123456789_123456789_123456789_123456789_123456789_

Ractor - Ruby's Actor-like concurrent abstraction

Ractor is designed to provide a parallel execution feature of Ruby without thread-safety concerns.

Summary

Multiple Ractors in an interpreter process

You can make multiple Ractors and they run in parallel.

Limited sharing between multiple ractors

Ractors don't share everything, unlike threads.

Two-types communication between Ractors

Ractors communicate with each other and synchronize the execution by message exchanging between Ractors. There are two message exchange protocols: push type (message passing) and pull type.

Copy & Move semantics to send messages

To send unshareable objects as messages, objects are copied or moved.

Thread-safety

Ractor helps to write a thread-safe concurrent program, but we can make thread-unsafe programs with Ractors.

Without Ractor, we need to trace all state-mutations to debug thread-safety issues. With Ractor, you can concentrate on suspicious code which are shared with Ractors.

Creation and termination

Ractor.new

# Ractor.new with a block creates new Ractor
r = Ractor.new do
  # This block will be run in parallel with other ractors
end

# You can name a Ractor with `name:` argument.
r = Ractor.new name: 'test-name' do
end

# and Ractor#name returns its name.
r.name #=> 'test-name'

Given block isolation

The Ractor executes given expr in a given block. Given block will be isolated from outer scope by the Proc#isolate method (not exposed yet for Ruby users). To prevent sharing unshareable objects between ractors, block outer-variables, self and other information are isolated.

Proc#isolate is called at Ractor creation time (when Ractor.new is called). If given Proc object is not able to isolate because of outer variables and so on, an error will be raised.

begin
  a = true
  r = Ractor.new do
    a #=> ArgumentError because this block accesses `a`.
  end
  r.take # see later
rescue ArgumentError
end
r = Ractor.new do
  p self.class #=> Ractor
  self.object_id
end
r.take == self.object_id #=> false

Passed arguments to Ractor.new() becomes block parameters for the given block. However, an interpreter does not pass the parameter object references, but send them as messages (see below for details).

r = Ractor.new 'ok' do |msg|
  msg #=> 'ok'
end
r.take #=> 'ok'
# almost similar to the last example
r = Ractor.new do
  msg = Ractor.receive
  msg
end
r.send 'ok'
r.take #=> 'ok'

An execution result of given block

Return value of the given block becomes an outgoing message (see below for details).

r = Ractor.new do
  'ok'
end
r.take #=> `ok`
# almost similar to the last example
r = Ractor.new do
  Ractor.yield 'ok'
end
r.take #=> 'ok'

Error in the given block will be propagated to the receiver of an outgoing message.

r = Ractor.new do
  raise 'ok' # exception will be transferred to the receiver
end

begin
  r.take
rescue Ractor::RemoteError => e
  e.cause.class   #=> RuntimeError
  e.cause.message #=> 'ok'
  e.ractor        #=> r
end

Communication between Ractors

Communication between Ractors is achieved by sending and receiving messages. There are two ways to communicate with each other.

Users can control program execution timing with (1), but should not control with (2) (only manage as critical section).

For message sending and receiving, there are two types of APIs: push type and pull type.

Sending/Receiving ports

Each Ractor has incoming-port and outgoing-port. Incoming-port is connected to the infinite sized incoming queue.

                  Ractor r
                 -------------------------------------------
                 | incoming                         outgoing |
                 | port                                 port |
   r.send(obj) ->*->[incoming queue]     Ractor.yield(obj) ->*-> r.take
                 |                |                          |
                 |                v                          |
                 |           Ractor.receive                  |
                 -------------------------------------------


Connection example: r2.send obj on r1、Ractor.receive on r2
  ----     ----
  * r1 |---->* r2 *
  ----     ----


Connection example: Ractor.yield(obj) on r1, r1.take on r2
  ----     ----
  * r1 *---->- r2 *
  ----     ----

Connection example: Ractor.yield(obj) on r1 and r2,
                    and waiting for both simultaneously by Ractor.select(r1, r2)

  ----
  * r1 *------+
  ----      |
              +----> Ractor.select(r1, r2)
  ----      |
  * r2 *------|
  ----
r = Ractor.new do
  msg = Ractor.receive # Receive from r's incoming queue
  msg # send back msg as block return value
end
r.send 'ok' # Send 'ok' to r's incoming port -> incoming queue
r.take      # Receive from r's outgoing port

The last example shows the following ractor network.

------        ---
* main |------> * r *---+
------        ---   |
    ^                   |
    -------------------

And this code can be simplified by using an argument for Ractor.new.

