Skip to content
Permalink
master
Go to file
 
 
Cannot retrieve contributors at this time
3403 lines (2518 sloc) 100 KB

Introduction

The opcode set is dynamic. This document is about core NQP opcodes. For Raku opcodes added by the rakudo compiler see docs/ops.markdown in the rakudo repository. They are of the form nqp::p6* (following the historical naming of Perl 6).

The tool tools/find-undocumented-ops.raku can be used to find undocumented opcodes. For generating an abstract tree that includes opcodes, see docs/qast.markdown.

Table of Contents

NQP Opcodes

Opcodes (ops) are used both directly when writing NQP, and during code generation in QAST nodes.

When invoking them directly, you'll need to prefix them with nqp::, e.g.

nqp::mul_i(6,9);

The ops are listed below by type. Each entry shows the name of the op, its variants, and their arguments and types, and may provide a short description. Some opcodes differ only by argument types - in that case, they are listed under their common name (e.g. mul), with each of their variants (e.g. mul_i, mul_n) together with a single description.

Opcode variants may contain a type suffix, which usually indicates:

  • _i argument is a native int
  • _u argument is an unsigned int
  • _n argument is a native float
  • _s argument is a native string
  • _o argument is an object
  • _b argument is a code block
  • _I argument is a Big Integer

They may also have a numeric suffix, which typically indicates the number of arguments required. (Note that some opcodes contain an underscore not being used to indicate a type suffix.)

In opcode signatures below, we use the following types, which may not correspond directly to NQP types.

  • int - native int
  • num - native float
  • str - native string
  • Int - BigInt
  • Mu - any NQP or VM object
  • Mu:T - a type object, e.g. Int
  • Exception - an Exception object
  • Handle - an I/O Handle object
  • Iterable - something iterable
  • Context - a Context object
  • LexPad - a Context object
  • @ - this sigil indicates an array parameter
  • % - this sigil indicates a hash parameter
  • ... - indicates variable args are accepted

VM-specific opcodes are denoted with a jvm, e.g. on the same line as the header. No annotation indicates this opcode should be supported on all nqp backends.

Some individual opcodes may be marked with Internal or Deprecated. Both of these indicate the opcodes are not intended to be used. Deprecated opcodes will eventually be removed from NQP. Internal opcodes are typically used at compile time to replace opcodes that take a variable number of arguments.

The opcodes are grouped into the following categories:

Arithmetic Opcodes

abs

  • abs_i(int $i --> int)
  • abs_n(num $n --> num)
  • abs_I(Int $i, Mu:T $type --> Int)

Return the absolute value of a number. _I variant returns an object of the given type.

add

  • add_i(int $l, int $r --> int)
  • add_n(num $l, num $r --> num)
  • add_I(Int $l, Int $r, Mu:T $type --> Int)

Add two numbers together, returning the result. _I variant returns an object of the given type.

div

  • div_i(int $l, int $r --> int)
  • div_n(num $l, num $r --> num)
  • div_I(Int $l, Int $r, Mu:T $type --> Int)
  • div_In(Int $l, Int $r --> num)

Divide $l by $r, returning the result. _I variant returns an object of the given type. The _In variant returns a native num, using a scale of 309, and a rounding mode equivalent to Java's ROUND_HALF_UP.

gcd

  • gcd_i(int $l, int $r --> int)
  • gcd_I(Int $l, Int $r, Mu:T $type --> Int)

Return the greatest common multiple of two numbers. _I variant returns an object of the given type.

lcm

  • lcm_i(int $l, int $r --> int)
  • lcm_I(Int $l, Int $r, Mu:T $type --> Int)

Return the lowest common multiple of two numbers. _I variant returns an object of the given type.

mod

  • mod_i(int $l, int $r --> int)
  • mod_n(num $l, num $r --> num)
  • mod_I(Int $l, Int $r, Mu:T $type --> Int)

Return the modulus of $l by $r. _I variant returns an object of the given type.

mul

  • mul_i(int $l, int $r --> int)
  • mul_n(num $l, num $r --> num)
  • mul_I(Int $l, Int $r, Mu:T $type --> Int)

Multiply two numbers, returning the result. _I variant returns an object of the given type.

neg

  • neg_i(int $i --> int)
  • neg_n(num $n --> num)
  • neg_I(Int $i, Mu:T $type --> Int)

Return the negative of a number. _I variant returns an object of the given type.

sub

  • sub_i(int $l, int $r --> int)
  • sub_n(num $l, num $r --> num)
  • sub_I(Int $l, Int $r, Mu:T $type --> Int)

Subtract $r from $l, returning the result. _I variant returns an object of the given type.

Numeric Opcodes

base

  • base_I(Int $i, int $radix --> str)

Returns a string representing the integer $i in base $radix

ceil

  • ceil_n(num $n --> num)

Return the ceiling of a number.

exp

  • exp_n(num $exponent --> num)

Return the value of e raised to $exponent.

floor

  • floor_n(num $n --> num)

Return the floor of a number.

inf

  • inf(--> num)

Return infinity.

log_n

  • log_n(num $n --> num)

Return the natural logarithm (base 𝑒) of a number.

expmod

  • expmod_I(Int $base, Int $exponent, Int $modulus, Mu:T $type --> Int)

Return a bigint that is $base raised to $exponent modulus $modulus. _I variant returns an object of the given type.

nan

  • nan(--> num)

Return NaN.

neginf

  • neginf(--> num)

Return negative infinity.

pow

  • pow_i(int $base, int $exponent --> int)
  • pow_n(num $base, num $exponent --> num)
  • pow_I(Int $base, Int $exponent, Mu:T $type_num, Mu:T $type_bigint --> Int)

Return the value of $base raised to $exponent; _I variant returns an object of $type_num for negative exponents, and of type $type_bigint for positive exponents.

rand

  • rand_n(num $n --> num)
  • rand_i(int $i, Mu:T $type --> int) moar
  • rand_I(Int $i, Mu:T $type --> Int)

Returns a psuedo-random bigint up to the value of the given number. _I variant returns an object of the given type.

srand

  • srand(int $n)

Sets and returns seed number for nqp::rand_* variants. Decimal numbers will be silently truncated, nqp::srand(1) and nqp::srand(1.1) are the same so always pass nqp::srand an integer.

sqrt

  • sqrt_n(num $n--> num)

Trigonometric Opcodes

Each opcode corresponds directly to the trigonometric function of the same name. h indicates a hyperbolic variant.

asec

  • asec_n(num $n --> num)

asin

  • asin_n(num $n --> num)

acos

  • acos_n(num $n --> num)

atan

  • atan_n(num $n --> num)

atan2

  • atan2_n(num $l, num $r --> num))

cos

  • cos_n(num $n --> num))

cosh

  • cosh_n(num $n --> num))

sin

  • sin_n(num $n --> num))

sinh

  • sinh_n(num $n --> num))

sec

  • sec_n(num $n --> num))

sech

  • sech_n(num $n --> num))

tan

  • tan_n(num $n --> num))

tanh

  • tanh_n(num $n --> num))

Relational / Logic Opcodes

cmp

  • cmp_i(int $l, int $r --> int)
  • cmp_n(num $l, num $r --> int)
  • cmp_s(str $l, str $r --> int)
  • cmp_I(Int $l, Int $r --> int)

Compare two values, returns -1 if $l is greater than $r, 0 if they are equal, and 1 if $r is greater than $l.

eqat

  • eqat(str $haystack, str $needle, int $pos --> int)

Return 1 if the string $haystack has the string $needle at position $pos, otherwise return 0.

eqatic

  • eqatic(str haystack, str $needle, int $pos --> int) Case-insensitive eqat

eqatim moar js

  • eqatim(str haystack, str $needle, int $pos --> int) Ignore-mark eqat, NFD decomposes and matches the base codepoint

Example: eqat("á", "a", 0) → 1

eqaticim moar js

  • eqaticim(str haystack, str $needle, int $pos --> int) Case-insensitive and ignore-mark eqat

falsey

  • falsey($a --> int)

Return 0 if the parameter has a truthy value, 1 otherwise.

iseq

  • iseq_i(int $l, int $r --> int)
  • iseq_n(num $l, num $r --> int)
  • iseq_s(str $l, str $r --> int)
  • iseq_I(Int $l, Int $r --> int)
  • iseq_snfg(str $l, str $r --> int) js

Return 1 if the two parameters are equal, 0 otherwise.

iseq_snfg is a JS specific opcode that first normalizes string arguments to NFC.

isgt

  • isgt_i(int $l, int $r --> int)
  • isgt_n(num $l, num $r --> int)
  • isgt_s(str $l, str $r --> int)
  • isgt_I(Int $l, Int $r --> int)

Return 1 if the two parameters are equal if $l is greater than $r, otherwise 0.

isge

  • isge_i(int $l, int $r --> int)
  • isge_n(num $l, num $r --> int)
  • isge_s(str $l, str $r --> int)
  • isge_I(Int $l, Int $r --> int)

Return 1 if $l is greater than or equal to $r, otherwise 0.

islt

  • islt_i(int $l, int $r --> int)
  • islt_n(num $l, num $r --> int)
  • islt_s(str $l, str $r --> int)
  • islt_I(Int $l, Int $r --> int)

Return 1 if $l is less than $r, otherwise 0.

isle

  • isle_i(int $l, int $r --> int)
  • isle_n(num $l, num $r --> int)
  • isle_s(str $l, str $r --> int)
  • isle_I(Int $l, Int $r --> int)

Return 1 if $l is less than or equal to $r, otherwise 0.

isne

  • isne_i(int $l, int $r --> int)
  • isne_n(num $l, num $r --> int)
  • isne_s(str $l, str $r --> int)
  • isne_I(Int $l, Int $r --> int)
  • isne_snfg(str $l, str $r --> int) js

Return 1 if the two parameters are not equal, otherwise 0.

isne_snfg is a JS specific opcode that first normalizes string arguments to NFC.

not_i

  • not_i(int $val --> int)

Return 1 if $val is 0, 0 otherwise.

