From @zoffixznet

This appears to hang more often than not, despite printing "Killing" messages.

perl6 -e 'await ^2 .map​: {start { with Proc​::Async.new​: $*EXECUTABLE, "-e", "sleep" -> $p { Promise.in(2).then​: {say "Killing"; $p.kill​: SIGTERM}; await $p.start } }}'

However, if we reduce it to just 1 Proc, then it appears to work without issues, killing the spawned Proc every time​:

perl6 -e 'await ^1 .map​: {start { with Proc​::Async.new​: $*EXECUTABLE, "-e", "sleep" -> $p { Promise.in(2).then​: {say "Killing"; $p.kill​: SIGTERM}; await $p.start } }}'

This looks to be the code handling the kill op, but I don't know if it looks wrong or not​:

https://github.com/MoarVM/MoarVM/blob/dff6a4198f44d0b2793863cc56cc41679c36677f/src/io/procops.c#L1072-L1085

void MVM_proc_kill_async(MVMThreadContext *tc, MVMObject *handle_obj, MVMint64 signal) {
  /* Ensure it's a handle for a process. */
  if (REPR(handle_obj)->ID == MVM_REPR_ID_MVMOSHandle) {
  MVMOSHandle *handle = (MVMOSHandle *)handle_obj;
  if (handle->body.ops == &proc_op_table) {
  /* It's fine; send the kill by cancelling the task. */
  MVMIOAsyncProcessData *data = (MVMIOAsyncProcessData *)handle->body.data;
  data->signal = signal;
  MVM_io_eventloop_cancel_work(tc, data->async_task, NULL, NULL);
  return;
  }
  }
  MVM_exception_throw_adhoc(tc, "killprocasync requires a process handle");
}

From @zoffixznet

FWIW, resolving this ticket will automatically resolve this roast Issue​: Raku/roast#158

From @zoffixznet

On Thu, 01 Jun 2017 18​:42​:03 -0700, cpan@​zoffix.com wrote​:

This appears to hang more often than not, despite printing "Killing"
messages.

perl6 -e 'await ^2 .map​: {start { with Proc​::Async.new​: $*EXECUTABLE,
"-e", "sleep" -> $p { Promise.in(2).then​: {say "Killing"; $p.kill​:
SIGTERM}; await $p.start } }}'

However, if we reduce it to just 1 Proc, then it appears to work
without issues, killing the spawned Proc every time​:

perl6 -e 'await ^1 .map​: {start { with Proc​::Async.new​: $*EXECUTABLE,
"-e", "sleep" -> $p { Promise.in(2).then​: {say "Killing"; $p.kill​:
SIGTERM}; await $p.start } }}'

This looks to be the code handling the kill op, but I don't know if it
looks wrong or not​:

https://github.com/MoarVM/MoarVM/blob/dff6a4198f44d0b2793863cc56cc41679c36677f/src/io/procops.c#L1072-
L1085

void MVM_proc_kill_async(MVMThreadContext *tc, MVMObject *handle_obj,
MVMint64 signal) {
/* Ensure it's a handle for a process. */
if (REPR(handle_obj)->ID == MVM_REPR_ID_MVMOSHandle) {
MVMOSHandle *handle = (MVMOSHandle *)handle_obj;
if (handle->body.ops == &proc_op_table) {
/* It's fine; send the kill by cancelling the task. */
MVMIOAsyncProcessData *data = (MVMIOAsyncProcessData
*)handle->body.data;
data->signal = signal;
MVM_io_eventloop_cancel_work(tc, data->async_task, NULL,
NULL);
return;
}
}
MVM_exception_throw_adhoc(tc, "killprocasync requires a process
handle");
}

Thank you for the report. This is now fixed.

Fix​: rakudo/rakudo@99421d4caa
Test​: Raku/roast@e484c9f27e

@zoffixznet - Status changed from 'new' to 'resolved'