42 };     # Syntax error! 

is an error,

my $x = do if True { 42 };

assigns the return value of the if statement (here 42) to the variable $x.

Terms

Terms are the basic nouns that, optionally together with operators, can form expressions. Examples for terms are variables ($x), barewords such as type names (Int), literals (42), declarations (sub f() { }) and calls (f()).

For example, in the expression 2 * $salary, 2 and $salary are two terms (an integer literal and a variable).

Variables

Variables typically start with a special character called the sigil, and are followed by an identifier. Variables must be declared before you can use them.

# declaration: 
my $number = 21;
# usage: 
say $number * 2;

See the documentation on variables for more details.

Barewords (constants, type names)

Pre-declared identifiers can be terms on their own. Those are typically type names or constants, but also the term self which refers to an object that a method was called on (see objects), and sigilless variables:

say Int;                # OUTPUT: «(Int)␤» 
#   ^^^ type name (built in) 
 
constant answer = 42;
say answer;
#   ^^^^^^ constant 
 
class Foo {
    method type-name {
        self.^name;
      # ^^^^ built-in term 'self' 
    }
}
say Foo.type-name;     # OUTPUT: «Foo␤» 
#   ^^^ type name 

Packages and qualified names

Named entities, such as variables, constants, classes, modules or subs, are part of a namespace. Nested parts of a name use :: to separate the hierarchy. Some examples:

$foo                # simple identifiers 
$Foo::Bar::baz      # compound identifiers separated by :: 
$Foo::($bar)::baz   # compound identifiers that perform interpolations 
Foo::Bar::bob(23)   # function invocation given qualified name 

See the documentation on packages for more details.

Literals

A literal is a representation of a constant value in source code. Perl 6 has literals for several built-in types, like strings, several numeric types, pairs and more.

String literals

String literals are surrounded by quotes:

say 'a string literal';
say "a string literal\nthat interprets escape sequences";

See quoting for many more options, including the escaping quoting q. Perl 6 uses the standard escape characters in literals: \a \b \t \n \f \r \e, with the same meaning as the ASCII escape codes, specified in the design document.

say "🔔\a";# OUTPUT: «🔔␇␤» 

Number literals

Number literals are generally specified in base ten, unless a prefix like 0x (hexadecimal, base 16), 0o (octal, base 8) or 0b (binary, base 2) or an explicit base in adverbial notation like :16<A0> specifies it otherwise. Unlike other programming languages, leading zeros do not indicate base 8; instead a compile-time warning is issued.

In all literal formats, you can use underscores to group digits; they don't carry any semantic information; the following literals all evaluate to the same number:

1000000
1_000_000
10_00000
100_00_00

Int literals

Integers default to signed base-10, but you can use other bases. For details, see Int.

# actually not a literal, but unary - operator applied to numeric literal 2 
-2
12345
0xBEEF      # base 16 
0o755       # base 8 
:3<1201>    # arbitrary base, here base 3 

Rat literals

Rat literals (rationals) are very common, and take the place of decimals or floats in many other languages. Integer division also results in a Rat.

1.0
3.14159
-2.5        # Not actually a literal, but still a Rat 
:3<21.0012> # Base 3 rational 
2/3         # Not actually a literal, but still a Rat 

Num literals

Scientific notation with an integer exponent to base ten after an e produces floating point number:

1e0
6.022e23
1e-9
-2e48
2e2.5       # error 

Complex literals

Complex numbers are written either as an imaginary number (which is just a rational number with postfix i appended), or as a sum of a real and an imaginary number:

1+2i
6.123e5i    # note that this is 6.123e5 * i and not 6.123 * 10 ** (5i) 

Pair literals

Pairs are made of a key and a value, and there are two basic forms for constructing them: key => 'value' and :key('value').

Arrow pairs

Arrow pairs can have an expression or an identifier on the left-hand side:

identifier => 42
"identifier" => 42
('a' ~ 'b'=> 1

Adverbial pairs (colon pairs)

Short forms without explicit values:

my $thing = 42;
:$thing                 # same as  thing => $thing 
:thing                  # same as  thing => True 
:!thing                 # same as  thing => False 

The variable form also works with other sigils, like :&callback or :@elements.

