leaking in the symbolic ring

[videlec]

The following operation creates a leak

pi * I * RR.one()

This can be tested with

import gc

def num_complex():
    gc.collect()
    return sum(1 for t in gc.get_objects() if type(t) is sage.rings.complex_number.ComplexNumber)

for _ in range(10):
    print num_complex()
    for _ in range(10):
        a = pi * I * RR.one()

Original report on sage-devel.

Depends on #24668

CC: @rwst

Component: memleak

Author: Ralf Stephan

Branch/Commit: a42d589

Reviewer: Jeroen Demeyer

Issue created by migration from https://trac.sagemath.org/ticket/24745

[rwst]

comment:1

Thanks for the excellent diagnostics. I will commit a Pynac patch which depends on #24668 (0.7.16). There was a superfluous Py_INCREF in the in-place multiplication code.

[rwst]

Branch: u/rws/leaking_in_the_symbolic_ring

[rwst]

Author: Ralf Stephan

[rwst]

comment:3

Note you need beta6 (or a copy of pynac-0.7.16 in upstream/) to test this.

---

New commits:

f5b3de4	24668: version/chzksum
67f9ddc	24668: doctest fixes
2568492	24668: fix doctest
9b2e16c	Merge branch 't/24668/upgrade_to_pynac_0_7_16' into t/24745/leaking_in_the_symbolic_ring
a42d589	24745: pynac patch: fix memleaks in in-place PyObject numerics

[rwst]

Commit: a42d589

[jdemeyer]

Dependencies: #24668

[jdemeyer]

Reviewer: Jeroen Demeyer

[vbraun]

Changed branch from u/rws/leaking_in_the_symbolic_ring to a42d589