Array Opcodes

atpos

  • atpos(@arr, int $i --> Mu)
  • atpos_i(@arr, int $i --> int)
  • atpos_n(@arr, int $i --> num)
  • atpos_s(@arr, int $i --> str)

Return whatever is bound to @arr at position $i.

atposnd

  • atposnd(@arr, @indices --> Mu)
  • atposnd_i(@arr, @indices --> int)
  • atposnd_n(@arr, @indices --> num)
  • atposnd_s(@arr, @indices --> str)

Return whatever is bound to the n-dimensional array @arr at @indices, where @indices is a 1-dimensional array of index values.

atpos2d

  • atpos2d(@arr, int $i, int $j --> Mu)
  • atpos2d_i(@arr, int $i, int $j --> int)
  • atpos2d_n(@arr, int $i, int $j --> num)
  • atpos2d_s(@arr, int $i, int $j --> str)

Return whatever is bound to the 2-dimensional array @arr at position $i, $j.

atpos3d

  • atpos3d(@arr, int $i, int $j, int $k --> Mu)
  • atpos3d_i(@arr, int $i, int $j, int $k --> int)
  • atpos3d_n(@arr, int $i, int $j, int $k --> num)
  • atpos3d_s(@arr, int $i, int $j, int $k --> str)

Return whatever is bound to the 3-dimensional array @arr at position $i, $j, $k.

bindpos

  • bindpos(@arr, int $i, Mu $v --> Mu)
  • bindpos_i(@arr, int $i, int $v --> int)
  • bindpos_n(@arr, int $i, num $v --> num)
  • bindpos_s(@arr, int $i, str $v --> str)

Bind $v to @arr at position $i and return $v.

bindposnd

  • bindposnd(@arr, @indices, Mu $v --> Mu)
  • bindposnd_i(@arr, @indices, int $v --> int)
  • bindposnd_n(@arr, @indices, num $v --> num)
  • bindposnd_s(@arr, @indices, str $v --> str)

Bind $v to @arr at the position specified by @indices and return $v. @indices is a 1-dimensional array of index values.

bindpos2d

  • bindpos2d(@arr, int $i, int $j, Mu $v --> Mu)
  • bindpos2d_i(@arr, int $i, int $j, int $v --> int)
  • bindpos2d_n(@arr, int $i, int $j, num $v --> num)
  • bindpos2d_s(@arr, int $i, int $j, str $v --> str)

Bind $v to the 2-dimensional arrray @arr at position $i, $j and return $v.

bindpos3d

  • bindpos3d(@arr, int $i, int $j, int $k, Mu $v --> Mu)
  • bindpos3d_i(@arr, int $i, int $j, int $k, int $v --> int)
  • bindpos3d_n(@arr, int $i, int $j, int $k, num $v --> num)
  • bindpos3d_s(@arr, int $i, int $j, int $k, str $v --> str)

Bind $v to the 3-dimensional arrray @arr at position $i, $j, $k and return $v.

atposref

  • atposref(@arr, int $idx --> Mu) js
  • atposref_i(@arr, int $idx --> int)
  • atposref_n(@arr, int $idx --> num)
  • atposref_s(@arr, int $idx --> str)

Returns a container (of type IntPosRef, NumPosRef, or StrPosRef) that you can assign to or read from which will directly access @arr at index $idx.

elems

  • elems(@arr --> int)
  • elems(%hash --> int)

Return the number of elements in @arr, or the number of keys in %hash.

existspos

  • existspos(@arr, int $i --> int)

Return 1 if anything is bound to @arr at position $i, 0 otherwise.

list

  • list(... --> Mu)
  • list_i(... --> Mu)
  • list_n(... --> Mu)
  • list_s(... --> Mu)
  • list_b(... --> Mu) moar

Create a list of the given parameters. If no arguments are passed, an empty list is created. If a typed variant is used, the parameters are coerced to the appropriate type.

push

  • push(@arr, Mu $v --> Mu)
  • push_i(Array int @arr, int $v --> int)
  • push_n(Array num @arr, num $v --> num)
  • push_s(Array str @arr, str $v --> str)

"Push $v onto the end of @arr." Bind $v to @arr at the position at the end of @arr, i.e., the position that is just after the last position of @arr that has been bound to.

Return value is not currently defined.

pop

  • pop(@arr --> Mu)
  • pop_i(@arr --> int)
  • pop_n(@arr --> num)
  • pop_s(@arr --> str)

"Pop the last value off the end of @arr." Return the value of @arr at its last bound position, and unbind @arr at that position.

setelems

  • setelems(@arr, int $i --> @arr)

Set the size of @arr to $i elements. If less than the current size, any elements after the new last position are unbound. If greater, the empty elements at the end are bound with potentially VM specific null entries.

shift

  • shift(@arr --> Mu)
  • shift_i(@arr --> int)
  • shift_n(@arr --> num)
  • shift_s(@arr --> str)
  • shift_o(@arr --> obj) moar

"Shift the first value from the beginning of @arr." Return the value of @arr at index 0, unbind @arr at index 0, and move all other bindings of @arr to the index one below what they were previously bound to.

slice

  • slice(@arr, int $start_pos, int $end_pos --> @copy)

Copy the elements in @arr starting at $start_pos and ending at $end_pos and return the resulting list. If $start_pos or $end_pos is -n it will translate into the nth position relative to the end of the list.

my @a := 'a', 'b', 'c';
print($_ ~ ', ') for nqp::slice(@a, 0, -2);

# OUTPUT: «a, b»

Will throw an exception if either $start_pos or $end_pos is out-of-bounds (after translation).

splice

  • splice(@arr, @from, int $offset, int $count --> @arr)

Remove the elements in @arr starting at $offset, for $count positions. Replace them with all the elements from @from.

unshift

  • unshift(@arr, Mu $v --> $v)
  • unshift_i(@arr, int $v --> $v)
  • unshift_n(@arr, num $v --> $v)
  • unshift_s(@arr, str $v --> $v)

"Shift $v into the beginning of @arr." Bind $v to @arr at index 0, move all other bindings of @arr to the index one above what they were previously bound to. Return the $v on JVM.

iterator

  • iterator()

Returns an iterator object to iterate over a list's items. For example:

my $list := nqp::list('a', 'b', 'c');
my $iter := nqp::iterator($list);

while $iter {
    say(nqp::shift($iter));
}

You can also use nqp::iterator() to iterate over a hash's key-value pairs.

Hash Opcodes

atkey

  • atkey(%hash, str $key --> Mu)
  • atkey_i(%hash, str $key --> int)
  • atkey_n(%hash, str $key --> num)
  • atkey_s(%hash, str $key --> str)
  • atkey_u(%hash, str $key --> uint) moar

Return the value of %hash at key $key.

Note, there's no bindkey_u yet since at the moment atkey_u is only used for getting values from the lexpad

bindkey

  • bindkey(%hash, str $key, Mu $v --> $v)
  • bindkey_i(%hash, str $key, int $v --> $v)
  • bindkey_n(%hash, str $key, num $v --> $v)
  • bindkey_s(%hash, str $key, str $v --> $v)

Bind key $key of %hash to $v and return $v.

existskey

  • existskey(%hash, str $key --> int)

Return 1 if %hash has key $key bound to something, otherwise 0.

deletekey

  • deletekey(%hash, str $key --> %hash)

Delete the given key from %hash.

hash

  • hash(... --> Mu)

Return a hash. The first argument is a key, the second its value, and so on. Be sure to pass an even number of arguments, a VM specific error may occur otherwise.

iterkey

  • iterkey_s($pair --> str)

Returns the key associated with the given key-value pair. For example:

for %hash {
    say(nqp::iterkey_s($_), ' => ', nqp::iterval($_));
}

Or alternately:

my $iter := nqp::iterator(%hash);
while $iter {
    my $pair := nqp::shift($iter);
    say(nqp::iterkey_s($pair), ' => ', nqp::iterval($pair));
}

iterval

  • iterval($pair --> Mu)

Returns the value associated with the given key-value pair.

Coercion opcodes

coerce_* opcodes do lower level conversion between int, num and str. intify, numify, strify respectively try to use the .Int, .Num or .Str method. If absent rely on lower level conversions.

coerce_in moar

  • coerce_in(int --> num)

coerce_ni moar

  • coerce_ni(num --> int)

coerce_is

  • coerce_is(int --> str)

coerce_ns moar

  • coerce_ns(num --> str)

coerce_sn moar

  • coerce_sn(str --> num)

coerce_si

  • coerce_si(str --> int)

intify moar

  • intify($o --> int)

numify moar

  • numify($o --> num)

strify moar

  • strify($o --> str)

String Opcodes

chars

  • chars(str $str --> int)
  • charsnfg(str $str --> int) js

Return the number of characters in the string.

chr

  • chr(int $codepoint --> str)

Given a unicode codepoint, return a str containing its character. Will throw an exception on invalid codepoints.

codepointfromname

  • codepointfromname(str $name --> int)

Returns the codepoint for the given unicode character name, or -1 if no match was found.

codes

  • codes(str $str --> int)

Return the number of codepoints in the string.

concat

  • concat(str $l, str $r --> str)

Return a string that is the concatenation of the two passed in strings.

decode

  • decode($buffer, str $encoding --> str)

Returns an (NFG) string resulting from decoding the specified buffer assuming the specified encoding.

decodetocodes moar

  • decodetocodes($buffer, str $encoding, int $normalization, $codes)

Decodes the bytes in the specified buffer using the provided encoding. Applies normalization as requested (must be one of the nqp::const::NORMALIZE_* values; use nqp::const::NORMALIZE_NONE to apply no normalization). Places the code points into $codes, which should be some VMArray holding 32-bit integers.

encode

  • encode(str $string, str $encoding, $buffer)

Encodes an (NFG) string into the specified encoding, writing into the buffer provided. The data written is normalized according to NFC.

encodefromcodes moar

  • encodefromcodes($codes, str $encoding, $buffer)

Takes a 32-bit integer array of Unicode codepoints, encodes them using the chosen encoding, and writes them into the buffer. No normalization is applied.

encodenorm

  • encode(str $string, str $encoding, int $normalization, $buffer)

Encodes an (NFG) string into the specified encoding, writing into the buffer provided. The data written is normalized according to the normalization value passed (which must be one of the nqp::const::NORMALIZE_* values). Specifying NORMALIZE_NONE is equivalent to NFC.

escape

  • escape(str $str --> str)

Given a string, return an escaped version that replaces the following characters with their escaped equivalents: "\", "\b", "\n", "\r", "\t", "\f", """, "\a", and "\e".

fc

  • fc(str $str --> str)

Returns a Unicode "fold case" operation copy of string, suitable for doing caseless string comparisons.

findcclass

  • findcclass(int $class, str $str, int $i, int $count --> int)

Search the string starting at the $ith character, for $count characters. Return the position of the first character that is of the specified class (nqp::const::CCLASS_*). If no characters match, return the position of the first character after the given range, or the length of the string, whichever is smaller.

findnotcclass

  • findnotcclass(int $class, str $str, int $i, int $count --> int)

Search the string starting at the $ith character, for $count characters. Return the position of the first character that is not of the specified class (nqp::const::CCLASS_*). If no characters match, return the position of the first character after the given range, or the length of the string, whichever is smaller.

flip

  • flip(str $str --> str)

Return a string with the characters of $string in reverse order.

index

  • index(str $haystack, str $needle --> int)
  • index(str $haystack, str $needle, int $pos --> int)

Return the position in $haystack at which $needle appears, or -1 if $needle does not appear. Begin searching at position $pos if specified, or at 0, otherwise.