Long forms with explicit values:

:thing($value)              # same as  thing => $value 
:thing<quoted list>         # same as  thing => <quoted list> 
:thing['some''values']    # same as  thing => ['some', 'values'] 
:thing{=> 'b'}            # same as  thing => { a => 'b' } 

Array literals

A pair of square brackets can surround an expression to form an itemized Array literal; typically there is a comma-delimited list inside:

say ['a''b'42].join(' ');   # OUTPUT: «a b 42␤» 
#   ^^^^^^^^^^^^^^ Array constructor 

If the constructor is given a single Iterable, it'll clone and flatten it. If you want an Array with just 1 element that is that Iterable, ensure to use a comma after it:

my @a = 12;
say [@a].perl;  # OUTPUT: «[1, 2]␤» 
say [@a,].perl# OUTPUT: «[[1, 2],]␤» 

The Array constructor does not flatten other types of contents. Use the Slip prefix operator (|) to flatten the needed items:

my @a = 12;
say [@a34].perl;  # OUTPUT: «[[1, 2], 3, 4]␤» 
say [|@a34].perl# OUTPUT: «[1, 2, 3, 4]␤» 

Hash literals

A leading associative sigil and pair of parenthesis %( ) can surround a List of Pairs to form a Hash literal; typically there is a comma-delimited List of Pairs inside. If a non-pair is used, it is assumed to be a key and the next element is the value. Most often this is used with simple arrow pairs.

say %=> 3=> 23:foo:dog<cat>"french""fries" );
# OUTPUT: «a => 3, b => 23, dog => cat, foo => True, french => fries␤» 
 
say %(=> 73foo => "fish").keys.join(" ");   # OUTPUT: «a foo␤» 
#   ^^^^^^^^^^^^^^^^^^^^^^^^^ Hash constructor 

When assigning to a % sigiled variable on the left-hand side, the sigil and parenthesis surrounding the right-hand side Pairs are optional.

my %ages = fred => 23jean => 87ann => 4;

By default, keys in %( ) are forced to strings. To compose a hash with non-string keys, use curly brace delimiters with a colon prefix :{ } :

my $when = :{ (now=> "Instant", (DateTime.now=> "DateTime" };

Note that with objects as keys, you cannot access non-string keys as strings:

say :{ -1 => 410 => 421 => 43 }<0>;  # OUTPUT: «(Any)␤» 
say :{ -1 => 410 => 421 => 43 }{0};  # OUTPUT: «42␤» 

Regex literals

A Regex is declared with slashes like /foo/. Note that this // syntax is shorthand for the full rx// syntax.

/foo/          # Short version 
rx/foo/        # Longer version 
Q :regex /foo/ # Even longer version
 
my $r = /foo/; # Regexes can be assigned to variables

Signature literals

Signatures can be used standalone for pattern matching, in addition to the typical usage in sub and block declarations. A standalone signature is declared starting with a colon:

say "match!" if 5"fish" ~~ :(IntStr); # OUTPUT: «match!␤» 
 
my $sig = :(Int $aStr);
say "match!" if (5"fish"~~ $sig# OUTPUT: «match!␤» 
 
given "foo"42 {
  when :(StrStr{ "This won't match" }
  when :(StrInt $n where $n > 20{ "This will!" }
}

See the Signatures documentation for more about signatures.

Declarations

Variable declaration

my $x;                          # simple lexical variable 
my $x = 7;                      # initialize the variable 
my Int $x = 7;                  # declare the type 
my Int:D $x = 7;                # specify that the value must be defined (not undef) 
my Int $x where { $_ > 3 } = 7# constrain the value based on a function 
my Int $x where * > 3 = 7;      # same constraint, but using Whatever shorthand 

See Variable Declarators and Scope for more details on other scopes (our, has).

Subroutine declaration

# The signature is optional 
sub foo { say "Hello!" }
 
sub say-hello($to-whom{ say "Hello $to-whom!" }

You can also assign subroutines to variables.

my &f = sub { say "Hello!" } # Un-named sub 
my &f = -> { say "Hello!" }  # Lambda style syntax. The & sigil indicates the variable holds a function 
my $f = -> { say "Hello!" }  # Functions can also be put into scalars 

Package, Module, Class, Role, and Grammar declaration

There are several types of package, each declared with a keyword, a name, some optional traits, and a body of subroutines, methods, or rules.

package P { }
 
module M { }
 
class C { }
 
role R { }
 
grammar G { }

You can declare a unit package without explicit curly braces. This must be at the start of the file (preceded only by comments or use statements), and the rest of the file will be taken as being the body of the package.

unit module M;
# ... stuff goes here instead of in {}'s 

Multi-dispatch declaration

See also Multi-dispatch.

