class Thread

Threads are the Ruby implementation for a concurrent programming model.

Programs that require multiple threads of execution are a perfect candidate for Ruby’s Thread class.

For example, we can create a new thread separate from the main thread’s execution using ::new.

thr = Thread.new { puts "What's the big deal" }

Then we are able to pause the execution of the main thread and allow our new thread to finish, using join:

thr.join #=> "What's the big deal"

If we don’t call thr.join before the main thread terminates, then all other threads including thr will be killed.

Alternatively, you can use an array for handling multiple threads at once, like in the following example:

threads = []
threads << Thread.new { puts "What's the big deal" }
threads << Thread.new { 3.times { puts "Threads are fun!" } }

After creating a few threads we wait for them all to finish consecutively.

threads.each { |thr| thr.join }

To retrieve the last value of a thread, use value

thr = Thread.new { sleep 1; "Useful value" }
thr.value #=> "Useful value"

`Thread` initialization¶ ↑

In order to create new threads, Ruby provides ::new, ::start, and ::fork. A block must be provided with each of these methods, otherwise a ThreadError will be raised.

When subclassing the Thread class, the initialize method of your subclass will be ignored by ::start and ::fork. Otherwise, be sure to call super in your initialize method.

`Thread` termination¶ ↑

For terminating threads, Ruby provides a variety of ways to do this.

The class method ::kill, is meant to exit a given thread:

thr = Thread.new { sleep }
Thread.kill(thr) # sends exit() to thr

Alternatively, you can use the instance method exit, or any of its aliases kill or terminate.

thr.exit

`Thread` status¶ ↑

Ruby provides a few instance methods for querying the state of a given thread. To get a string with the current thread’s state use status

thr = Thread.new { sleep }
thr.status # => "sleep"
thr.exit
thr.status # => false

You can also use alive? to tell if the thread is running or sleeping, and stop? if the thread is dead or sleeping.

`Thread` variables and scope¶ ↑

Since threads are created with blocks, the same rules apply to other Ruby blocks for variable scope. Any local variables created within this block are accessible to only this thread.

Fiber-local vs. Thread-local¶ ↑

Each fiber has its own bucket for Thread#[] storage. When you set a new fiber-local it is only accessible within this Fiber. To illustrate:

Thread.new {
  Thread.current[:foo] = "bar"
  Fiber.new {
    p Thread.current[:foo] # => nil
  }.resume
}.join

This example uses [] for getting and []= for setting fiber-locals, you can also use keys to list the fiber-locals for a given thread and key? to check if a fiber-local exists.

When it comes to thread-locals, they are accessible within the entire scope of the thread. Given the following example:

Thread.new{
  Thread.current.thread_variable_set(:foo, 1)
  p Thread.current.thread_variable_get(:foo) # => 1
  Fiber.new{
    Thread.current.thread_variable_set(:foo, 2)
    p Thread.current.thread_variable_get(:foo) # => 2
  }.resume
  p Thread.current.thread_variable_get(:foo)   # => 2
}.join

You can see that the thread-local :foo carried over into the fiber and was changed to 2 by the end of the thread.

This example makes use of thread_variable_set to create new thread-locals, and thread_variable_get to reference them.

There is also thread_variables to list all thread-locals, and thread_variable? to check if a given thread-local exists.

When an unhandled exception is raised inside a thread, it will terminate. By default, this exception will not propagate to other threads. The exception is stored and when another thread calls value or join, the exception will be re-raised in that thread.

t = Thread.new{ raise 'something went wrong' }
t.value #=> RuntimeError: something went wrong

An exception can be raised from outside the thread using the Thread#raise instance method, which takes the same parameters as Kernel#raise.

Setting Thread.abort_on_exception = true, Thread#abort_on_exception = true, or $DEBUG = true will cause a subsequent unhandled exception raised in a thread to be automatically re-raised in the main thread.

With the addition of the class method ::handle_interrupt, you can now handle exceptions asynchronously with threads.

Scheduling¶ ↑

Ruby provides a few ways to support scheduling threads in your program.

The first way is by using the class method ::stop, to put the current running thread to sleep and schedule the execution of another thread.