  • indexfrom(str $haystack, str $needle, int $pos) moar jvm Internal

index is converted to this internal opcode by the compiler.

indexic moar

  • indexic(str $haystack, str $needle, int $pos --> int)

This op has the same arguments and functionality as nqp::index, except it is case-insensitive. For now we only have it under MoarVM, but the plan is to support it on other platforms as well.

On MoarVM uses proper Unicode foldcase type comparison.

indexim moar

  • indexim(str $haystack, str $needle, int $pos --> int)

Like index but decomposes and matches against the base character.

Example: indexim("bcá", "a", 0) → 2

indexicim moar

  • indexicim(str $haystack, str $needle, int $pos)

Ignorecase and ignoremark index

iscclass

  • iscclass(int $class, str $str, int $i --> int)
  • iscclassnfg(int $class, str $str, int $i --> int) js

Return 1 if the $ith character of $str is of the specified class, (nqp::const::CCLASS_*), 0 otherwise.

join

  • join(str $delim, @arr --> str)

Joins the separate strings of @arr into a single string with fields separated by the value of EXPR, and returns that new string.

lc

  • lc(str $str --> str)

Return lowercase copy of string.

normalizecodes

  • normalizecodes($codes-in, int $normalization, $codes-out)

Takes the codepoints in $codes-in, applies the specified normalization, and places the result into the $codes-out array. Both arrays of codepoints must be 32-bit integer arrays.

numify

use nqp; say nqp::numify(nqp::unbox_s("42e0"));

Converts a string to one of the numeric types. Only used in HLL space, as in pure nqp +"42e0" performs the same numification. The op can be used, for example, to parse a Num out of a string.

ord

  • ord(str $str --> int)
  • ord(str $str, int $i --> int)

Return the unicode codepoint of the first character in $str, or at the $ith character, if it's specified.

  • ordat(str $str, int $i --> int) Internal
  • ordfirst(str $str --> int) moar jvm Internal

ord is converted to these internal opcodes by the compiler.

ordbaseat

  • ordbaseat(str $str, int $pos --> int)

Returns the Unicode codepoint which is the base (non extend/prepend character at that position). If it is a degenerate, and contains no base character, it then returns the first codepoint in that grapheme.

radix

  • radix(int $radix, str $str, int $pos, int $flags --> Mu)
  • radix_I(int $radix, str $str, int $pos, int $flags, Mu:T $type --> Mu)

Convert string $str into a number starting at offset $pos and using radix $radix. The result of the conversion returns an array with

out[0] = converted value
out[1] = $radix ** $number-of-digits-converted
out[2] = offset after consuming digits, -1 if no digits consumed

The opcode skips single underscores between pairs of digits, per the Raku specification.

The $flags is a bitmask that modifies the parse and/or result:

0x01: negate the result (useful if you've already parsed a minus)
0x02: parse a leading +/- and negate the result on -
0x04: parse trailing zeroes but do not include in result
      (for parsing values after a decimal point)

replace

  • replace(str $str, int $offset, int $count, str $replacement --> str)

Return a copy of $str where the characters starting at $offset for $count characters have been replaced with the $replacement string.

rindex

  • rindex(str $haystack, str $needle --> int)
  • rindex(str $haystack, str $needle, int $pos --> int)

Searching backwards through the $haystack, return the position at which $needle appears, or -1 if it does not. Begin searching at $pos if specified, otherwise start from the last position.

  • rindexfrom(str $haystack, str $needle, int $pos) moar jvm Internal
  • rindexfromend(str $haystack, str $needle) jvm Internal

rindex is converted to this internal opcode by the compiler.

split

  • split(str $delimiter, str $string --> Mu)

Splits the string $string into an array of elements; these elements are the substrings between delimiters in the original string.

If the original string begins or ends with the delimiter, the resulting array will begin or end with an empty element.

sprintf

  • sprintf(str $pattern, @values --> str)

Returns a string formatted by the printf conventions similar to Perl 5 / C. Machine sized numeric types, their limits and therefore overflows are not implemented though.

sprintfdirectives

  • sprintfdirectives(str $pattern)

This takes the same pattern as sprintf does, and computes the needed value-count that sprintf would have to provide.

sprintfaddargumenthandler

  • sprintfaddargumenthandler(Mu $handler)

Lets you register a handler-instance that supports the sprintf op when it has to numify custom types. This handler has to provide two methods, mine and int. mine gets the the value in question and returns true if this handler is in charge for this type, false otherwise. The method int does the conversion for patterns like %d.

my class MyHandler {
    method mine($x) { $x ~~ MyType }
    method int($x) { $x.Int }
}

strfromcodes

  • strfromcodes($codes --> str)

Returns an (NFG) string built from the specified codepoints, which must be provided as a 32-bit integer array.

strfromname

  • strfromname(str $name --> str)

Like codepointfromname except it returns a string instead of a codepoint. This function is able to return not just Unicode codepoints by name, but also Unicode Named Sequences, including Emoji Sequences and Emoji ZWJ Sequences and Name Aliases.

In addition it is also case-insensitive, unlike codepointfromname

See these links for a full list of Named Sequences, Emoji Sequences, Emoji ZWJ Sequences and Name Aliases.

strtocodes

  • strtocodes(str $str, int $normalization, $codes)

Takes an NFG string, and places the codepoints from it into the codes array, which must be a 32-bit integer array. Applies the specified normalization, specified as one of the nqp::const::NORMALIZE_* values; NORMALIZE_NONE is not allowed.

substr

  • substr(str $str, int $position --> str)
  • substr(str $str, int $position, int $length --> str)
  • substr_s(str $str, int $position --> str) moar
  • substr_s(str $str, int $position, int $length --> str) moar
  • substr2(str $str, int $position) jvm Internal
  • substr3(str $str, int $position, int $length) jvm Internal
  • substrnfg(str $str, int $position --> str) js
  • substrnfg(str $str, int $position, int $length --> str) js

Return the portion of the string starting at the given position. If $length is specified, only return that many characters. The numbered variants required the args specified - the unnumbered version may use either signature.

tc

  • tc(str $str --> str)

Return titlecase copy of string.

tclc

  • tclc(str $str --> str)

Return copy of string with first character titlecased, and remaining characters lowercased.

uc

  • uc(str $str --> str)

Return uppercase copy of string.

unicmp_s moar js

  • unicmp_s(str $str1, str Str2, int $mode, int $iso639, int $iso3166 --> int)

Compares strings using the Unicode Collation Algorithm (UCA).

Parameters:

$str1, $str2  # strings to compare
$mode         # collation mode (bitmask)
$iso639       # ISO 639 Language code
$iso3166      # ISO 3166 Country code

The collation mode defines whether we use Primary, Secondary, Tertiary and/or Quaternary sorting.

Compares two strings, using the Unicode Collation Algorithm Return values: 0 The strings are identical for the collation levels requested -1/1 String a is less than string b/String a is greater than string b

collation_mode acts like a bitfield. Each of primary, secondary and tertiary collation levels can be either: disabled, enabled, reversed. In the table below, where + designates sorting normal direction and

  • indicates reversed sorting for that collation level.
 Collation level | bitfield value
        Primary+ |   1
        Primary- |   2
      Secondary+ |   4
      Secondary- |   8
       Tertiary+ |  16
       Tertiary- |  32
     Quaternary+ |  64
     Quaternary- | 128

While the Primary, Secondary and Tertiary mean different things for different scripts, for the Latin script used in English they mostly correspond with Primary being Alphabetic, Secondary being Diacritics and Tertiary being Case.

Setting 0 for language and country will collate all scripts according to their own distinctions for Primary, Secondary, and Tertiary, although it will not take into account certain languages.

For example, some language based differences in collation:

For more information see Unicode TR10.

*** Note ***

  • Currently only language and country insensitive sorting methods are implemented.

x

  • x(str $str, int $count --> str)

Return a new string containing $count copies of $str.

Unicode Property Opcodes

getuniname

  • getuniname(int $codepoint --> str)

Translate a codepoint to its Unicode name.

getuniprop_int moar

  • getuniprop_int(int $codepoint, int $propcode --> int)

Uses the table found by unipropcode to look up an integer property value for a given codepoint. Note that many properties that are officially numeric are really stored as strings, and if you try to use this op on such a property, you'll get a meaningless position in an enum table instead of the value you want.

getuniprop_str

  • getuniprop_str(int $codepoint, int $propcode --> str)

Same thing, but fetches a string property value.

getuniprop_bool moar

  • getuniprop_bool(int $codepoint, int $propcode --> int)

Same thing, but fetches a boolean property value.

matchuniprop moar

  • matchuniprop(int $codepoint, int $propcode, int $pvalcode --> int)

Looks up a codepoint property and return 1 if it matches the pval, 0 otherwise. The propcode and pvalcode may be looked up with the opcodes above. (Note that you can use the property value name (e.g. Nd) for both lookups.)

unipropcode

  • unipropcode(str $propname --> int)

Translates a property name to the backend's property code. This is not distinct across backends and is expected to change over time. For the most part only useful for calling getuniprop_int, get_uniprop_str or get_uniprop_bool or comparing whether two unicode property names resolve to the same propcode, for example 'Alpha', 'alpha', 'alphabetic' and 'Alphabetic' should return the same property code.

unipvalcode moar

  • unipvalcode(int $propcode, str $propname --> int)

Looks up a property name in its property category, and returns which table within that category to use.

hasuniprop moar

  • hasuniprop(str $string, int offset, int propcode, int pvalcode --> int)

Checks if the string has a specific property value at a specific offset. Requires both the propcode and the pvalcode to work.

