Is it possible to enable scalene profiling for specific subprocess(es)?

[dralley]

We have a task running engine inside our code which spawns new processes like so:

task_process = multiprocessing.Process(target=_perform_task, args=(task.pk, task_working_dir_rel_path))
task_process.start()

It spawns new processes because it needs to parse externally-hosted data, sometimes substantial quantities of it, using native libraries, so there are both security and performance reasons.

It would be kind of nice if we could enable a flag inside of our app which would turn on automatic scalene profiling of tasks and dump the results in a unique temporary directory. Each task would then have its own separate profile which we could look at. The problem is that we don't want to be running our entire app under scalene all the time in production.

Is it completely not-possible to enable scalene purely from within the code at runtime, without running the code itself using scalene? The docs give the impression that:

from scalene import scalene_profiler

# Turn profiling on
scalene_profiler.start()

# Turn profiling off
scalene_profiler.stop()

still requires running the code using scalene.

[dralley]

@emeryberger