Working on fix for BytesWarning now.

Thanks for updating that PR, I've just added this to my TODO list, but it may take a while.

Hi @berkerpeksag, circling back to this one, do you still have time and interest in the topic, or should we find another reviewer?

Is this going to be merged any time soon?

Sorry, I will try to review this PR this weekend.

any update?

this bug currently blocks any integretion with the linkerd2 proxy as this throws a 400 bad request for connections inititaded by python apps

https://github.com/linkerd/linkerd2/pull/1200/files

It crashes in _tunnel on response line:

Can you show the exact error for this?

«crash» usually describes a segfault of the interpreter, is that what happended or was it a regular Python exception?

Please answer: what exactly do you see when you say «it crashes»?

Hi folks -- would love to try again to get this merged. :-) cc @berkerpeksag who I solicited the original review from, and tagging in @orsenthil and @SethMichaelLarson (I think this is the same GitHub identity for the person who approved #20959, that's who I am looking for, sorry if I tagged the wrong person) who have also reviewed PRs for Lib/http/client.py in the past...

Sigh, I did mean @sethmlarson not @SethMichaelLarson, sorry about that, the GitHub autocompleter was determined to mislead me!

Hi Michael, Thanks for following up. I will be able to review and bring it to a closure.

any updates on this? would love to see this merge as it prevents our services from using certain proxies. thanks

Please merge this PR as HTTP/1.0 does not define CONNECT. It breaks the compatibility with some proxies and it is Python's fault. This PR fixes this bug but it has been delayed for more than a year.

This PR needs to have the fle conflicts resolved and get reviewed by an HTTP domain expert.

A note: I don’t think the module is intended to run on its own for production-grade applications. nginx, waitress, uwsgi are all full-featured capable servers that can run python apps and/or work as reverse proxy.

This PR needs to have the fle conflicts resolved and get reviewed by an HTTP domain expert.

A note: I don’t think the module is intended to run on its own for production-grade applications. nginx, waitress, uwsgi are all full-featured capable servers that can run python apps and/or work as reverse proxy.

Reply to the note above: HTTP CONNECT is also widely used as a client instead of a production-ready server. For example, people who live in a country that blocks some websites needs an proxy client for their daily uses, including writing some Python scripts.

My bad, this PR is for http.client not http.server!

If you want this included the target (versionchanged) needs to be 3.12 now.

It is weird this is still not merged.

Effectively this is a couple of lines change w/o comments and tests:

--- /usr/lib/python3.X/http/client.py
+++ client.py
@@ -855,10 +855,20 @@

         self._tunnel_host, self._tunnel_port = self._get_hostport(host, port)
         if headers:
-            self._tunnel_headers = headers
+            self._tunnel_headers = headers.copy()
         else:
             self._tunnel_headers.clear()

+        saw_host_header = False
+        for header in self._tunnel_headers.keys():
+            if header.lower() == "host":
+                saw_host_header = True
+                break
+        if not saw_host_header:
+            encoded_host = self._tunnel_host.encode("idna").decode("ascii")
+            self._tunnel_headers["Host"] = "%s:%d" % (
+                encoded_host, self._tunnel_port)
+
     def _get_hostport(self, host, port):
         if port is None:
             i = host.rfind(':')
@@ -883,8 +893,8 @@
         self.debuglevel = level

     def _tunnel(self):
-        connect = b"CONNECT %s:%d HTTP/1.0\r\n" % (
-            self._tunnel_host.encode("ascii"), self._tunnel_port)
+        connect = b"CONNECT %s:%d %s\r\n" % (
+            self._tunnel_host.encode("ascii"), self._tunnel_port, self._http_vsn_str.encode("ascii"))
         headers = [connect]
         for header, value in self._tunnel_headers.items():
             headers.append(f"{header}: {value}\r\n".encode("latin-1"))

Thanks for making a progress!

@merwok @orsenthil @ammaraskar @berkerpeksag This PR has been refreshed to be mergeable into the current state of HEAD, and all checks are passing. As you folks are Python organizations members, can any of you help review or provide suggestions as next steps? Thank you for your time! 🙏🏻