Once a thread is asleep, you can use the instance method wakeup to mark your thread as eligible for scheduling.

You can also try ::pass, which attempts to pass execution to another thread but is dependent on the OS whether a running thread will switch or not. The same goes for priority, which lets you hint to the thread scheduler which threads you want to take precedence when passing execution. This method is also dependent on the OS and may be ignored on some platforms.

Public Class Methods

abort_on_exception → true or false

Returns the status of the global “abort on exception” condition.

The default is false.

When set to true, if any thread is aborted by an exception, the raised exception will be re-raised in the main thread.

Can also be specified by the global $DEBUG flag or command line option -d.

NOTE¶ ↑

Asynchronous interrupts are difficult to use.

If you need to communicate between threads, please consider to use another way such as Queue.

Or use them with deep understanding about this method.

Usage¶ ↑

In this example, we can guard from Thread#raise exceptions.

Using the :never TimingSymbol the RuntimeError exception will always be ignored in the first block of the main thread. In the second ::handle_interrupt block we can purposefully handle RuntimeError exceptions.

th = Thread.new do
  Thread.handle_interrupt(RuntimeError => :never) {
    begin
      # You can write resource allocation code safely.
      Thread.handle_interrupt(RuntimeError => :immediate) {
        # ...
      }
    ensure
      # You can write resource deallocation code safely.
    end
  }
end
Thread.pass
# ...
th.raise "stop"

While we are ignoring the RuntimeError exception, it’s safe to write our resource allocation code. Then, the ensure block is where we can safely deallocate your resources.

Stack control settings¶ ↑

It’s possible to stack multiple levels of ::handle_interrupt blocks in order to control more than one ExceptionClass and TimingSymbol at a time.

Thread.handle_interrupt(FooError => :never) {
  Thread.handle_interrupt(BarError => :never) {
     # FooError and BarError are prohibited.
  }
}

Inheritance with ExceptionClass¶ ↑

All exceptions inherited from the ExceptionClass parameter will be considered.

Thread.handle_interrupt(Exception => :never) {
  # all exceptions inherited from Exception are prohibited.
}

For handling all interrupts, use Object and not Exception as the ExceptionClass, as kill/terminate interrupts are not handled by Exception.

static VALUE
rb_thread_s_handle_interrupt(VALUE self, VALUE mask_arg)
{
    VALUE mask = Qundef;
    rb_execution_context_t * volatile ec = GET_EC();
    rb_thread_t * volatile th = rb_ec_thread_ptr(ec);
    volatile VALUE r = Qnil;
    enum ruby_tag_type state;

    if (!rb_block_given_p()) {
        rb_raise(rb_eArgError, "block is needed.");
    }

    mask_arg = rb_to_hash_type(mask_arg);

    if (OBJ_FROZEN(mask_arg) && rb_hash_compare_by_id_p(mask_arg)) {
        mask = Qnil;
    }

    rb_hash_foreach(mask_arg, handle_interrupt_arg_check_i, (VALUE)&mask);

    if (UNDEF_P(mask)) {
        return rb_yield(Qnil);
    }

    if (!RTEST(mask)) {
        mask = mask_arg;
    }
    else if (RB_TYPE_P(mask, T_HASH)) {
        OBJ_FREEZE(mask);
    }

    rb_ary_push(th->pending_interrupt_mask_stack, mask);
    if (!rb_threadptr_pending_interrupt_empty_p(th)) {
        th->pending_interrupt_queue_checked = 0;
        RUBY_VM_SET_INTERRUPT(th->ec);
    }

    EC_PUSH_TAG(th->ec);
    if ((state = EC_EXEC_TAG()) == TAG_NONE) {
        r = rb_yield(Qnil);
    }
    EC_POP_TAG();

    rb_ary_pop(th->pending_interrupt_mask_stack);
    if (!rb_threadptr_pending_interrupt_empty_p(th)) {
        th->pending_interrupt_queue_checked = 0;
        RUBY_VM_SET_INTERRUPT(th->ec);
    }

    RUBY_VM_CHECK_INTS(th->ec);

    if (state) {
        EC_JUMP_TAG(th->ec, state);
    }

    return r;
}

ignore_deadlock → true or false

Returns the status of the global “ignore deadlock” condition. The default is false, so that deadlock conditions are not ignored.

