#11339: Illegal use of __deallocate__ in cython (pyx) code
-----------------------------------------------+----------------------------
Reporter: gagern | Owner: drkirkby
Type: defect | Status: needs_work
Priority: major | Milestone: sage-4.7.1
Component: porting | Keywords: sd31
Work_issues: | Upstream: N/A
Reviewer: François Bissey, Steven Trogdon | Author: Volker Braun,
Martin von Gagern
Merged: | Dependencies:
-----------------------------------------------+----------------------------
Comment(by vbraun):
The Python docs say "The `tp_clear` member function is used to break
reference cycles in cyclic garbage detected by the garbage collector.", so
there must have been a cycle involving the `GroebnerStrategy` instance.
I looked a bit at the source and we definitely have cycles
ideal<->`GroebnerStrategy`. Both refer to the parent ring, and this
explains why `tp_clear` was called on `GroebnerStrategy`. I haven't found
a place where the parent refers back to the cycle. In this situation
Martin's patch should work as we can get rid of the cycle while the parent
has positive refcount. It could be that Python 2.7 got smarter and deletes
the parent together with the cycle while Python 2.6 first deleted the
cycle and only then noticed that nothing else refers to the parent any
more. This would explain why Martin's patch works but my attempt of
blindly accessing the ring C structure fails on Python 2.7.
There is a lot of stuff that could conceivably create a larger cycle
involving the parent even though I haven't found one. For one, the parent
keeps a reference to the "one" element. Even then, I think it is too
dangerous to make the hidden assumption that there are no complete cycles
involving the parents. Somebody is bound to trip over this sooner or
later.
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/11339#comment:23>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
