On Wednesday, December 26, 2012 8:22:24 PM UTC+1, Jean-Pierre Flori wrote:
>
>
>
> On Wednesday, December 26, 2012 8:20:03 PM UTC+1, Jean-Pierre Flori wrote:
>>
>>
>>
>> On Wednesday, December 26, 2012 8:08:24 PM UTC+1, Jean-Pierre Flori wrote:
>>>
>>>
>>>
>>> On Wednesday, December 26, 2012 6:49:52 PM UTC+1, Jean-Pierre Flori
>>> wrote:
>>>>
>>>> The offending object seems to be a weakref which is not refcounted
>>>> correctly.
>>>> Not sure which yet.
>>>>
>>> Not sure why, but it might be the one initialized by the Expect class at
>>> the line:
>>> quit.expect_objects.append(weakref.ref(self))
>>>
>>> If I add:
>>> import gc
>>> print gc.get_referrers(quit.expect_objects[-1])
>>>
>>> I get:
>>> [Mathematica, [<weakref at 0x21c07c0; to 'Mathematica' at 0x21bd490>]]
>>>
>>> Strange that Mathematica points to its weakref.
>>> The second member is quit.expect_object I assume.
>>>
>> Really strange.
>> If I del the weakref and then mathematica (from
>> sage.interfaces.mathematica), everything is fine (until it fails later in
>> InfinityRing).
>> But if I try to del mathematica directly, then bang!
>>
> (Using sys.getrefcount shows there is no problem until the call to del)
>
In fact it may be a Cython misuse (or bug?).
The following minimal example seems to reproduce the issue:
In some .pxd file:
cdef class CFake:
cdef object __weakref__
In some .py file:
import gc, weakref, quit
from sage.structure.cfake import CFake
class Fake(CFake):
def __init__(self):
L.append(weakref.ref(self))
a=Fake()
import gc
print gc.get_referrers(quit.expect_objects[-1])
import sys
print sys.getrefcount(quit.expect_objects[-1])
gc.collect()
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To post to this group, send email to sage-devel@googlegroups.com.
To unsubscribe from this group, send email to
sage-devel+unsubscr...@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel?hl=en.