Usage¶ ↑

th = Thread.new{
  Thread.handle_interrupt(RuntimeError => :on_blocking){
    while true
      ...
      # reach safe point to invoke interrupt
      if Thread.pending_interrupt?
        Thread.handle_interrupt(Object => :immediate){}
      end
      ...
    end
  }
}
...
th.raise # stop thread

This example can also be written as the following, which you should use to avoid asynchronous interrupts.

flag = true
th = Thread.new{
  Thread.handle_interrupt(RuntimeError => :on_blocking){
    while true
      ...
      # reach safe point to invoke interrupt
      break if flag == false
      ...
    end
  }
}
...
flag = false # stop thread

static VALUE
rb_thread_s_pending_interrupt_p(int argc, VALUE *argv, VALUE self)
{
    return rb_thread_pending_interrupt_p(argc, argv, GET_THREAD()->self);
}

report_on_exception → true or false

Returns the status of the global “report on exception” condition.

The default is true since Ruby 2.5.

All threads created when this flag is true will report a message on $stderr if an exception kills the thread.

Thread.new { 1.times { raise } }

will produce this output on $stderr:

#<Thread:...> terminated with exception (report_on_exception is true):
Traceback (most recent call last):
        2: from -e:1:in `block in <main>'
        1: from -e:1:in `times'

This is done to catch errors in threads early. In some cases, you might not want this output. There are multiple ways to avoid the extra output:

If the exception is not intended, the best is to fix the cause of the exception so it does not happen anymore.
If the exception is intended, it might be better to rescue it closer to where it is raised rather then let it kill the Thread.
If it is guaranteed the Thread will be joined with Thread#join or Thread#value, then it is safe to disable this report with Thread.current.report_on_exception = false when starting the Thread. However, this might handle the exception much later, or not at all if the Thread is never joined due to the parent thread being blocked, etc.

Public Instance Methods

thr[sym] → obj or nil

Attribute Reference—Returns the value of a fiber-local variable (current thread’s root fiber if not explicitly inside a Fiber), using either a symbol or a string name. If the specified variable does not exist, returns nil.

[
  Thread.new { Thread.current["name"] = "A" },
  Thread.new { Thread.current[:name]  = "B" },
  Thread.new { Thread.current["name"] = "C" }
].each do |th|
  th.join
  puts "#{th.inspect}: #{th[:name]}"
end

This will produce:

#<Thread:0x00000002a54220 dead>: A
#<Thread:0x00000002a541a8 dead>: B
#<Thread:0x00000002a54130 dead>: C

Thread#[] and Thread#[]= are not thread-local but fiber-local. This confusion did not exist in Ruby 1.8 because fibers are only available since Ruby 1.9. Ruby 1.9 chooses that the methods behaves fiber-local to save following idiom for dynamic scope.

def meth(newvalue)
  begin
    oldvalue = Thread.current[:name]
    Thread.current[:name] = newvalue
    yield
  ensure
    Thread.current[:name] = oldvalue
  end
end

The idiom may not work as dynamic scope if the methods are thread-local and a given block switches fiber.

f = Fiber.new {
  meth(1) {
    Fiber.yield
  }
}
meth(2) {
  f.resume
}
f.resume
p Thread.current[:name]
#=> nil if fiber-local
#=> 2 if thread-local (The value 2 is leaked to outside of meth method.)

For thread-local variables, please see thread_variable_get and thread_variable_set.

static VALUE
rb_thread_aref(VALUE thread, VALUE key)
{
    ID id = rb_check_id(&key);
    if (!id) return Qnil;
    return rb_thread_local_aref(thread, id);
}

thr[sym] = obj → obj

Attribute Assignment—Sets or creates the value of a fiber-local variable, using either a symbol or a string.

See also stop? and status.

static VALUE
rb_thread_alive_p(VALUE thread)
{
    return RBOOL(!thread_finished(rb_thread_ptr(thread)));
}

backtrace → array or nil

Returns the current backtrace of the target thread.

static VALUE
rb_thread_backtrace_m(int argc, VALUE *argv, VALUE thval)
{
    return rb_vm_thread_backtrace(argc, argv, thval);
}

backtrace_locations(*args) → array or nil

Returns the execution stack for the target thread—an array containing backtrace location objects.

