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
13222	unexpected HttpResponse iteration behavior	teepark	Aymeric Augustin	"For all the builtin iterables, and most from libraries, the `__iter__` method returns something representing one and only one iteration. `django.http.HttpResponse.__iter__` sets an iterator attribute on itself, then returns self (it also defines a `next` method so it is its own iterator). This is bad behavior for an iterable and causes problems when trying to do anything but basic iteration.

A use case (and where I encountered this issue) is in a WSGI server that pulls off the first item in the iterable app return value, prefixes anything that was sent via `write` calls, and then rebuilds the iterable with `itertools.chain`:

{{{
if prefix:
    body_iterator = iter(body)
    try:
        first_chunk = body_iterator.next()
    except StopIteration:
        first_chunk = ''
    body = itertools.chain((prefix + first_chunk,), body_iterator)
}}}

Using `django.core.handlers.wsgi.WSGIHandler` (which just returns the HttpResponse, since it's iterable) with this server, the first chunk of the HttpResponse gets doubled because `itertools.chain` calls `iter` on its arguments, and that resets the HttpResponse's iterator.

To compare with some builtins and other libraries:

{{{
>>> it = iter([""foobar""])
>>> list(it), list(it)
(['foobar'], [])
>>> it = iter(""foobar"" for i in [1])
>>> list(it), list(it)
(['foobar'], [])
>>> import werkzeug
>>> it = iter(werkzeug.Response(""foobar"").response)
>>> list(it), list(it)
(['foobar'], [])
>>> import webob
>>> it = iter(webob.Response(""foobar"").app_iter)
>>> list(it), list(it)
(['foobar'], [])
>>> import django.http
>>> it = iter(django.http.HttpResponse(""foobar""))
>>> list(it), list(it)
(['foobar'], ['foobar'])
}}}"	Bug	closed	HTTP handling	1.3-rc	Normal	fixed			Ready for checkin	1	0	0	0	0	0