NEWS.md
$task_info
method to retrieve more detailed information about where a task was run (mrc-4058)collect_timeout
in bulk functions becomes timeout_task_wait
worker_stop_timeout
in the rrq_controller
’s destroy
method becomes timeout_worker_stop
task_timeout
in queuing a task on a separate process, in bulk submission functions becomes timeout_task_run
timeout
in enqueue
becomes timeout_task_run
timeout_task_wait
and option rrq.timeout_task_wait
to control the default time to wait for tasks to be returned from task_wait
and bulk task retrieval methods. The default behaviour is unchanged (blocking indefinitely) but this can now easily be changed at a global or queue-scoped level.task_wait
, task_results
and the bulk interfaces. The default remains not to error.at_front
argument (introduced in rrq 0.2.14), tasks can no longer jump the queue (mrc-4069)timeout_poll
and timeout_die
configurable via the worker config.rrq_
prefix; you must now use rrq_worker_spawn()
(not worker_spawn
), rrq_worker_wait
, rrq_heartbeat
and rrq_heartbeat_kill
(mrc-2682)rrq::rrq_worker_from_config
, which is a simpler way of constructing worker objects. The worker object itself (rrq::rrq_worker
) loses its helper constructor function and is documented (mrc-2297)enqueue
can add a task with dependencies i.e. tasks which must be complete before this task can be run via depends_on
(mrc-2255)separate_process = TRUE
to $enqueue()
. Use this to ensure isolation between tasks (mrc-2068)task_preceeding
function to controller to list tasks in front of a particular task in the queue (vimc-4502)$lapply
and friends restored after being removed during the refactor for version 0.2.0 (mrc-558)$worker_detect_exited
for detecting exited workers when a heartbeat is used (mrc-1231)$task_cancel
if running with a heartbeat enabled (mrc-734)rrq::rrq_task_progress_update
function, which can be called from any task run from rrq
and queried with $task_progress
from a rrq_controller
(mrc-600)