aboutsummaryrefslogtreecommitdiff
path: root/scripts/async-http.py
diff options
context:
space:
mode:
authorRob Austein <sra@hactrn.net>2009-04-29 18:55:48 +0000
committerRob Austein <sra@hactrn.net>2009-04-29 18:55:48 +0000
commit309dad3781b2259d9dccf59a87c9896c91f56ebd (patch)
tree8131a4ba693aac3dbe64b5b1bc1393f26ae15985 /scripts/async-http.py
parent2c7bea62c4e8a8a25c2e0e23033d37820a9165a3 (diff)
Notes on test environments
svn path=/scripts/async-http.py; revision=2375
Diffstat (limited to 'scripts/async-http.py')
-rw-r--r--scripts/async-http.py17
1 files changed, 17 insertions, 0 deletions
diff --git a/scripts/async-http.py b/scripts/async-http.py
index 66c502de..fe7847b7 100644
--- a/scripts/async-http.py
+++ b/scripts/async-http.py
@@ -21,6 +21,23 @@ PERFORMANCE OF THIS SOFTWARE.
# Command that may be useful for testing server side of this:
#
# lynx -post_data -mime_header -source http://127.0.0.1:8000/
+#
+# Testing the client side of this is more entertaining, both because
+# we need to be liberal in what we accept and also because servers do
+# wildly different things depending both on HTTP version number and
+# options signaled by the client and also on internal details in the
+# server context (such as whether the content is static or generated
+# by CGI). It's useful to test against static content, but also to
+# test against CGI-generated output, eg, the following trivial script:
+#
+# print "Content-Type: text/plain; charset=US-ASCII\r"
+# print "\r"
+# for i in xrange(100):
+# print "%08d" % i, "." * 120, "\r"
+#
+# At least with Apache 2.0, result of running this differs
+# significantly depending on whether client signals HTTP 1.0 or 1.1;
+# the latter produces chunked output.
import sys, os, time, socket, asyncore, asynchat, traceback, urlparse
import rpki.async