Created on 2010-10-20 04:01 by James.Bowman, last changed 2010-10-20 06:22 by georg.brandl. This issue is now closed.
| Messages (3) | |||
|---|---|---|---|
| msg119187 - (view) | Author: James Bowman (James.Bowman) | Date: 2010-10-20 04:01 | |
import sys
def foo():
x = [o] * 100
raise ArithmeticError
o = "something"
print sys.getrefcount(o)
try:
foo()
except ArithmeticError:
pass
print sys.getrefcount(o)
-------------------------------------------
Gives:
4
104
Looking at the CPython source, FrameObject's deallocator does actually decrement refcounts of its locals and arguments. Guessing that the FrameObject is not being deallocated.
|
|||
| msg119188 - (view) | Author: Alex Gaynor (alex) *  |
Date: 2010-10-20 04:58 | |
That's because in Python 2.5 at that point in the code sys.exc_info() still points at the traceback (and by extension, the frame) thus x is not deallocated yet. I don't think this is a bug. |
|||
| msg119190 - (view) | Author: Georg Brandl (georg.brandl) * |
Date: 2010-10-20 06:22 | |
Alex is correct. (You can prove that by raising and catching another exception before the second getrefcount().) |
|||
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2010-10-20 06:22:07 | georg.brandl | set | status: open -> closed nosy: + georg.brandl messages: + msg119190 resolution: wont fix |
| 2010-10-20 04:58:36 | alex | set | nosy:
+ alex messages: + msg119188 |
| 2010-10-20 04:01:45 | James.Bowman | create | |