I have code that already handles an “gzip” encoded response from urlopen(). All three patches leave the Content-Encoding header intact, so I suspect my code would try to decompress the body a second time. Deleting this header (as already suggested) would work for me.