VM-Provided Streaming Decoder Opcodes

decoderconfigure

  • decoderconfigure(Decoder $dec, str $encoding, VMHash $config)

Configures the decoder with an encoding. The $config hash parameter is currently unused, and an empty hash or an nqp::null should be passed.

decodersetlineseps

  • decodersetlineseps(Decoder $dec, VMArray $separators)

Sets the line separators to be used for line-based reading. It should be a string array (nqp::list_s(...)).

decoderaddbytes

  • decoderaddbytes(Decoder $dec, VMArray $blob)

Adds bytes to the decoder's internal buffer. Must have VMArray REPR, and must have elements of type int8 or uint8.

decodertakechars

  • decodertakechars(Decoder $dec, int $num-chars --> str)

Returns an NFG string consisting of $num-chars graphemes, provided that many are available after decoding. If less than $num-chars characters can be decoded, then nqp::null_s will be returned. Note that a decoded codepoint at the end of a byte buffer may not be available as a character if the encoding allows the next character to be a combining character.

decodertakeallchars

  • decodertakeallchars(Decoder $dec --> str)

Decodes all remaining undecoded bytes, and flushes the normalization buffer. Returns an NFG string consisting of the decoded characters. This is suitable to use when the end of a stream of bytes to decode has been reached (for example, EOF when reading a file).

decodertakeavailablechars

  • decodertakeavailablechars(Decoder $dec -->int)

Decodes all remaining undecoded bytes. Returns an NFG string consisting of the decoded characters. Does not flush the normalization buffer. This is suitable when performing streaming decoding, and a later byte buffer may provide a combining character.

decodertakeline

  • decodertakeline(Decoder $dec, int $chomp, int $incomplete-ok --> str)

Decodes bytes until a line separator is reached, or all bytes have been decoded. If $incomplete-ok is zero and the separator was not found, then nqp::null_s will be returned. (Thus, $incomplete-ok is appropriate only when knowing that the end of the stream has been reached.) If $chomp is non-zero, then the separator--if present--will not be included in the resulting string.

decoderbytesavailable

  • decoderbytesavailable(Decoder $dec --> int)

Returns the number of undecoded bytes available inside of the decoder. This is useful in the case that chunks of the input should also be pulled out as bytes, and may be useful for doing tuning or pre-fetching in various other cases. Note that the result does not include bytes that were decoded but have not yet been taken as characters, or that were decoded to code points that are still in the normalization buffer. Thus the result is only accurate before reading any chars or after decodertakechars or after decodertakeline with $incomplete-ok passed a non-zero value.

decodertakebytes

  • decodertakebytes(Decoder $dec, VMArray $blob_type, int $bytes)

Takes up to $bytes bytes from the decode stream's undecoded buffer, makes an instance of the $blob_type, and places the bytes in it. The same set of caveats about decoded-but-untaken bytes in decoderbytesavailable apply.

decoderempty

  • decoderempty(Decoder $dec --> int)

Returns 1 if the decoder is empty (this means that there are no undecoded bytes, no decoded but untaken chars, and nothing in the normalization buffer). Otherwise returns 0.

Conditional Opcodes

if

  • if(Block $condition, Block $then)
  • if(Block $condition, Block $then, Block $else)

If the $condition evaluates to a non-zero value, run the $then block. If not, and an $else block is present, run that instead, if it's absent, return result of $condition.

unless

  • unless(Block $condition, Block $then)
  • unless(Block $condition, Block $then, Block $else)

If the $condition evaluates to 0, run the $then block. If not, and an $else block is present, run that instead, if it's absent, return result of $condition.

Loop/Control Opcodes

control

  • QAST::Op.new(:op<control>, :name<next>);
  • QAST::Op.new(:op<control>, :name<last>);
  • QAST::Op.new(:op<control>, :name<redo>);

Not callable directly from NQP, but used in languages via QAST to perform loop control. The specific kind of loop control desired is specified via the :name attribute; either next, last, or redo.

defor

  • defor(Block $cond, Block $body)

If the $cond evaluates to defined value, return it, otherwise, evaluate the $body.

for

  • for(Iterable $iter, Block $body)

Invoke the $body for every item available in $iter.

ifnull

  • ifnull(Block $cond, Block $body)

If the $cond evaluates to null, evaluate the $body, otherwise return the result of $cond.

repeat_until

  • repeat_until(Block $condition, Block $body)
  • repeat_until(Block $condition, Block $body, Block $post)

First run the $body. Then, enter a loop, running the $body only if the condition returns 0.

If a $post block is present, run that at the end, regardless of $condition.

repeat_while

  • repeat_while(Block $condition, Block $body)
  • repeat_while(Block $condition, Block $body, Block $post)

First run the $body. Then, enter a loop, running the $body only if the condition returns a non-0 value.

If a $post block is present, run that at the end, regardless of $condition.

stmts

  • stmts(...)

Executes the given statements sequentially. For example:

nqp::stmts((my $a := nqp::chars("foo")), say($a), say("bar"));
# 3
# bar

Note that := statements must be surrounded by parentheses.

until

  • until(Block $condition, Block $body)
  • until(Block $condition, Block $body, Block $post)

Enter a loop, running the $body only if the condition returns 0.

If a $post block is present, run that at the end, regardless of $condition.

while

  • while(Block $condition, Block $body)
  • while(Block $condition, Block $body, Block $post)

Enter a loop, running the $body only if the condition returns a non-0 value.

If a $post block is present, run that at the end, regardless of $condition.

Exceptional Opcodes

backtrace

  • backtrace(Exception $ex)

Return an array of hashes, describing the backtrace of the given exception.

backtracestrings

  • backtracestrings(Exception $ex)

Return an array of strings, describing the backtrace of the given exception.

die

  • die(str $message)
  • die_s(str $message)

Create and throw an exception using the given message.

exception

  • exception()

Return the current exception object.

getextype

  • getextype(Exception $ex)

Gets the exception category (nqp::const::CONTROL_*)

getmessage

  • getmessage(Exception $ex)

Gets the exception message.

getpayload

  • getpayload(Exception $ex)

Gets the exception payload.

newexception

  • newexception()

Return a new exception object.

resume

  • resume(Exception $ex)

Resume the exception, if possible.

rethrow

  • rethrow(Exception $ex)

Re-throw the exception.

setextype

  • setextype(Exception $ex, int $type)

Sets the exception category (nqp::const::CONTROL_*)

setmessage

  • setmessage(Exception $ex, str $message)

Sets the exception message.

setpayload

  • setpayload(Exception $ex, Mu $obj)

Sets the exception payload.

throw

  • throw(Exception $ex)

Throw the exception.

Input/Output Opcodes

closefh

  • closefh(Handle $fh)

Close the filehandle.

eoffh

  • eoffh(Handle $fh --> int)

Return 1 if this filehandle is at the end of the file, otherwise 0.

filenofh

  • filenofh(Handle $fh --> int) Returns the filehandle number. Not usable on the JVM (always returns -1).

flushfh

  • flushfh(Handle $fh)

Flushes the filehandle, forcing it to write any buffered output.

getstderr

  • getstderr()

Return the filehandle for standard error.

getstdin

  • getstdin()

Return the filehandle for standard input.

getstdout

  • getstdout()

Return the filehandle for standard output.

open

  • open(str $filename, str $mode)

Open the specified file in the given mode. Valid modes include r for read, w for write, and wa for write with append. Returns a filehandle.

print

  • print(str $str)

Output the given string to stdout.

readfh

  • readfh(Handle $fh, @arr, long $count)

Given a readable $fh, and an array of Buf[int8] or a Buf[uint8], read in the next $count bytes from the filehandle and store them in the array.

say

  • say(str $str)

Output the given string to stdout, followed by a newline.

seekfh

  • seekfh(Handle $fh, int $offset, int $whence)

Seek in the filehandle to the location specified by the offset and whence.

* `0` - from beginning of file
* `1` - from current position
* `2` - from end of file

tellfh

  • tellfh(Handle $fh --> int)

Return current access position for an open filehandle.

writefh

  • writefh(Handle $fh, Mu $str --> int)

Output the given object to the filehandle. Returns the number of bytes written.

External command Opcodes

execname moar js

  • execname(--> str)

It's used to implement $*EXECUTABLE in Raku, and is the name of the current "executable". So if you run ./raku-m .... then it'll be the ./raku-m. It's actually set at present by passing a --execname= argument to MoarVM, since raku is actually a shell script. But when we do get to providing a fake executable for raku instead, then it'd just initialize it to argv[0].

File / Directory / Network Opcodes

chdir

  • chdir(str $path --> str)

Change the working directory to the given path.

chmod

  • chmod(str $path, int $mode --> int)

Change the permissions of $path to the posix style permissions of $mode. Returns 0 on success, throws an exception on failure.

closedir

  • closedir(Handle $)

Close the given directory handle.

copy

  • copy(str $from, str $to --> int)

Copy file $from to file $to. Return 0 on success, throw an exception on failure.

cwd

  • cwd(--> str)

Return a string containing the current working directory.

fileexecutable

  • fileexecutable(str $str --> int)

If the specified filename refers to an executable file, returns 1. If not, returns 0. If an error occurs, return -1.

fileislink

  • fileislink(str $str --> int)

If the specified filename refers to a symbolic link, returns 1. If not, returns 0. If an error occurs, return -1.

filereadable

  • filereadable(str $str --> int)

If the specified filename refers to a readable file, returns 1. If not, returns 0. If an error occurs, return -1.

filewritable

  • filewritable(str $str --> int)

If the specified filename refers to a writeable file, returns 1. If not, returns 0. If an error occurs, return -1.

gethostname

  • gethostname(str $str --> str)

Returns the hostname of the system where it is run.

getport moar jvm

  • getport($obj --> int)

If the specified object is an IO::Handle, return the integer port number the object is listening on. If an error occurs, return -1.

link

  • link(str $before, str $after --> int)

Create a link from $after to $before

mkdir

  • mkdir(str $name, int $mode --> int)

Create a directory of the given name. Use posix-style mode on non-windows platforms. Returns 0, or throws an exception.

nextfiledir

  • nextfiledir(handle $iterator)

Given the result of an opendir, return the next path from that directory. When no more items are available, return a null string. (check with null_s)

opendir

  • opendir(str $path --> Mu)

Return a directory handle on the given directory path. Throw an exception if $path is not a directory.

rename

  • rename(str $from, str $to --> int)

Rename file $from to file $to. Return 0 on success, throw an exception on failure.

rmdir

  • rmdir(str $path --> int)

Delete the given directory $path. Returns 0 on success, -2 if the directory didn't exist. May throw an exception.

stat

  • stat(str $path, int $code --> int)

Given a path and a code, return an int describing that path using the OS's stat() function. Any of these variants may throw an exception if the platform does not support them. (JVM does not support STAT_PLATFORM_BLOCKSIZE or STAT_PLATFORM_BLOCKS).

* `nqp::const::STAT_EXISTS`

Returns 1 if the path exists, 0 otherwise.

