Author tim.peters
Recipients Mark.Shannon, benjamin.peterson, christian.heimes, inada.naoki, jdemeyer, larry, lukasz.langa, miss-islington, nascheme, ned.deily, pablogsal, petr.viktorin, pitrou, tim.peters, vstinner
Date 2019-10-15.00:19:05
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1571098745.91.0.653023783074.issue38006@roundup.psfhosted.org>
In-reply-to
Content
> I'm often amazed it works at all, let alone perfectly. ;-P

Indeed!  Every time I take a break from gc and come back, I burn another hour wondering why it doesn't recycle _everything_ ;-)

> But what happens if the GC doesn't see that WZ is trash?
> Then it will not be cleared.  Hang it off Y so the GC
> can't find it.

In that case it won't think C is trash either (a weakref holds a strong reference to its callback), so C won't be tp_clear'ed - if WZ isn't in the unreachable set, C can't be either.

> We can set things up so that Z is freed before WZ (e.g.
> WZ in a second and newer cycle).  Then, the callback
> might still run.

Since C hasn't been damaged, and nothing reachable from C would have been considered trash either, I don't see a problem.

> On further thought, this seems safe (but maybe broken) because
> of the handle_weakrefs() logic.  The GC will think WZ is going
> to outlive Z so it will call it before doing any tp_clear calls.

Don't think that applies.  As above, WZ isn't in the unreachable set, so gc knows nothing about it.

Z will eventually be reclaimed via refcounting side effect, C _will_ run, and then WZ, and then C, will be reclaimed by refcounting.

Or am I missing something?
History
Date User Action Args
2019-10-15 00:19:06tim.peterssetrecipients: + tim.peters, nascheme, pitrou, vstinner, larry, christian.heimes, benjamin.peterson, ned.deily, petr.viktorin, inada.naoki, lukasz.langa, Mark.Shannon, jdemeyer, pablogsal, miss-islington
2019-10-15 00:19:05tim.peterssetmessageid: <1571098745.91.0.653023783074.issue38006@roundup.psfhosted.org>
2019-10-15 00:19:05tim.peterslinkissue38006 messages
2019-10-15 00:19:05tim.peterscreate