[issue23377] HTTPResponse may drop buffer holding next response

Sun Apr 12 05:02:20 CEST 2015

Martin Panter added the comment:

HTTP pipelining is not supported in general. According to the module doc string, you can actually pipeline a second request if you have read the first response’s headers but not its body:

>>> conn = HTTPSConnection("bugs.python.org")
>>> conn.request("GET", "/")
>>> response1 = conn.getresponse()  # Read header section only
>>> conn.request("GET", "/issue23377")
>>> response1.read()[:100]  # Read first body after pipelining second request
b'<!--\n This is the default body that is displayed when people visit the\n tracker. The tag below lists'
>>> response2 = conn.getresponse()
>>> response2.read()[:100]
b'\n<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xht'

But I suspect this might deadlock with some servers (that don’t properly support pipelining themselves) if the second request was a large POST or something and filled up the buffers that this mode relies on.

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue23377>
_______________________________________