Raymond Hettinger: > If this is new, pairwise() is the most likely cause. I reproduced the issue without involving the itertools module. It seems to be a zip_next() micro-optimization which keeps an object alive longer than epxected. This micro-optimization is not new. It's just really hard to trigger the issue. It seems to depend if the zip object is deleted or not. The builtin zip type implements a traverse function which visits Py_VISIT(lz->result). The GC is supposed to be able to break such cycle, no?