* `nqp::const::STAT_FILESIZE`

Returns the size of the file in bytes.

* `nqp::const::STAT_ISDIR`

Returns 1 if the path is a directory, 0 otherwise, -1 if an exception occurred processing the request.

* `nqp::const::STAT_ISREG`

Returns 1 if the path is a regular file, 0 otherwise, -1 if an exception occurred processing the request.

* `nqp::const::STAT_ISDEV`

Returns 1 if the path is a special file, 0 otherwise, -1 if an exception occurred processing the request.

* `nqp::const::STAT_ISLNK`

Returns 1 if the path is a symbol link, 0 otherwise, -1 if an exception occurred processing the request.

* `nqp::const::STAT_CREATETIME`
* `nqp::const::STAT_ACCESSTIME`
* `nqp::const::STAT_MODIFYTIME`
* `nqp::const::STAT_CHANGETIME`

Returns respective time attribute in seconds since epoch, or -1 if an exception occurred.

* `nqp::const::STAT_BACKUPTIME`

Returns -1.

* `nqp::const::STAT_GID`
* `nqp::const::STAT_UID`

Returns the user id and group id of the path, respectively. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_DEV`

Returns the device number of filesystem associated with the path. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_INODE`

Returns the inode. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_MODE`

Returns unix style mode. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_NLINKS`

Returns number of hard links to the path. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_DEV`

Returns the device identifier. Returns -1 if an exception occurred.

* `nqp::const::STAT_PLATFORM_BLOCKSIZE`

Returns preferred I/O size in bytes for interacting with the file.

* `nqp::const::STAT_PLATFORM_BLOCKS`

Returns number of system-specific blocks allocated on disk.

stat_time

  • stat_time(str $path, int $code --> num)

Given a path and one of the STAT_*TIME codes, return that time attribute as a num, using the OS's stat() function.

lstat

  • lstat(str $path, int $code --> int)

Same as stat, but internally uses the OS's lstat() function, which does not follow symlinks.

lstat_time

  • stat_time(str $path, int $code --> num)

Same as stat_time, but internally uses the OS's lstat() function, which does not follow symlinks.

symlink

  • symlink(str $before, str $after)

Create a symbolic link from $after to $before

unlink

  • unlink(str $path --> int)

Delete the given file $path. Returns 0 on success, -2 if the file didn't exist. May throw an exception.

Type/Conversion Opcodes

bool

  • bool_I(Int $val)

Returns 0 if $val is 0, otherwise 1.

bootarray moar jvm

  • bootarray()

Returns a VM specific type object for a native array.

boothash moar jvm

  • boothash()

Returns a VM specific type object for a native hash.

bootint moar jvm

  • bootint()

Returns a VM specific type object that can box a native int.

bootintarray moar jvm

  • bootintarray()

Returns a VM specific type object for a native array of int.

bootnum moar jvm

  • bootnum()

Returns a VM specific type object that can box a native num.

bootnumarray moar jvm

  • bootnumarray()

Returns a VM specific type object for a native array of num.

bootstr moar jvm

  • bootstr()

Returns a VM specific type object that can box a native str.

bootstrarray moar jvm

  • bootstrarray()

Returns a VM specific type object for a native array of str.

box

  • box_i(int $val, Mu:T $type)
  • box_n(num $val, Mu:T $type)
  • box_s(str $val, Mu:T $type)

Given a native value, return a Raku object of the given type with the same value.

decont

decont(Mu $val --> Mu)

Extract, or de-containerize, a value from a Scalar container. If the argument is not a container, the argument is returned unchanged :

use nqp;
my $a = (1, 2, 3);
.say for $a;              # OUTPUT: «(1, 2, 3)␤»
.say for nqp::decont($a); # OUTPUT: «1␤2␤3␤»

defined

  • defined(Mu $obj --> int)

Returns 1 if the object is not null and is not a Type object, 0 otherwise.

fromnum

  • fromnum_I(num $val, Mu:T $type --> Int)

Convert float value to a Big Integer of the given type, discarding any decimal portion.

fromstr

  • fromstr_I(str $val, Mu:T $type --> Int)

Convert string value to a Big Integer of the given type.

isbig

  • isbig_I(Int $obj --> int)

Returns a 1 if the object's numerical representation requires a big int, 0 otherwise.

iscoderef moar

  • iscoderef($obj --> int)

Returns a 1 if the object contains a code reference, 0 otherwise.

isconcrete

  • isconcrete(Mu $obj --> int)

Returns a 1 if the object is not a type object, 0 otherwise.

isconcrete_nd

  • isconcrete_nd(Mu $obj --> int)

Returns a 1 if the object is not a type object, 0 otherwise. Does not decontainerize the object.

iscont

  • iscont(Mu $obj --> int)
  • iscont_i(int $int --> int)
  • iscont_n(num $int --> int)
  • iscont_s(str $int --> int)

Returns a 1 if the object is a container type, 0 otherwise.

isfalse

  • isfalse(Mu $obj --> int)

Returns a 0 if the object has a truthy value, 1 otherwise.

ishash

  • ishash(Mu $obj --> int)

Returns a 1 if the object is a Hash, 0 otherwise.

isint

  • isint(Mu $obj --> int)

Returns a 1 if the object is an int type, 0 otherwise.

isinvokable

  • isinvokable(Mu $obj --> int)

Returns a 1 if the object represents something executable, 0 otherwise.

islist

  • islist(Mu $obj)

Returns a 1 if the object is an Array, 0 otherwise.

isnanorinf

  • isnanorinf(num $n --> int)

Return truth value indicating if this number represents any of the special values, positive infinity, negative infinity, or NaN.

isnull

  • isnull(Mu $obj --> int)
  • isnull_s(str $obj --> int)

Returns a 1 if the object is a null, 0 otherwise.

isnum

  • isnum(Mu $obj --> int)

Returns a 1 if the object is a float type, 0 otherwise.

isprime

  • isprime_I(Int $obj --> int)

Returns a 1 if the integer value of the object is prime, 0 otherwise.

isstr

  • isstr(Mu $obj --> int)

Returns a 1 if the object is a str type, 0 otherwise.

istrue

  • istrue(Mu $obj --> int)

Returns a 1 if the object has a truthy value, 0 otherwise.

istype

  • istype(Mu $obj, Mu:T $obj --> int)

Returns a 1 if the object is of the given type, 0 otherwise.

isttyfh

  • isttyfh(Mu $obj --> int)

Returns a 1 if the object is an IO::Handle object that is a tty, 0 otherwise.

null

  • null(--> Mu)
  • null_s(--> str)

Generate a null value.

null_s returns a null string value that can be stored in a native str.

The value returned by null_s is VM dependant. Notably, it may stringify differently depending on the backend.

tostr

  • tostr_I(Int $val --> str)

Convert Big Integer value to a native string.

tonum

  • tonum_I(Int $val --> num)

Convert Big Integer value to a native number.

unbox

  • unbox_i(Mu $val --> int)
  • unbox_n(Mu $val --> num)
  • unbox_s(Mu $val --> str)
  • unbox_u(Mu $val --> str) moar

Given a Raku object, return a native with the same value, of the type indicated by the opcode suffix.

Binary Data Opcodes

For these definitions, buffer refers to a concrete object with a REPR of either VMArray or MultiDimArray, the latter being constrained to a single dimension. (Note: dimensionality is a property of the type, meaning that type specialization is already sufficient to optimize out both the REPR and shape checks.) In either case, the array must be an 8-bit integer array (as a Raku Blob or Buf will be).

Constants

The following new nqp::const entries are defined for use with the new ops, and specify sizes to use in reads and writes:

  • BINARY_SIZE_8_BIT
  • BINARY_SIZE_32_BIT
  • BINARY_SIZE_16_BIT
  • BINARY_SIZE_64_BIT

These nqp::const entries are defined for specifying the endianness of the data to read or write:

  • BINARY_ENDIAN_LITTLE
  • BINARY_ENDIAN_BIG
  • BINARY_ENDIAN_NATIVE

Operations not configured with one of these options will assume native endian. Reading or writing little endian on a little endian machine will, of course, carry no transformation overhead.

writeint moar js

  • writeint(buffer $target, int $offset, int $value, int $flags)

Writes the signed integer $value at $offset into the buffer $target, with the size and endianness specified by $flags.

writeuint moar js

  • writeuint(buffer $target, int $offset, uint $value, int $flags)

Writes the unsigned integer $value at $offset into the buffer $target, with the size and endianness specified by $flags.

writenum moar js

  • writenum(buffer $target, int $offset, num $value, int $flags)

Writes the floating point $value at $offset into the buffer $target. Only 32-bit and 64-bit sizes are supported.

readint moar js

  • readint(buffer $source, int $offset, int $flags --> int)

Reads a signed integer at offset $offset from $source with size and endianness specified by $flags. Returns that value, widened to a 64-bit int.

readuint moar js

  • readuint(buffer $source, int $offset, int $flags --> uint)

Reads an unsigned integer at offset $offset from $source with size and endianness specified by $flags. Returns that value, widened to a 64-bit uint.

readnum moar js

  • readnum(buffer $source, int $offset, int $flags --> num)

Reads a floating point number at offset $offset from $source with the size specified by $flags. Returns that value, widened to a 64-bit num.

OO/SixModel Opcodes

attrinited

  • attrinited(Mu $obj. Mu:T $type, str $attributename --> int)

Test if the attribute of name $attributename of object $obj has been bound, see bindattr. Note that any access to the atribute that results in a getattr call causes it to be inited.

bindattr

  • bindattr(Mu $obj, Mu:T $type, str $attributename, Mu $new_value)
  • bindattr_i(Mu $obj, Mu:T $type, str $attributename, int $new_value)
  • bindattr_n(Mu $obj, Mu:T $type, str $attributename, num $new_value)
  • bindattr_s(Mu $obj, Mu:T $type, str $attributename, str $new_value)

Binds $new_value to the attribute of name $attributename of object $obj, where the attribute was declared in type $type. The notes in the getattr documentation also apply to bindattr.

bindcomp

  • bindcomp(str $base-class, Mu $compiler --> Mu)

Registers $compiler as the compiler for the language named $base-class, as in:

my $lang = My::Lang::Compiler.new();
nqp::bindcomp('My::Lang', $lang);

In general, though, $lang will inherit from HLL::Compiler, and the above will be achieved via:

$lang.language('My::Lang');

call

  • call()

This method is not documented yet.