# Actual argument 'ok' for {Ractor.new()} will be sent to created Ractor.
r = Ractor.new 'ok' do |msg|
  # Values for formal parameters will be received from incoming queue.
  # Similar to: msg = Ractor.receive

  msg # Return value of the given block will be sent via outgoing port
end

# receive from the r's outgoing port.
r.take #=> `ok`

Return value of a block for Ractor.new

As already explained, the return value of Ractor.new (an evaluated value of expr in Ractor.new{ expr }) can be taken by Ractor#take.

Ractor.new{ 42 }.take #=> 42

When the block return value is available, the Ractor is dead so that no ractors except taken Ractor can touch the return value, so any values can be sent with this communication path without any modification.

r = Ractor.new do
  a = "hello"
  binding
end

r.take.eval("p a") #=> "hello" (other communication path can not send a Binding object directly)

Wait for multiple Ractors with Ractor.select

You can wait multiple Ractor's yield with Ractor.select(*ractors). The return value of Ractor.select() is [r, msg] where r is yielding Ractor and msg is yielded message.

Wait for a single ractor (same as Ractor.take):

r1 = Ractor.new{'r1'}

r, obj = Ractor.select(r1)
r == r1 and obj == 'r1' #=> true

Wait for two ractors:

r1 = Ractor.new{'r1'}
r2 = Ractor.new{'r2'}
rs = [r1, r2]
as = []

# Wait for r1 or r2's Ractor.yield
r, obj = Ractor.select(*rs)
rs.delete(r)
as << obj

# Second try (rs only contain not-closed ractors)
r, obj = Ractor.select(*rs)
rs.delete(r)
as << obj
as.sort == ['r1', 'r2'] #=> true

Complex example:

pipe = Ractor.new do
  loop do
    Ractor.yield Ractor.receive
  end
end

RN = 10
rs = RN.times.map{|i|
  Ractor.new pipe, i do |pipe, i|
    msg = pipe.take
    msg # ping-pong
  end
}
RN.times{|i|
  pipe << i
}
RN.times.map{
  r, n = Ractor.select(*rs)
  rs.delete r
  n
}.sort #=> [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

Multiple Ractors can send to one Ractor.

# Create 10 ractors and they send objects to pipe ractor.
# pipe ractor yield received objects

pipe = Ractor.new do
  loop do
    Ractor.yield Ractor.receive
  end
end

RN = 10
rs = RN.times.map{|i|
  Ractor.new pipe, i do |pipe, i|
    pipe << i
  end
}

RN.times.map{
  pipe.take
}.sort #=> [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]

TODO: Current Ractor.select() has the same issue of select(2), so this interface should be refined.

TODO: select syntax of go-language uses round-robin technique to make fair scheduling. Now Ractor.select() doesn't use it.

Closing Ractor's ports

Example (try to take from closed Ractor):

r = Ractor.new do
  'finish'
end
r.take # success (will return 'finish')
begin
  o = r.take # try to take from closed Ractor
rescue Ractor::ClosedError
  'ok'
else
  "ng: #{o}"
end

Example (try to send to closed (terminated) Ractor):

r = Ractor.new do
end

r.take # wait terminate

begin
  r.send(1)
rescue Ractor::ClosedError
  'ok'
else
  'ng'
end

When multiple Ractors are waiting for Ractor.yield(), Ractor#close_outgoing will cancel all blocking by raising an exception (ClosedError).

Send a message by copying

Ractor#send(obj) or Ractor.yield(obj) copy obj deeply if obj is an unshareable object.

obj = 'str'.dup
r = Ractor.new obj do |msg|
  # return received msg's object_id
  msg.object_id
end

obj.object_id == r.take #=> false

Some objects are not supported to copy the value, and raise an exception.

obj = Thread.new{}
begin
  Ractor.new obj do |msg|
    msg
  end
rescue TypeError => e
  e.message #=> #<TypeError: allocator undefined for Thread>
else
  'ng' # unreachable here
end

Send a message by moving

Ractor#send(obj, move: true) or Ractor.yield(obj, move: true) move obj to the destination Ractor. If the source Ractor touches the moved object (for example, call the method like obj.foo()), it will be an error.

# move with Ractor#send
r = Ractor.new do
  obj = Ractor.receive
  obj << ' world'
end

str = 'hello'
r.send str, move: true
modified = r.take #=> 'hello world'

# str is moved, and accessing str from this Ractor is prohibited

begin
  # Error because it touches moved str.
  str << ' exception' # raise Ractor::MovedError
rescue Ractor::MovedError
  modified #=> 'hello world'
else
  raise 'unreachable'
end
# move with Ractor.yield
r = Ractor.new do
  obj = 'hello'
  Ractor.yield obj, move: true
  obj << 'world'  # raise Ractor::MovedError
end

str = r.take
begin
  r.take
rescue Ractor::RemoteError
  p str #=> "hello"
end

Some objects are not supported to move, and an exception will be raised.

r = Ractor.new do
  Ractor.receive
end

r.send(Thread.new{}, move: true) #=> allocator undefined for Thread (TypeError)

To achieve the access prohibition for moved objects, class replacement technique is used to implement it.