Subroutines can be declared with multiple signatures.

multi sub foo() { say "Hello!" }
multi sub foo($name{ say "Hello $name!" }

Inside of a class, you can also declare multi-dispatch methods.

multi method greet { }
multi method greet(Str $name{ }

Subroutine calls

Subroutines are created with the keyword sub followed by an optional name, an optional signature and a code block. Subroutines are lexically scoped, so if a name is specified at the declaration time, the same name can be used in the lexical scope to invoke the subroutine. A subroutine is an instance of type Sub and can be assigned to any container.

foo;   # Invoke the function foo with no arguments 
foo(); # Invoke the function foo with no arguments 
&f();  # Invoke &f, which contains a function 
&f.(); # Same as above, needed to make the following work 
my @functions = ({say 1}{say 2}{say 3});
@functions>>.(); # hyper method call operator 

When declared within a class, a subroutine is named "method": methods are subroutines invoked against an object (i.e., a class instance). Within a method the special variable self contains the object instance (see Methods).

# Method invocation. Object (instance) is $person, method is set-name-age 
$person.set-name-age('jane'98);   # Most common way 
$person.set-name-age: 'jane'98;   # Precedence drop 
set-name-age($person: 'jane'98);  # Invocant marker 
set-name-age $person: 'jane'98;   # Indirect invocation 

For more information see functions.

Precedence drop

In the case of method invocation (i.e., when invoking a subroutine against a class instance) it is possible to apply the precedence drop, identified by a colon : just after the method name and before the argument list. The argument list takes precedence over the method call, that on the other hand "drops" its precedence. In order to better understand consider the following simple example (extra spaces have been added just to align method calls):

my $band = 'Foo Fighters';
say $band.substr03 ) .substr01 ); # F 
say $band.substr: 03   .substr01 ); # Foo 

In the second method call the rightmost substr is applied to "3" and not to the result of the leftmost substr, which on the other hand yields precedence to the rightmost one.

Operators

See Operators for lots of details.

Operators are functions with a more symbol heavy and composable syntax. Like other functions, operators can be multi-dispatch to allow for context-specific usage.

There are five types (arrangements) for operators, each taking either one or two arguments.

++$x           # prefix, operator comes before single input 
5 + 3          # infix, operator is between two inputs 
$x++           # postfix, operator is after single input 
<the blue sky> # circumfix, operator surrounds single input 
%foo<bar>      # postcircumfix, operator comes after first input and surrounds second 

Metaoperators

Operators can be composed. A common example of this is combining an infix (binary) operator with assignment. You can combine assignment with any binary operator.

$x += 5     # Adds 5 to $x, same as $x = $x + 5 
$x min= 3   # Sets $x to the smaller of $x and 3, same as $x = $x min 3 
$x .= child # Equivalent to $x = $x.child 

Wrap an infix operator in [ ] to create a new reduction operator that works on a single list of inputs, resulting in a single value.

say [+] <1 2 3 4 5>;    # OUTPUT: «15␤» 
(((1 + 2+ 3+ 4+ 5 # equivalent expanded version 

Wrap an infix operator in « » (or the ASCII equivalent ) to create a new hyper operator that works pairwise on two lists.

say <1 2 3> «+» <4 5 6> # OUTPUT: «(5 7 9)␤» 

The direction of the arrows indicates what to do when the lists are not the same size.

@a «+« @b # Result is the size of @b, elements from @a will be re-used 
@a »+» @b # Result is the size of @a, elements from @b will be re-used 
@a «+» @b # Result is the size of the biggest input, the smaller one is re-used 
@a »+« @b # Exception if @a and @b are different sizes 

You can also wrap a unary operator with a hyper operator.

say -« <1 2 3> # OUTPUT: «(-1 -2 -3)␤»