Example:

nqp::call(nqp::getlexcaller('&say'), 'foo')

callmethod

  • callmethod(Mu $obj, str $methodname, *@pos, *%named --> Mu)

Uses findmethod to locate method $methodname on object $obj, and call to invoke the method with positional arguments *@pos and named arguments *%named.

Example:

class A { method x($a, $b, :$c) { say("$a $b $c") } }

nqp::callmethod(A, 'x', '4', '2', c => 'foo');
# same as: A.x(4, 2, c => 'foo')

can

  • can(Mu $obj, str $method --> int)

If the object has a method of the given name, return 1. Otherwise, return 0.

Returns 1 if $obj object has FALLBACK method.

clone

  • clone(Mu $obj --> Mu)

Return a clone of the passed in object.

create

  • create(Mu:T $type --> Mu)

Returns a newly allocated instance of type $type.

eqaddr

  • eqaddr(Mu $l, Mu $r --> int)

Returns 1 if the objects are the same object in the underlying VM, 0 otherwise.

findmethod

  • findmethod(Mu $obj, str $method --> Mu)

If the object has a method of the given name, return it. Otherwise, throw an exception.

getattr

  • getattr(Mu $obj, Mu:T $type, str $attributename --> Mu)
  • getattr_i(Mu $obj, Mu:T $type, str $attributename --> int)
  • getattr_n(Mu $obj, Mu:T $type, str $attributename --> num)
  • getattr_s(Mu $obj, Mu:T $type, str $attributename --> str)

Returns the attribute of name $attributename of object $obj, where the object was declared in class $type. The _n, _i, and _s variants are for natively typed attributes.

The following example demonstrates why the type object needs to passed along, and cannot be inferred from the object:

class A      { has str $!x }
class B is A { has str $!x }

my $obj := nqp::create(B);
nqp::bindattr_s($obj, A, '$!x', 'A.x');
nqp::bindattr_s($obj, B, '$!x', 'B.x');

nqp::say(nqp::getattr_s($obj, A, '$!x'));
nqp::say(nqp::getattr_s($obj, B, '$!x'));

Throws an exception if there is no such attribute in the class, the attribute is of the wrong type, or the object doesn't conform to the type.

Note that in languages that support a full-blown container model, you might need to decontainerize $obj before passing it to getattr, unless you actually want to access an attribute of the container.

getcomp

  • getcomp(str $base-class --> Mu)

Returns the compiler class registered for that $base-class. See bindcomp for more information.

how

  • how(Mu $obj --> Mu)

NQP equivalent for Raku's $obj.HOW.

rebless

  • rebless(Mu $obj, Mu:T $type --> Mu)

Convert $obj to be an object of the new $type.

reprname

  • reprname(Mu $obj --> str)

Return the name of the REPR for the given object.

setwho

  • setwho(Mu $obj, Mu $who --> Mu)

Replace $obj's WHO. Return $obj.

who

  • who(Mu $obj --> Mu)

NQP equivalent for Raku's $obj.WHO.

what

  • what(Mu $obj --> Mu)

NQP equivalent for Raku's $obj.WHAT.

where

  • where(Mu $obj --> int)

Return a unique ID for this $obj.

Bit Opcodes

bitand

  • bitand_i(int $l, int $r)
  • bitand_s(str $l, str $r --> str)
  • bitand_I(Int $l, Int $r, Mu:T $type --> Int)

AND the bits in $l and $r. _I variant returns an object of the given type.

bitneg

  • bitneg_i(int $bits --> int)
  • bitneg_I(Int $bits, Mu:T $type --> Int)

Negate the bits in $bits. _I variant returns an object of the given type.

bitor

  • bitor_i(int $l, int $r)
  • bitor_s(str $l, str $r --> str)
  • bitor_I(Int $l, Int $r, Mu:T $type --> Int)

OR the bits in $l and $r. _I variant returns an object of the given type.

bitshiftl

  • bitshiftl_i(int $bits, int $count --> int)
  • bitshiftl_I(Int $bits, int $count, Mu:T $type --> Int)

Signed left shift of $bits by $count. _I variant returns an object of the given type.

bitshiftr

  • bitshiftr_i(int $bits, int $count --> int)
  • bitshiftr_I(Int $bits, int $count, Mu:T $type --> Int)

Signed right shift of $bits by $count. _I variant returns an object of the given type.

bitxor

  • bitxor_i(int $l, int $r --> int)
  • bitxor_s(str $l, str $r --> str)
  • bitxor_I(Int $l, Int $r, Mu:T $type --> Int)

XOR the bits in $l and $r. _I variant returns an object of the given type.

Context Introspection Opcodes

ctx

  • ctx(--> Context)

Return the object representing the current context.

ctxcaller

  • ctxcaller(Context $ctx)

Given a context, return the caller context, or null.

ctxlexpad

  • ctxlexpad(Context $ctx)

Given a context, return its lexpad

curlexpad

  • curlexpad()

Return the current lexpad.

ctxouter

  • ctxouter(Context $ctx)

Given a context, return the outer context, or null.

lexprimspec

  • lexprimspec(LexPad $pad, str $name --> int)

Given a lexpad and a name, return the name's primitive type.

The primitive types are 1 for int, 2 for num and 3 for str. 0 is any object.

savecapture

  • savecapture()

Gets hold of the argument capture passed to the current block. Commits to doing enough copying that the list is valid any amount of time. See usecapture for a version of the op that doesn't promise that. Used by the multi-dispatcher.

usecapture

  • usecapture()

Gets hold of the argument capture passed to the current block. (a future usecapture may invalidate it) It's valid to implement this exactly the same way as savecapture if there's no performance benefit to be had in a split. Used by the multi-dispatcher.

getlex moar jvm

  • getlex(str $name)
  • getlex_i(str $name)
  • getlex_n(str $name)
  • getlex_s(str $name)

Looks up the lexical with the specified name and the specified type. Searching in the outer frames, starting at the current. An error is thrown if it does not exist or if the type is incorrect.

getlexref moar jvm

  • getlexref_i(str $name)
  • getlexref_n(str $name)
  • getlexref_s(str $name)

Looks up the native type lexical with the specified name and the specified type. Searching in the outer frames, starting at the current. An error is thrown if it does not exist or if the type is incorrect.

Lexicalref is a mechanism that allows us to treat a native value stored in a lexpad as if it were a read-writable container we can pass around. the lexicalref that gets created holds a reference to the frame in question and any access to it acts like getlex from the frame it originated in

bindlex moar jvm

  • bindlex(str $name, Mu $value)
  • bindlex_i(str $name, int $value)
  • bindlex_n(str $name, num $value)
  • bindlex_s(str $name, str $value)

Binds $value to the lexical specified by name and type. Searching in the outer frames, starting at the current. An error is thrown if it does not exist or if the type is incorrect.

getlexdyn

  • getlexdyn(str $name)

Looks up the contextual with the specified name in the caller chain, starting at the calling frame.

bindlexdyn

  • bindlexdyn(str $name, Mu $value)

Binds $value to the contextual with the specified name, searching for it in the call-chain, starting at the calling frame.

getlexouter

  • getlexouter(str $name)

Looks up the lexical with the specified name and the specified type. Searching in the outer frames, starting at outer.

getlexcaller

  • getlexcaller(str $name)

Looks up the lexical with the specified name, starting at the calling frame. It checks all outer frames of the caller chain.

getlexrel

  • getlexrel(Mu $context, str $name)

Looks up the lexical with the specified name and the specified type. Searching in the outer frames, starting at the given $context.

getlexreldyn

  • getlexreldyn(Mu $context, str $name)

Looks up the contextual with the specified name in the caller chain, starting at the given $context.

getlexrelcaller

  • getlexrelcaller(Mu $context, str $name)

Looks up the lexical with the specified name, starting at the given $context. It checks all outer frames of the caller chain.

Variable Opcodes

bind

  • bind(Mu $variable, Mu $value)

Binds $value to the $variable. Dies if $variable isn't actually a variable. Same as the := operator in NQP.

Miscellaneous Opcodes

locallifetime

  • QAST::Op.new(:op<locallifetime>, :node($/), QAST::Stmt.new(...))

Defines when local variables can be considered dead. E.g. the temporary setting of $_ on the right side of ~~ uses that.

const

  • const()

Not actually an opcode, but a collection of several constants. Each of the constants below can be used in nqp as (e.g.) nqp::const::CCLASS_ANY.

* CCLASS_ANY
* CCLASS_UPPERCASE Lu
* CCLASS_LOWERCASE Ll
* CCLASS_ALPHABETIC Lo | Ll | Lu | Lt | Lm
* CCLASS_NUMERIC Nd
* CCLASS_HEXADECIMAL
* CCLASS_WHITESPACE (9..13,32,133,160,5760,8192..8202,8232,8233,8239,8287,12228)
* CCLASS_PRINTING !(0..31, 127..159)
* CCLASS_BLANK Zs
* CCLASS_CONTROL (0..31, 127..159)
* CCLASS_PUNCTUATION
* CCLASS_ALPHANUMERIC Lo | Ll | Lu | Lt | Lm | Nd
* CCLASS_NEWLINE Zl Zp
* CCLASS_WORD Lo | Ll | Lu | Lt | Lm | Nd + "_"

* HLL_ROLE_NONE
* HLL_ROLE_INT
* HLL_ROLE_NUM
* HLL_ROLE_STR
* HLL_ROLE_ARRAY
* HLL_ROLE_HASH
* HLL_ROLE_CODE

* CONTROL_TAKE
* CONTROL_LAST
* CONTROL_NEXT
* CONTROL_REDO
* CONTROL_SUCCEED
* CONTROL_PROCEED
* CONTROL_WARN

* STAT_EXISTS
* STAT_FILESIZE
* STAT_ISDIR
* STAT_ISREG
* STAT_ISDEV
* STAT_CREATETIME
* STAT_ACCESSTIME
* STAT_MODIFYTIME
* STAT_CHANGETIME
* STAT_BACKUPTIME
* STAT_UID
* STAT_GID
* STAT_ISLNK
* STAT_PLATFORM_DEV
* STAT_PLATFORM_INODE
* STAT_PLATFORM_MODE
* STAT_PLATFORM_NLINKS
* STAT_PLATFORM_DEVTYPE
* STAT_PLATFORM_BLOCKSIZE
* STAT_PLATFORM_BLOCKS

* BINARY_SIZE_8_BIT
* BINARY_SIZE_32_BIT
* BINARY_SIZE_16_BIT
* BINARY_SIZE_64_BIT
* BINARY_ENDIAN_LITTLE
* BINARY_ENDIAN_BIG
* BINARY_ENDIAN_NATIVE