Shareable objects

The following objects are shareable.

Implementation: Now shareable objects (RVALUE) have FL_SHAREABLE flag. This flag can be added lazily.

To make shareable objects, Ractor.make_shareable(obj) method is provided. In this case, try to make sharaeble by freezing obj and recursively traversable objects. This method accepts copy: keyword (default value is false).Ractor.make_shareable(obj, copy: true) tries to make a deep copy of obj and make the copied object shareable.

Language changes to isolate unshareable objects between Ractors

To isolate unshareable objects between Ractors, we introduced additional language semantics on multi-Ractor Ruby programs.

Note that without using Ractors, these additional semantics is not needed (100% compatible with Ruby 2).

Global variables

Only the main Ractor (a Ractor created at starting of interpreter) can access global variables.

$gv = 1
r = Ractor.new do
  $gv
end

begin
  r.take
rescue Ractor::RemoteError => e
  e.cause.message #=> 'can not access global variables from non-main Ractors'
end

Note that some special global variables are ractor-local, like $stdin, $stdout, $stderr. See [Bug #17268] for more details.

Instance variables of shareable objects

Instance variables of classes/modules can be get from non-main Ractors if the referring values are shareable objects.

class C
  @iv = 1
end

p Ractor.new do
  class C
     @iv
  end
end.take #=> 1

Otherwise, only the main Ractor can access instance variables of shareable objects.

class C
  @iv = [] # unshareable object
end

Ractor.new do
  class C
    begin
      p @iv
    rescue Ractor::IsolationError
      p $!.message
      #=> "can not get unshareable values from instance variables of classes/modules from non-main Ractors"
    end

    begin
      @iv = 42
    rescue Ractor::IsolationError
      p $!.message
      #=> "can not set instance variables of classes/modules by non-main Ractors"
    end
  end
end.take
shared = Ractor.new{}
shared.instance_variable_set(:@iv, 'str')

r = Ractor.new shared do |shared|
  p shared.instance_variable_get(:@iv)
end

begin
  r.take
rescue Ractor::RemoteError => e
  e.cause.message #=> can not access instance variables of shareable objects from non-main Ractors (Ractor::IsolationError)
end

Note that instance variables for class/module objects are also prohibited on Ractors.

Class variables

Only the main Ractor can access class variables.

class C
  @@cv = 'str'
end

r = Ractor.new do
  class C
    p @@cv
  end
end


begin
  r.take
rescue => e
  e.class #=> Ractor::IsolationError
end

Constants

Only the main Ractor can read constants which refer to the unshareable object.

class C
  CONST = 'str'
end
r = Ractor.new do
  C::CONST
end
begin
  r.take
rescue => e
  e.class #=> Ractor::IsolationError
end

Only the main Ractor can define constants which refer to the unshareable object.

class C
end
r = Ractor.new do
  C::CONST = 'str'
end
begin
  r.take
rescue => e
  e.class #=> Ractor::IsolationError
end

To make multi-ractor supported library, the constants should only refer shareable objects.

TABLE = {a: 'ko1', b: 'ko2', c: 'ko3'}

In this case, TABLE references an unshareable Hash object. So that other ractors can not refer TABLE constant. To make it shareable, we can use Ractor.make_shareable() like that.

TABLE = Ractor.make_shareable( {a: 'ko1', b: 'ko2', c: 'ko3'} )

To make it easy, Ruby 3.0 introduced new shareable_constant_value Directive.

# shareable_constant_value: literal

TABLE = {a: 'ko1', b: 'ko2', c: 'ko3'}
#=> Same as: TABLE = Ractor.make_shareable( {a: 'ko1', b: 'ko2', c: 'ko3'} )

shareable_constant_value directive accepts the following modes (descriptions use the example: CONST = expr):

Except the none mode (default), it is guaranteed that the assigned constants refer to only shareable objects.

See doc/syntax/comments.rdoc for more details.

