Message153355
| Author |
neologix |
| Recipients |
Aaron.Staley, amaury.forgeotdarc, neologix, pitrou |
| Date |
2012年02月14日.18:10:30 |
| SpamBayes Score |
3.322041e-08 |
| Marked as misclassified |
No |
| Message-id |
<CAH_1eM2JAEO40YBDK7ikQ43q9vXHwsuuhT3m5Lac4cJ2bN_5uw@mail.gmail.com> |
| In-reply-to |
<1329192115.13.0.211613762365.issue13992@psf.upfronthosting.co.za> |
| Content |
I'm running out of ideas to debug this, maybe Antoine or Amaury can help :-)
One last idea (not sure it will work though):
If Channel's finalizer gets called twice, inside Channel.__del__, you
could save a string representation of the current backtrace as an
attribute, and print the previous and current backtrace is there is
already one such attribute (i.e. print the backtrace if __del__ is
called twice).
There may be false positives (for example a finalizer resurecting the
object would end up in __del__ being called twice), but it might yield
useful info.
You might also want to save the thread ID and time.
But this is really strange... |
|