* RUSAGE_UTIME_SEC
* RUSAGE_UTIME_MSEC
* RUSAGE_STIME_SEC
* RUSAGE_STIME_MSEC

* UNAME_SYSNAME
* UNAME_RELEASE
* UNAME_VERSION
* UNAME_MACHINE

* TYPE_CHECK_CACHE_DEFINITIVE
* TYPE_CHECK_CACHE_THEN_METHOD
* TYPE_CHECK_NEEDS_ACCEPTS

cpucores

  • cpucores()

Returns a native integer for the number of CPU cores that are reported to be available.

decodelocaltime

  • decodelocaltime(int $epoch --> int @tm)

Returns an integer array with localtime information, formatted like the C struct tm: $sec,$min,$hour,$mday,$mon,$year,$weekday,$yearday,$isdst. Note that contrary to C's localtime() function, the $year contains the actual year (A.D), and the $month has been normalized to 1..12.

force_gc moar jvm

  • force_gc()

Force the garbage collector to run.

getcodename

  • getcodename($obj --> str)

Returns the name of the given concrete code object. Throws an exception if an object of the wrong type is passed.

getrusage

  • getrusage(int @rusage)

Accepts an integer array and fills it with usage data, of which the following elements are currently defined:

  • RUSAGE_UTIME_SEC Userland CPU usage (seconds part)
  • RUSAGE_UTIME_MSEC Userland CPU usage (micro-seconds part)
  • RUSAGE_STIME_SEC System CPU usage (seconds part)
  • RUSAGE_STIME_MSEC System CPU usage (micro-seconds part)
  • RUSAGE_MAXRSS Maximum resident set size (in bytes/Kbytes)
  • RUSAGE_IXRSS Integral shared text memory size (in bytes/Kbytes)
  • RUSAGE_IDRSS Integral unshared data size (in bytes/Kbytes)
  • RUSAGE_ISRSS Integral unshared stack size (in bytes/Kbytes)
  • RUSAGE_MINFLT Number of page reclaims (lower part)
  • RUSAGE_MAJFLT Number of page reclaims (upper part)
  • RUSAGE_NSWAP Number of swaps
  • RUSAGE_INBLOCK Number of block input operations
  • RUSAGE_OUBLOCK Number of block output operations
  • RUSAGE_MSGSND Number of messages sen
  • RUSAGE_MSGRCVA Number of messages received
  • RUSAGE_NSIGNALS Number of signals received
  • RUSAGE_NVCSW Number of voluntary context switches
  • RUSAGE_NIVCSW Number of involuntary context switches

Currently, the elements ending in RSS appear to return values in bytes rather than Kbytes on MacOS.

Elements may be 0 if it is impossible to determine that value in the current system.

uname moar js

  • uname(--> Mu)

Returns a string array and fills it with uname data, of which the following elements are currently defined:

  • UNAME_SYSNAME Name of the operating system implementation
  • UNAME_RELEASE Release level of the operating system
  • UNAME_VERSION Version level of the operating system
  • UNAME_MACHINE Machine hardware platform

debugnoop jvm

  • debugnoop(Mu $a)

Returns $a. Does nothing, exists only to provide a breakpoint location for debugging.

exit

  • exit(int $status)

Exit nqp, using the given status as the compiler's exit value.

getenvhash

  • getenvhash(--> Mu)

Returns a hash containing the environment variables. Changing the hash doesn't affect the environment variables

getsignals

  • getsignals(--> Mu)

Returns a list containing signal names interleaved with the associated signum integer on the host platform (MacOSX, Linux, BSD, etc).

If the current backend does not support the registering of a signal handler for a given signal, the hash value will be a negative integer. For instance, the JVM only supports signal handlers for SIGINT and SIGKILL, so all the values will be negative except 2 (SIGINT) and 9 (SIGKILL). If a signal is not available on the host system, the hash value will be set to 0.

The complete list of signal entries is as follows:

* SIGHUP
* SIGINT
* SIGQUIT
* SIGILL
* SIGTRAP
* SIGABRT
* SIGEMT
* SIGFPE
* SIGKILL
* SIGBUS
* SIGSEGV
* SIGSYS
* SIGPIPE
* SIGALRM
* SIGTERM
* SIGURG
* SIGSTOP
* SIGTSTP
* SIGCONT
* SIGCHLD
* SIGTTIN
* SIGTTOU
* SIGIO
* SIGXCPU
* SIGXFSZ
* SIGVTALRM
* SIGPROF
* SIGWINCH
* SIGINFO
* SIGUSR1
* SIGUSR2
* SIGTHR
* SIGSTKFLT
* SIGPWR
* SIGBREAK

backendconfig

  • backendconfig(--> Mu)

Returns a hash containing backend-specific information, like backend-version, configure and build flags.

getpid

  • getpid(--> int)

Return the current process id, an int.

getppid moar

  • getppid(--> int)

Return the process id of the parent process, an int.

indexingoptimized

  • indexingoptimized(str --> str)

If the input string is made up of strands, then returns a flattened string that is otherwise identical. If not, returns the input string.

Intended for strings that will be indexed into often, for example, when evaluating regexes.

js moar js

  • js(str)

Execute the string of JavaScript code passed in.

While this opcode exists in moar, it throws an exception declaring it is not implemented.

jvmclasspaths jvm

  • jvmclasspaths(--> Mu)

Converts the JVM property java.class.path into a list of paths, returns it.

jvmgetproperties jvm

  • jvmgetproperties(--> Hash)

Map the JVM's System.getProperties into a Hash usable in NQP. Normalizes some OS names (key: 'os.name'), returns all other data as is.

setcodename

  • setcodename($obj, str)

Sets the name of the given code object. Throws an exception if an object of the wrong type is passed.

sha1

  • sha1(str $str -> str)

Given a UTF-8 string, return the SHA-1 digest for that string. This op is built for the specific purpose of hashing source code for dependency management purposes, and isn't intended to be used more widely.

sleep

  • sleep(num $seconds --> num)

Sleep for the given number of seconds (no guarantee is made how exact the time sleeping is spent.) Returns the passed in number.

takeclosure

  • takeclosure(Block $innerblock)

Creates a lexical closure from the block's outer scope.

time

  • time_i(--> int)
  • time_n(--> num)

Return the time in seconds since January 1, 1970 UTC. _i variant returns an integral number of seconds, _n returns a fractional amount.

totalmem

  • totalmem(--> int)

Returns the number of bytes of memory in use by the VM.

mvmstartprofile moar

  • mvmstartprofile(hash $config)

Turns on one of MoarVM's profilers. The configuration must have a kind key that specifies which profiler will be turned on:

  • instrumented

    takes no further configuration options. records call graph, garbage collection, and object allocation information.

  • heap

    takes a path / filename in the path key. writes a snapshot of the heap's structure (objects and their connections) to the file every time the GC runs.

If a profiler is already active, an exception will be thrown; only one profiler can run at a time.

mvmendprofile moar

  • mvmendprofile(--> Object)

Turns off the profiler and returns data gathered.

  • heap

    currently doesn't return anything. it does, however, cause one heap snapshot to be taken immediately.

  • instrumented

The first element is an array of arrays with information about the types that have been allocated. At the moment of writing, this array appears to have information about objects that were created, but for which there is no allocation information. It has the following structure:

0                                  - array with type information
├ 0 = 140415871842064               - unique ID for this type
└ 1                                 - hash with additional information
  ├ repr => P6opaque                  - name of the REPR of this type
  ├ type => Block                     - type object of type (aka, the .WHAT)
  ├ managed_size => 72                - size in bytes of instance
  └ has_unmanaged_data => 1           - is there additional data on heap?

The second element of the list returned by nqp::mvmendprofile, is a list of hashes, one for each thread on which data has been collected. It has the following structure (times are in microseconds, sizes are in bytes):

0                                 - hash with info of thread
├ thread => 1                       - OS thread ID
├ parent => 0                       - OS thread ID of parent thread
├ spesh_time => 0                   - amount of time spent in spesh
├ start_time => 0                   - when thread was started
├ total_time => 21004               - total time spent in thread
├ call_graph                        - hash with first Callee info
│ ├ id => 140328666076608             - unique ID of this Callee
│ ├ first_entry_time => 0             - when Callee was first called
│ ├ inclusive_time => 2               - time spent here + all sub-Callees
│ ├ exclusive_time => 2               - time spent in this Callee
│ ├ entries => 97897                  - number of times Callee was called
│ ├ inlined_entries => 56757          - times called when inlined
│ ├ jit_entries => 6566               - times called when jitted
│ ├ osr => 1                          - times Callee was OSR'd
│ ├ name => foo                       - name of Callee (if available)
│ ├ file => gen/moar/BOOTSTRAP.nqp    - filename of Callee
│ ├ line => 2070                      - line of Callee in file
│ ├ allocations => (2)                - array with Allocations
│ │ ├ 0                                 - hash with Allocation info
│ │ │ ├ count => 100                      - number of allocations
│ │ │ ├ replaced => 1                     - scalar replacements stopping alloc
│ │ │ └ id => 140329083232016             - type ID
│ └ callees => (2)                    - array with Callees called here
└ gcs                               - array with Garbage Collections
  └ 0                                 - hash with GC info
    ├ sequence => 0                     - ordinal number of GC
    ├ start_time => 1964                - when GC was started
    ├ time => 7222                      - time spent doing GC
    ├ full => 0                         - whether or not a full GC
    ├ responsible => 1                  - thread ID that triggered this GC
    ├ promoted_bytes => 212960          - bytes promoted from the nuresery
    ├ promoted_bytes_unmanaged => 54781 - additional bytes promoted
    ├ retained_bytes => 76576           - bytes *not* promoted
    ├ cleared_bytes => 3228716          - bytes cleared from the nursery
    ├ gen2 => 18402                     -
    ├ gen2_roots => 18402               - gen2 allocs rooted in nursery
    ├ deallocs                          - array with Deallocations
      ├ 0                                 - hash with deallocation info
        ├ id => 140329080607960             - type ID being deallocated
        ├ nursery_seen => 10                - seen before in a GC
        └ nursery_fresh => 6                - *not* seen before in a GC

Native Call / Interoperability Opcodes

nativecallrefresh

Refresh the C-based data backing the Perl 6 object. This op should only be used if changes have been made to the C-data, and these changes are not being reflected in the Perl 6 object.