Implementation note

Examples

Traditional Ring example in Actor-model

RN = 1_000
CR = Ractor.current

r = Ractor.new do
  p Ractor.receive
  CR << :fin
end

RN.times{
  r = Ractor.new r do |next_r|
    next_r << Ractor.receive
  end
}

p :setup_ok
r << 1
p Ractor.receive

Fork-join

def fib n
  if n < 2
    1
  else
    fib(n-2) + fib(n-1)
  end
end

RN = 10
rs = (1..RN).map do |i|
  Ractor.new i do |i|
    [i, fib(i)]
  end
end

until rs.empty?
  r, v = Ractor.select(*rs)
  rs.delete r
  p answer: v
end

Worker pool

require 'prime'

pipe = Ractor.new do
  loop do
    Ractor.yield Ractor.receive
  end
end

N = 1000
RN = 10
workers = (1..RN).map do
  Ractor.new pipe do |pipe|
    while n = pipe.take
      Ractor.yield [n, n.prime?]
    end
  end
end

(1..N).each{|i|
  pipe << i
}

pp (1..N).map{
  _r, (n, b) = Ractor.select(*workers)
  [n, b]
}.sort_by{|(n, b)| n}

Pipeline

# pipeline with yield/take
r1 = Ractor.new do
  'r1'
end

r2 = Ractor.new r1 do |r1|
  r1.take + 'r2'
end

r3 = Ractor.new r2 do |r2|
  r2.take + 'r3'
end

p r3.take #=> 'r1r2r3'
# pipeline with send/receive

r3 = Ractor.new Ractor.current do |cr|
  cr.send Ractor.receive + 'r3'
end

r2 = Ractor.new r3 do |r3|
  r3.send Ractor.receive + 'r2'
end

r1 = Ractor.new r2 do |r2|
  r2.send Ractor.receive + 'r1'
end

r1 << 'r0'
p Ractor.receive #=> "r0r1r2r3"

Supervise

# ring example again

r = Ractor.current
(1..10).map{|i|
  r = Ractor.new r, i do |r, i|
    r.send Ractor.receive + "r#{i}"
  end
}

r.send "r0"
p Ractor.receive #=> "r0r10r9r8r7r6r5r4r3r2r1"
# ring example with an error

r = Ractor.current
rs = (1..10).map{|i|
  r = Ractor.new r, i do |r, i|
    loop do
      msg = Ractor.receive
      raise if /e/ =~ msg
      r.send msg + "r#{i}"
    end
  end
}

r.send "r0"
p Ractor.receive #=> "r0r10r9r8r7r6r5r4r3r2r1"
r.send "r0"
p Ractor.select(*rs, Ractor.current) #=> [:receive, "r0r10r9r8r7r6r5r4r3r2r1"]
r.send "e0"
p Ractor.select(*rs, Ractor.current)
#=>
#<Thread:0x000056262de28bd8 run> terminated with exception (report_on_exception is true):
Traceback (most recent call last):
        2: from /home/ko1/src/ruby/trunk/test.rb:7:in `block (2 levels) in <main>'
        1: from /home/ko1/src/ruby/trunk/test.rb:7:in `loop'
/home/ko1/src/ruby/trunk/test.rb:9:in `block (3 levels) in <main>': unhandled exception
Traceback (most recent call last):
        2: from /home/ko1/src/ruby/trunk/test.rb:7:in `block (2 levels) in <main>'
        1: from /home/ko1/src/ruby/trunk/test.rb:7:in `loop'
/home/ko1/src/ruby/trunk/test.rb:9:in `block (3 levels) in <main>': unhandled exception
        1: from /home/ko1/src/ruby/trunk/test.rb:21:in `<main>'
<internal:ractor>:69:in `select': thrown by remote Ractor. (Ractor::RemoteError)
# resend non-error message

r = Ractor.current
rs = (1..10).map{|i|
  r = Ractor.new r, i do |r, i|
    loop do
      msg = Ractor.receive
      raise if /e/ =~ msg
      r.send msg + "r#{i}"
    end
  end
}

r.send "r0"
p Ractor.receive #=> "r0r10r9r8r7r6r5r4r3r2r1"
r.send "r0"
p Ractor.select(*rs, Ractor.current)
[:receive, "r0r10r9r8r7r6r5r4r3r2r1"]
msg = 'e0'
begin
  r.send msg
  p Ractor.select(*rs, Ractor.current)
rescue Ractor::RemoteError
  msg = 'r0'
  retry
end

#=> <internal:ractor>:100:in `send': The incoming-port is already closed (Ractor::ClosedError)
# because r == r[-1] is terminated.
# ring example with supervisor and re-start

def make_ractor r, i
  Ractor.new r, i do |r, i|
    loop do
      msg = Ractor.receive
      raise if /e/ =~ msg
      r.send msg + "r#{i}"
    end
  end
end

r = Ractor.current
rs = (1..10).map{|i|
  r = make_ractor(r, i)
}

msg = 'e0' # error causing message
begin
  r.send msg
  p Ractor.select(*rs, Ractor.current)
rescue Ractor::RemoteError
  r = rs[-1] = make_ractor(rs[-2], rs.size-1)
  msg = 'x0'
  retry
end

#=> [:receive, "x0r9r9r8r7r6r5r4r3r2r1"]