ddetommaso
commented
4 years ago As it is right now it seems working so I'll merge with master, but we'll need to rethink the concept of task_id.
As far as I see it's not an information of the Task itself but more a tracking id of the logger. In fact the id start with 0 for all the tasks, then it changes once the task starts.
... 2020-11-04 08:45:02,071 - DEBUG - T2: Starting task foo2() by MainThread_0_T1_0 I am alive! ...
Although the information of the parent, the parent_id has to be more clear
Reasons: get_native_id is not useful when the thread has not started yet, so tracking a dependency tree for threads becomes problematic also, for readability - unique and consecutive id is quicker to match when the console output is moving I think we can move to (some_thread).native_id from (some_thread).name if we overhaul the launching a bit and make the Task also carry over the proper native_id... but again, they will not be unique and will get reused by the system
the dictionary config is temporary, should be moved actually to a couple of optional configfiles, in order to allow for example a rolling config with names "pykron-20201103-1518.log"