Asynchronous Operations

The various asynchronous operations, such as timers and asynchronous I/O, take a concurrent queue to push a work item into at an appropriate time. This may be a code object to be invoked, or it may be an array of a code item and some arguments to supply to it. Asynchronous operations are represented by some object with the AsyncTask REPR, the exact details of which are highly specific to a given backend. The type to use for that is given as $handle_type.

[As of 2014.04, these are very new and subject to revision and additions.]

permit

  • permit(AsyncTask $handle, int $channel, int $permits)

Takes something with the AsyncTask REPR (the $handle parameter) and permits it to emit up to $permits more notifications. This is used as a back-pressure mechanism for asynchronous tasks that produce a stream of events, such as representing data arriving over a socket. Some kinds of tasks may emit on multiple channels, for example an asynchronous process may emit events for STDOUT (channel 1) and STDERR (channel 2) if both are of interest. The $channel argument is used to specify which channel is to get the permits if needed (use a separate permit stament for each channel of interest).

If $permits is less than zero (e.g., permit($task, $channel, -1), then it means there is no limit to the emits.

If $permits is set to any value greater than or equal to zero, then:

  • In the case unlimited emits were permitted previously, the permits will be set to the new value. If the new value is zero, then the reader will be stopped.
  • Otherwise the number of permits will be incremented by the specified value. If the resulting number of permits allowed is greater than zero and the reader is not running, it will be started.

cancel moar jvm

  • cancel(AsyncTask $handle)

Takes something with the AsyncTask REPR and tries to cancel it, if it is possible to do so. If it is somehow not possible (for example, the operation already completed anyway), then nothing will happen. This is to avoid race conditions.

timer moar jvm

  • timer($queue, $schedulee, int $timeout, int $repeat, $handle_type)

Starts a timer. If timeout is zero, the $schedulee is immediately pushed to the queue. Otherwise, it is pushed after the timeout period. If repeat is non-zero, after the initial timeout period it will then be pushed again at the repeat interval. Returns an object of type $handle_type, which has a AsyncTask REPR. Cancellation stops the timer ever repeating again.

signal moar jvm

  • signal($queue, $schedulee, int [nqp::cosnt::SIG_], $handle_type)

Sets up a signal handler for the given signal. Whenever it occurs, an array is pushed to the queue containing the schedulee and the signal number. Cancel to stop handling it.

watchfile moar jvm

  • watchfile($queue, $schedulee, str $filename, $handle_type)

Watches an individual file for changes. Pushes an array to the queue when a change is detected, consisting of the schedulee, the filename that changed if provided by the underlying watcher mechanism, a 0 if the file changed, and a 1 if it was renamed. Cancel to stop watching.

asyncconnect

  • asyncconnect($queue, $schedulee, str $host, int $port, $handle_type)

Creates an asynchronous client socket and commences a connection operation. Upon connection, the queue will be passed an array consisting of the schedulee, a handle if the connection was successful (a type object if not) and an error string (some type object if no error). Returns an AsyncTask representing the connection attempt.

asynclisten

  • asynclisten($queue, $schedulee, str $host, int $port, $handle_type)

Creates an asynchronous server socket listening on the specified host and port. Each time a connection arrives, the queue will be passed an array consisting of the schedulee and the newly created asynchronous socket, for communicating with the connecting client. Returns an AsyncTask that can be cancelled to stop listening, or throws an exception if there is an error starting to listen.

asyncwritebytes

  • asyncwritebytes($handle, $queue, $schedulee, $to_write, $handle_type)

Writes a byte array to some handle capable of asynchronous operations. Once the write is complete, the queue will be passed an array consisting of the schedulee, an integer containing the number of bytes written or a type object if there was an error, and a string containing an error or some type object if none.

asyncreadbytes moar jvm

  • asyncreadbytes($handle, $queue, $schedulee, $buf_type, $handle_type)

Starts reading bytes from the handle. When a packet is received, a $buf_type will be constructed and point to the received memory. An array will be pushed to the queue containing the schedulee, a sequence number that starts at 0, the buffer or just its type object on error, and an error string (type object if no error). If EOF is reached, a sequence number of -1 is sent. Cancel to stop reading.

spawnprocasync

  • spawnprocasync($queue, $args, $cwd, %env, $callbacks)

Replaced shell and spawn. See t/nqp/111-spawnprocasync.t for an example of use.

killprocasync

  • killprocasync($handle, $signal)

HLL-Specific Operations

hllbool

  • hllbool(int -> obj)

If passed 0, return a HLL specific Boolean false value, otherwise, a true one. For Raku, this maps to Bool::False and Bool::True, respectively.

hllboxtype

  • hllboxtype_i(Mu)
  • hllboxtype_n(Mu)
  • hllboxtype_s(Mu)

Ignores any args passed and returns the HLL specific type objects for each basic type. For Raku, this maps to Int, Num, and Str.

hllhash moar jvm

  • hllhash(Mu)

Returns HLL specific type object for a hash. Ignores optional argument.

hlllist moar jvm

  • hlllist(Mu)

Returns HLL specific type object for a list. Ignores optional argument.

bindhllsym / bindcurhllsym

  • bindhllsym(str $hllname, str $symname, $value)
  • bindcurhllsym(str $symname, $value)

Store a value in a specified HLL's symbol hash at a given key; the cur variant uses the hll the code that has the op in it was compiled for.

gethllsym / getcurhllsym

  • gethllsym(str $hllname, str $symname --> Mu)
  • getcurhllsym(str $symname --> Mu)

Retrieve a value from a specified HLL's symbol hash at a given key; The cur variant uses the hll the code that has the op in it was compiled for.

usecompilerhll / usecompileehll

  • usecompilerhllconfig
  • usecompileehllconfig

Increases or decreases the "compilee depth" value. When the compilee depth is greater than one, every hll access will hit the "compilee's HLL config", otherwise every access will hit the "compiler's HLL config".

This serves, for example, to seperate a running NQP compiler from an NQP compiler it's compiling, when compile-time evaluation happens.

Atomic Operations

cas moar

  • cas(ObjectContainer $cont, Mu $expected, Mu $new --> Mu)

Takes an object which has a container spec set on it that knows how to do an atomic compare and swap, and performs an atomic compare and swap operation. The operation atomically compares the $expected object with what is currently held in the container. If they are the same object, then it replaces it with $new. If not, no change takes place. The original object stored in the container is returned, which can be used with eqaddr to check if it is the same as the $expected object. The container may perform type checks on the $new object before it attempts the operation.

cas_i moar

  • cas_i(NativeIntRef $i, int64 $expected, int64 $new --> int)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Casts the expected and new parameters to the machine's atomic operation size, and then uses them to perform an atomic compare and swap operation on the referenced integer. The operation atomically compares the $expected value with the value currently at the referenced location. If they are equal, it replaces the value with $new. If they are not equal, nothing happens. The operation evaluates to the value originally at the location (which can be compared with $expected to see if the operation was a success).

atomicinc_i moar

  • atomicinc_i(NativeIntRef $i --> int)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Performs an atomic increment of the referenced integer. Returns the value before it was incremented.

atomicdec_i moar

  • atomicdec_i(NativeIntRef $i --> int)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Performs an atomic decrement of the referenced integer. Returns the value before it was decremented.

atomicadd_i moar

  • atomicadd_i(NativeIntRef $i, int $value --> int)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Performs an atomic addition of the provided value, which will be cast to the machine's atomic operation size before the operation is performed. Returns the value at the memory location before the addition was performed.

atomicload moar

  • atomicload(ObjectContainer $c)

Takes an object which has a container spec set on it that knows how to do an atomic load (that is, with appropriate barriering to ensure the latest value is read). Performs the atomic load, and returns the loaded object.

atomicload_i moar

  • atomicload_i(NativeIntRef $i --> int)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Performs an atomic load (that is, with appropriate barriering to ensure the latest value is read).

atomicstore moar

  • atomicstore(ObjectContainer $c, Mu $value)

Takes an object which has a container spec set on it that knows how to do an atomic load. Performs the atomic store, which may fail if the value being stored does not, for example, meet type constraints. Evaluates to the stored value. The store performs appropriate barriering to ensure the changed value is "published".

atomicstore_i moar

  • atomicstore_i(NativeIntRef $i, int64 $value)

Takes an object with the NativeRef representation, which must point to an integer of the machine's atomic operation size. Performs an atomic store (that is, with appropriate barriering to ensure the changed value is "published").

barrierfull moar

  • barrierfull()

Performs a full memory barrier.

Serialization context

Abbreviated as SC. You probably don't need any of these. When creating a new language and possibly a new World class, you will inherit serialization code that use these opcodes. For test examples, see t/serialization/

createsc

  • createsc($handle-string) creates a serialization context and returns it.

scsetdesc

  • scsetdesc($sc, $descriptor-string)

Set a descriptor for $sc created by createsc()

scgetdesc

  • scgetdesc($sc)

Get the descriptor set by scsetdec

scgethandle

  • scgethandle($sc)

Get the handle string used by createsc to create the SC $sc

pushcompsc

  • pushcompsc($sc)

popcompsc

  • popcompsc($sc)

scsetobjc

  • scsetobj($sc, $idx, $obj)

setobjsc

  • setobjsc($obj, $sc)

getobjsc

  • getobjsc($obj)

scgetobjidx

  • scgetobjidx()

serialize

  • serialize()

deserialize

  • deserialize()

scobjcount

  • scobjcount()

freshcoderef

  • freshcoderef($code-object) Creates a clone of the given code object and a clone of its static frame and connects the two. Returns the clone of the code object. Used for creating a fresh copy of a statically compiled piece of code so each of some high level code object can gets its own low level executable.

markcodestatic

  • markcodestatic($code-object) Marks the code as "static" meaning it's not a closure and thus no closure will be serialized.

scsetcode

  • scsetcode($sc, $index, $code-object) Adds $code-object to the serialization context at block index $index

forceouterctx

  • forceouterctx($code-object, $context) Sets the code object's outer to the context's frame and also the code object's static frame's outer to the context's static frame. Used to embed a separately compiled code object in a given context, e.g. to give an EVALed code a surrounding lexical scope.

neverrepossess

  • neverrepossess($obj) Prevents the object from ever getting repossessed. Repossession means that an object from a different serialization context, i.e. something we got from loading a module, gets added to our own serialization context as well. This is done to keep modifications to the object. Of course if different versions of the same object are loaded from different serialization contexts, there's a conflict that requires resolution -> resolve_reposession_conflicts. In the common case of the object being a Stash, we can just merge the different versions unless the keys overlap. For some objects we do not want repossession even if they were modified, i.e. they were only needed for compilation and/or there wouldn't be a way to resolve conflicts.
You can’t perform that action at this time.