See Thread::Backtrace::Location for more information.

This method behaves similarly to Kernel#caller_locations except it applies to a specific thread.

static VALUE
rb_thread_backtrace_locations_m(int argc, VALUE *argv, VALUE thval)
{
    return rb_vm_thread_backtrace_locations(argc, argv, thval);
}

exit → thr

Terminates thr and schedules another thread to be run, returning the terminated Thread. If this is the main thread, or the last thread, exits the process.

Alias for: kill

fetch(sym) → obj

fetch(sym) { } → obj

fetch(sym, default) → obj

Returns a fiber-local for the given key. If the key can’t be found, there are several options: With no other arguments, it will raise a KeyError exception; if default is given, then that will be returned; if the optional code block is specified, then that will be run and its result returned. See Thread#[] and Hash#fetch.

static VALUE
rb_thread_fetch(int argc, VALUE *argv, VALUE self)
{
    VALUE key, val;
    ID id;
    rb_thread_t *target_th = rb_thread_ptr(self);
    int block_given;

    rb_check_arity(argc, 1, 2);
    key = argv[0];

    block_given = rb_block_given_p();
    if (block_given && argc == 2) {
        rb_warn("block supersedes default value argument");
    }

    id = rb_check_id(&key);

    if (id == recursive_key) {
        return target_th->ec->local_storage_recursive_hash;
    }
    else if (id && target_th->ec->local_storage &&
             rb_id_table_lookup(target_th->ec->local_storage, id, &val)) {
        return val;
    }
    else if (block_given) {
        return rb_yield(key);
    }
    else if (argc == 1) {
        rb_key_err_raise(rb_sprintf("key not found: %+"PRIsVALUE, key), self, key);
    }
    else {
        return argv[1];
    }
}

group → thgrp or nil

Returns the ThreadGroup which contains the given thread.

Thread.main.group   #=> #<ThreadGroup:0x4029d914>

VALUE
rb_thread_group(VALUE thread)
{
    return rb_thread_ptr(thread)->thgroup;
}

inspect

Alias for: to_s

join → thr

join(limit) → thr

The calling thread will suspend execution and run this thr.

Does not return until thr exits or until the given limit seconds have passed.

If the time limit expires, nil will be returned, otherwise thr is returned.

Any threads not joined will be killed when the main program exits.

If thr had previously raised an exception and the ::abort_on_exception or $DEBUG flags are not set, (so the exception has not yet been processed), it will be processed at this time.

a = Thread.new { print "a"; sleep(10); print "b"; print "c" }
x = Thread.new { print "x"; Thread.pass; print "y"; print "z" }
x.join # Let thread x finish, thread a will be killed on exit.
#=> "axyz"

The following example illustrates the limit parameter.

y = Thread.new { 4.times { sleep 0.1; puts 'tick... ' }}
puts "Waiting" until y.join(0.15)

This will produce:

tick...
Waiting
tick...
Waiting
tick...
tick...

static VALUE
thread_join_m(int argc, VALUE *argv, VALUE self)
{
    VALUE timeout = Qnil;
    rb_hrtime_t rel = 0, *limit = 0;

    if (rb_check_arity(argc, 0, 1)) {
        timeout = argv[0];
    }

    // Convert the timeout eagerly, so it's always converted and deterministic
    /*
     * This supports INFINITY and negative values, so we can't use
     * rb_time_interval right now...
     */
    if (NIL_P(timeout)) {
        /* unlimited */
    }
    else if (FIXNUM_P(timeout)) {
        rel = rb_sec2hrtime(NUM2TIMET(timeout));
        limit = &rel;
    }
    else {
        limit = double2hrtime(&rel, rb_num2dbl(timeout));
    }

    return thread_join(rb_thread_ptr(self), timeout, limit);
}

key?(sym) → true or false

Returns true if the given string (or symbol) exists as a fiber-local variable.

