On Thu, Oct 7, 2010 at 22:01, Tim McNamara <[email protected]> wrote: > On 8 October 2010 03:34, Martin Langhoff <[email protected]> wrote: >> >> See http://dev.laptop.org/ticket/10386 for details. The sugar-session >> process in 10.1.2 grows slowly... >> >> There's some form of leak somewhere. Maybe we are triggerin a real >> python leak, maybe we have reference loops. How do we trace this? > > This post seems pretty good [1]. It cites a tool that creates an object > graph that visually represents what is happening in memory.[2] > > [1] http://www.lshift.net/blog/2008/11/14/tracing-python-memory-leaks > [2] http://mg.pov.lt/blog/python-object-graphs.html
Those are good links, in this case I suspect those icons, or any other object referenced inside a cycle that also contains an AP icon, are listening for a signal in a longer-lived object, which prevents the whole cycle to be freed. We need to either disconnect from that signal when the icon is no longer needed, or use a weak reference. I will be looking at a similar issue in the next few days but in 0.90, I will know by then. Regards, Tomeu > _______________________________________________ > Sugar-devel mailing list > [email protected] > http://lists.sugarlabs.org/listinfo/sugar-devel > > _______________________________________________ Devel mailing list [email protected] http://lists.laptop.org/listinfo/devel
