#12313: Fix yet another memory leak caused by caching of coercion data
---------------------------------------------------------------------------+
Reporter: SimonKing |
Owner:
Type: defect |
Status: needs_work
Priority: major |
Milestone: sage-5.8
Component: memleak |
Resolution:
Keywords: coercion weak dictionary |
Work issues:
Report Upstream: N/A |
Reviewers: Simon King, Jean-Pierre Flori, John Perry, Nils Bruin
Authors: Simon King, Jean-Pierre Flori |
Merged in:
Dependencies: #715, #11521, #12215, #13746, #13378; merge with #13387 |
Stopgaps:
---------------------------------------------------------------------------+
Changes (by SimonKing):
* status: positive_review => needs_work
Old description:
> The following could not be fixed in #715:
>
> {{{
> sage: K = GF(1<<55,'t')
> sage: for i in range(50):
> ....: a = K.random_element()
> ....: E = EllipticCurve(j=a)
> ....: b = K.has_coerce_map_from(E)
> ....:
> sage: import gc
> sage: gc.collect()
> 0
> }}}
> The problem is that any parent has a dictionary that stores any coerce
> map (and a different dictionary for conversion maps) ending at this
> parent. The keys are given by the domains of the maps. So, in the example
> above, the field `K` has an attribute that is a dictionary whose keys are
> the different elliptic curves.
>
> In coercion, it is usually best to compare parents not by equality but by
> identity. Therefore, I suggest to implement a container called `MonoDict`
> that works similar to the new weak `TripleDict` (see #715), but takes a
> single item as a key.
>
> First tests show that one can actually gain a lot of speed: `MonoDict`
> can access its items much faster than a usual dictionary.
>
> __Apply__
>
> * [attachment:trac_12313-mono_dict-combined-random-sk.patch]
New description:
The following could not be fixed in #715:
{{{
sage: K = GF(1<<55,'t')
sage: for i in range(50):
....: a = K.random_element()
....: E = EllipticCurve(j=a)
....: b = K.has_coerce_map_from(E)
....:
sage: import gc
sage: gc.collect()
0
}}}
The problem is that any parent has a dictionary that stores any coerce map
(and a different dictionary for conversion maps) ending at this parent.
The keys are given by the domains of the maps. So, in the example above,
the field `K` has an attribute that is a dictionary whose keys are the
different elliptic curves.
In coercion, it is usually best to compare parents not by equality but by
identity. Therefore, I suggest to implement a container called `MonoDict`
that works similar to the new weak `TripleDict` (see #715), but takes a
single item as a key.
First tests show that one can actually gain a lot of speed: `MonoDict` can
access its items much faster than a usual dictionary.
__Apply__
* [attachment:trac_12313-mono_dict-combined-random-sk.patch] (apparently
merged in sage-5.8.beta0)
* [attachment:trac_12313-revert_callback_from_11521.patch]
--
Comment:
I have added a patch, so that no callback is used for homsets. The
rationale is given in comment:322.
For the patchbot (in an attempt to trick it, because the main patch seems
to be in 5.8.beta0...):
Apply trac_12313-revert_callback_from_11521.patch to sage-5.8.beta0
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/12313#comment:325>
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?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