me = Thread.current
me[:oliver] = "a"
me.key?(:oliver)    #=> true
me.key?(:stanley)   #=> false

static VALUE
rb_thread_key_p(VALUE self, VALUE key)
{
    VALUE val;
    ID id = rb_check_id(&key);
    struct rb_id_table *local_storage = rb_thread_ptr(self)->ec->local_storage;

    if (!id || local_storage == NULL) {
        return Qfalse;
    }
    return RBOOL(rb_id_table_lookup(local_storage, id, &val));
}

keys → array

Returns an array of the names of the fiber-local variables (as Symbols).

thr = Thread.new do
  Thread.current[:cat] = 'meow'
  Thread.current["dog"] = 'woof'
end
thr.join   #=> #<Thread:0x401b3f10 dead>
thr.keys   #=> [:dog, :cat]

static VALUE
rb_thread_keys(VALUE self)
{
    struct rb_id_table *local_storage = rb_thread_ptr(self)->ec->local_storage;
    VALUE ary = rb_ary_new();

    if (local_storage) {
        rb_id_table_foreach(local_storage, thread_keys_i, (void *)ary);
    }
    return ary;
}

kill → thr

Terminates thr and schedules another thread to be run, returning the terminated Thread. If this is the main thread, or the last thread, exits the process.

VALUE
rb_thread_kill(VALUE thread)
{
    rb_thread_t *target_th = rb_thread_ptr(thread);

    if (target_th->to_kill || target_th->status == THREAD_KILLED) {
        return thread;
    }
    if (target_th == target_th->vm->ractor.main_thread) {
        rb_exit(EXIT_SUCCESS);
    }

    RUBY_DEBUG_LOG("target_th:%u", rb_th_serial(target_th));

    if (target_th == GET_THREAD()) {
        /* kill myself immediately */
        rb_threadptr_to_kill(target_th);
    }
    else {
        threadptr_check_pending_interrupt_queue(target_th);
        rb_threadptr_pending_interrupt_enque(target_th, RUBY_FATAL_THREAD_KILLED);
        rb_threadptr_interrupt(target_th);
    }

    return thread;
}

Also aliased as: terminate, exit

name → string

show the name of the thread.

static VALUE
rb_thread_getname(VALUE thread)
{
    return rb_thread_ptr(thread)->name;
}

name=(name) → string

set given name to the ruby thread. On some platform, it may set the name to pthread and/or kernel.

static VALUE
rb_thread_setname(VALUE thread, VALUE name)
{
    rb_thread_t *target_th = rb_thread_ptr(thread);

    if (!NIL_P(name)) {
        rb_encoding *enc;
        StringValueCStr(name);
        enc = rb_enc_get(name);
        if (!rb_enc_asciicompat(enc)) {
            rb_raise(rb_eArgError, "ASCII incompatible encoding (%s)",
                     rb_enc_name(enc));
        }
        name = rb_str_new_frozen(name);
    }
    target_th->name = name;
    if (threadptr_initialized(target_th) && target_th->has_dedicated_nt) {
        native_set_another_thread_name(target_th->nt->thread_id, name);
    }
    return name;
}

native_thread_id → integer

Return the native thread ID which is used by the Ruby thread.

The ID depends on the OS. (not POSIX thread ID returned by pthread_self(3))

On Linux it is TID returned by gettid(2).
On macOS it is the system-wide unique integral ID of thread returned by pthread_threadid_np(3).
On FreeBSD it is the unique integral ID of the thread returned by pthread_getthreadid_np(3).
On Windows it is the thread identifier returned by GetThreadId().
On other platforms, it raises NotImplementedError.

NOTE: If the thread is not associated yet or already deassociated with a native thread, it returns nil. If the Ruby implementation uses M:N thread model, the ID may change depending on the timing.

static VALUE
rb_thread_native_thread_id(VALUE thread)
{
    rb_thread_t *target_th = rb_thread_ptr(thread);
    if (rb_threadptr_dead(target_th)) return Qnil;
    return native_thread_native_thread_id(target_th);
}

pending_interrupt?(error = nil) → true/false

Returns whether or not the asynchronous queue is empty for the target thread.

If error is given, then check only for error type deferred events.

