Cyclops 2.0

Tim Peters tim_one at
Wed Feb 21 01:06:57 EST 2001

[Robin Becker]
> I suppose a lot of the need for T Peters' Cyclops module has gone away
> with gc in Python 2.0,

Depends on whether you care whether you're creating cycles.  There's less
need to care now, but objects with __del__ methods in cycles still aren't
reclaimed by magic.  I personally care about the latter only to the extent
that I would consider it to be a bug in my code if I ever created such a
thing ("a bug" because I intend never *to* create such a thing).

> but Cyclops provide a convenient framework for the analysis. Is
> there any equivalent for 2.0?

Sure:  Cyclops <0.9 wink>.  Nothing in 2.0 or 2.1 broke it, AFAIK.

> If I just use Cyclops is it telling me a lot more than the
> gc.garbage list?

Depends on *how* you use Cyclops.  It has methods for delivering many kinds
of info, from grouping by connection type to partitioning into strongly
connected components.  It can also be taught how to chase new kinds of
objects without delving into C.  gc.garbage is just a list, and of the kinds
of objects gc knows how to chase on its own.  If "just a list" is all you
need, cool.

More information about the Python-list mailing list