id summary reporter owner description type status component version severity resolution keywords cc stage has_patch needs_docs needs_tests needs_better_patch easy ui_ux 6308 QueryDict won't release objects between requests when request.POST is copy() ed Yasushi Masuda Yasushi Masuda "Hi, There seems to be the population of container argument in QueryDict.__deepcopy__(), which should cause incremental (and unreleased) population of POST data between requests. Here is the relevant code in django.http.__init__.py: {{{ 174 def __deepcopy__(self, memo={}): 175 import copy 176 result = self.__class__('', mutable=True) 177 memo[id(self)] = result 178 for key, value in dict.items(self): 179 dict.__setitem__(result, copy.deepcopy(key, memo), copy.deepcopy(value, memo)) }}} The behaviour can be easily demonstrated as follows: {{{ from django.conf.urls.defaults import * from django.http import HttpResponse, QueryDict from django.template import Template, Context def demo(request): request.POST.copy() t = Template( '' '' '{% for item in memo %}llama at {{ item.0 }}, {% endfor %}' '') memo = QueryDict.__deepcopy__.im_func.func_defaults[0].items() c = Context({'memo': memo}) return HttpResponse(t.render(c)) urlpatterns = patterns('', url(r'^$', demo)) }}} You should see llamas are populating between requests, and after enormous POST requests, You can see the process size is significantly increased. I think (I'm not sure the reason why memo is directly populated) it can be fixed rewriting __deepcopy__ as follows: {{{ def __deepcopy__(self, memo={}): import copy result = self.__class__('', mutable=True) m = dict(memo.items()) m[id(self)] = result for key, value in dict.items(self): dict.__setitem__(result, copy.deepcopy(key, m), copy.deepcopy(value, m)) return result }}} Thanks." closed Core (Other) dev fixed request.POST.copy, memory issue Unreviewed 0 0 0 0 0 0