See ::pending_interrupt? for more information.

static VALUE
rb_thread_pending_interrupt_p(int argc, VALUE *argv, VALUE target_thread)
{
    rb_thread_t *target_th = rb_thread_ptr(target_thread);

    if (!target_th->pending_interrupt_queue) {
        return Qfalse;
    }
    if (rb_threadptr_pending_interrupt_empty_p(target_th)) {
        return Qfalse;
    }
    if (rb_check_arity(argc, 0, 1)) {
        VALUE err = argv[0];
        if (!rb_obj_is_kind_of(err, rb_cModule)) {
            rb_raise(rb_eTypeError, "class or module required for rescue clause");
        }
        return RBOOL(rb_threadptr_pending_interrupt_include_p(target_th, err));
    }
    else {
        return Qtrue;
    }
}

priority → integer

Returns the priority of thr. Default is inherited from the current thread which creating the new thread, or zero for the initial main thread; higher-priority thread will run more frequently than lower-priority threads (but lower-priority threads can also run).

This is just hint for Ruby thread scheduler. It may be ignored on some platform.

Thread.current.priority   #=> 0

static VALUE
rb_thread_priority(VALUE thread)
{
    return INT2NUM(rb_thread_ptr(thread)->priority);
}

priority= integer → thr

Sets the priority of thr to integer. Higher-priority threads will run more frequently than lower-priority threads (but lower-priority threads can also run).

This is just hint for Ruby thread scheduler. It may be ignored on some platform.

count1 = count2 = 0
a = Thread.new do
      loop { count1 += 1 }
    end
a.priority = -1

b = Thread.new do
      loop { count2 += 1 }
    end
b.priority = -2
sleep 1   #=> 1
count1    #=> 622504
count2    #=> 5832

static VALUE
rb_thread_priority_set(VALUE thread, VALUE prio)
{
    rb_thread_t *target_th = rb_thread_ptr(thread);
    int priority;

#if USE_NATIVE_THREAD_PRIORITY
    target_th->priority = NUM2INT(prio);
    native_thread_apply_priority(th);
#else
    priority = NUM2INT(prio);
    if (priority > RUBY_THREAD_PRIORITY_MAX) {
        priority = RUBY_THREAD_PRIORITY_MAX;
    }
    else if (priority < RUBY_THREAD_PRIORITY_MIN) {
        priority = RUBY_THREAD_PRIORITY_MIN;
    }
    target_th->priority = (int8_t)priority;
#endif
    return INT2NUM(target_th->priority);
}

raise

raise(string)

raise(exception [, string [, array]])

Raises an exception from the given thread. The caller does not have to be thr. See Kernel#raise for more information.

Thread.abort_on_exception = true
a = Thread.new { sleep(200) }
a.raise("Gotcha")

This will produce:

prog.rb:3: Gotcha (RuntimeError)
 from prog.rb:2:in `initialize'
 from prog.rb:2:in `new'
 from prog.rb:2

static VALUE
thread_raise_m(int argc, VALUE *argv, VALUE self)
{
    rb_thread_t *target_th = rb_thread_ptr(self);
    const rb_thread_t *current_th = GET_THREAD();

    threadptr_check_pending_interrupt_queue(target_th);
    rb_threadptr_raise(target_th, argc, argv);

    /* To perform Thread.current.raise as Kernel.raise */
    if (current_th == target_th) {
        RUBY_VM_CHECK_INTS(target_th->ec);
    }
    return Qnil;
}

report_on_exception → true or false

Returns the status of the thread-local “report on exception” condition for this thr.

The default value when creating a Thread is the value of the global flag Thread.report_on_exception.

class Thread

Thread initialization¶ ↑

Thread termination¶ ↑

Thread status¶ ↑

Thread variables and scope¶ ↑

Fiber-local vs. Thread-local¶ ↑

Exception handling¶ ↑

Scheduling¶ ↑

NOTE¶ ↑

Usage¶ ↑

Stack control settings¶ ↑

Inheritance with ExceptionClass¶ ↑

Usage¶ ↑

`Thread` initialization¶ ↑

`Thread` termination¶ ↑

`Thread` status¶ ↑

`Thread` variables and scope¶ ↑

`Exception` handling¶ ↑