Issue #20082 has been updated by zverok (Victor Shepelev). UPD: OTOH, without resuming, the code doesn't see any problem at all (despite the claim about killing other thread's fibers): ```ruby fibers = [] Thread.new { f = Fiber.new { (1..).each { sleep(0.1) } } fibers << f } sleep(0.1) # to make sure the thread have started p fibers.last.kill #=> #<Fiber:0x00007f73f682c078 examples/fiber_kill.rb:7 (terminated)> ``` ---------------------------------------- Bug #20082: Killing fibers across threads: unexpected exception https://bugs.ruby-lang.org/issues/20082#change-105839 * Author: zverok (Victor Shepelev) * Status: Open * Priority: Normal * Assignee: ioquatix (Samuel Williams) * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN ---------------------------------------- For providing the example in a changelog, I tried to imitate killing fibers belonging to other threads. Documentation [claims](https://docs.ruby-lang.org/en/master/Fiber.html#method-i-kill)

Raises FiberError if called on a fiber belonging to another thread.

So, I created this artificial example to check how it works: ```ruby fibers = [] Thread.new { f = Fiber.new { (1..).each { sleep(0.1) } } fibers << f f.resume } sleep(0.1) # to make sure the thread has started fibers.last.kill ``` The example indeed fails with `FiberError`, but the error message is confusing: ``` in `kill': attempt to resume a resumed fiber (double resume) (FiberError) ``` Is this an expected message for such case? Or am I misunderstanding something? -- https://bugs.ruby-lang.org/

Issue #20082 has been updated by ioquatix (Samuel Williams).

Raises FiberError if called on a fiber belonging to another thread.

I agree the error is a little confusing, but the reality is it's probably okay. "in `kill': attempt to resume a resumed fiber (double resume) (FiberError)" You are literally reaching across into a fiber which is currently resumed and executing sleep. This is a text book definition of a double resume. To get the error message I imagine you are after: ```ruby fibers = [] Thread.new { f = Fiber.new { Fiber.yield } f.resume fibers << f }.join fibers.last.kill ```

without resuming, the code doesn't see any problem at all (despite the claim about killing other thread's fibers):

A fiber is lazily constructed according to the first thread it is resumed on. So, this behaviour looks okay to me. ---------------------------------------- Bug #20082: Killing fibers across threads: unexpected exception https://bugs.ruby-lang.org/issues/20082#change-105858 * Author: zverok (Victor Shepelev) * Status: Open * Priority: Normal * Assignee: ioquatix (Samuel Williams) * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN ---------------------------------------- For providing the example in a changelog, I tried to imitate killing fibers belonging to other threads. Documentation [claims](https://docs.ruby-lang.org/en/master/Fiber.html#method-i-kill)

Raises FiberError if called on a fiber belonging to another thread.

So, I created this artificial example to check how it works: ```ruby fibers = [] Thread.new { f = Fiber.new { (1..).each { sleep(0.1) } } fibers << f f.resume } sleep(0.1) # to make sure the thread has started fibers.last.kill ``` The example indeed fails with `FiberError`, but the error message is confusing: ``` in `kill': attempt to resume a resumed fiber (double resume) (FiberError) ``` Is this an expected message for such case? Or am I misunderstanding something? -- https://bugs.ruby-lang.org/