[Python-Dev] Evil reference cycles caused Exception.traceback (original) (raw)

Nick Coghlan ncoghlan at gmail.com
Mon Sep 18 06:35:02 EDT 2017

On 18 September 2017 at 20:18, Victor Stinner <victor.stinner at gmail.com> wrote:

2017-09-18 12:07 GMT+02:00 Nick Coghlan <ncoghlan at gmail.com>:

I wonder if it might be reasonable to have tracebacks only hold a weak reference to their frame objects when "debug == False". Please don't change the Python behaviour depending on debug, or it will be a nightmare to debug it :-( ("Why does it work on my computer?")

Yeah, that's always a challenge :)

Rather than being thread local or context local state, whether or not to keep the full frames in the traceback could be a yet another setting on the exception object, whereby we tweaked the logic that drops the reference at the end of an except clause as follows:

  1. If the exception ref count == 1, just drop the reference and let the traceback die with it
  2. If the exception ref count > 1, check for a new "keep_locals" attribute
  3. If it's true, just drop the reference and otherwise leave the exception alone
  4. If it's false, recursively clear all the already terminated frames in traceback, context, and cause

For now, we'd default to "keep_locals = True", but we'd attempt to figure out some way to migrate to the default being "keep_locals = False".

Cheers, Nick.

-- Nick Coghlan | ncoghlan at gmail.com | Brisbane, Australia

More information about the Python-Dev mailing list