#15424: A coercion-related memory leak
-----------------------------+-------------------------
Reporter: SimonKing | Owner:
Type: defect | Status: new
Priority: major | Milestone: sage-5.13
Component: memleak | Resolution:
Keywords: | Merged in:
Authors: | Reviewers:
Report Upstream: N/A | Work issues:
Branch: | Commit:
Dependencies: | Stopgaps:
-----------------------------+-------------------------
Comment (by nbruin):
I tried to see where the reference can be coming from:
{{{
%cpaste
import gc
from sage.structure.coerce_dict import *
def all_referrers(c,X):
found_IDs=set()
R=[]
new=gc.get_referrers(c)
found_IDs.add(id(new))
found_IDs.add(id(R))
found_IDs.add(id(globals()))
while len(new)>0:
r=new.pop()
if id(r) in found_IDs or type(r) not in X:
print "skipping",type(r)
continue
R.append(r)
found_IDs.add(id(r))
new.extend(gc.get_referrers(r))
print "type(r)=%s len(R)=%s len(new)=%s"%(type(r),len(R),len(new))
return R
def getR():
K = IntegerModRing(111115)
C = type(K)
phi = K.coerce_map_from(ZZ)
del phi
#vary this line:
x = K.one()*2
#del K,x
_ = gc.collect()
X=set([list,dict,tuple,
RingHomset_generic_with_category,
sage.rings.finite_rings.integer_mod.Integer_to_IntegerMod,
TripleDict, MonoDict, C])
R=all_referrers(list(c for c in gc.get_objects() if
isinstance(c,C))[0],X)
return R
--
RingHomset_generic_with_category=type(ZZ.Hom(QQ))
R=getR()
}}}
and I found that with `x=K.one()*2` there is also a !TripleDict that show
up. It's fairly big (44 entries) and the entries all seem to be of the
form `D[domain,codomain,None]=morphism from domain to codomain` (or `None`
entries). Since the morphism has a strong reference to the codomain, this
would keep our ring alive [it doesn't seem like the kind of dictionary
that can afford to be weak on its values]. Judging from the entries, this
dictionary is a global one. I think there is a ticket somewhere that
mentions the same phenomenon (also doing arithmetic), in the context of
finite fields. Somewhere from the era when we started to work on #715 in
earnest.
--
Ticket URL: <http://trac.sagemath.org/ticket/15424#comment:1>
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 unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/groups/opt_out.
