Issue 2249 - Support UDP sockets in multiplexer.fcgi

multiplexer.fcgi

Index: multiplexer.fcgi
diff --git a/multiplexer.fcgi b/multiplexer.fcgi
index d72b2c18a6d7e04b211354ec3a3e38895f74ed65..34019b821e032080445c0968f2ac4b99a2eaaa57 100755
--- a/multiplexer.fcgi
+++ b/multiplexer.fcgi
@@ -21,10 +21,12 @@ from flup.server.fcgi import WSGIServer
 
 from sitescripts.web import multiplex
 
-bindAddress = None
-if 'FCGI_BIND_ADDRESS' in os.environ:
-  match = re.match(r'^(.*?):(\d+)$', os.environ['FCGI_BIND_ADDRESS'])
-  bindAddress = (match.group(1), int(match.group(2)))
+bindAddress = os.environ.get('FCGI_BIND_ADDRESS', None)

[Sebastian Noack, 2015/04/01 07:43:27]

Nit: .get() defaults to None anyway. So there is no need to specify a custom
default value here.

[mathias, 2015/04/01 08:03:11]

Done.

+if bindAddress:
+  match = re.match(r'^(.*?):(\d+)$', bindAddress)

[Sebastian Noack, 2015/04/01 07:43:27]

Didn't we agree to don't use re.match() anymore in new code due to it's counter
intuitive behavior, but re.search() instead?

[Sebastian Noack, 2015/04/01 08:01:34]

What's about this?

[mathias, 2015/04/01 08:03:11]

Frankly I have no idea what you mean by that.
What behavior? When did that happen? Who is we?

[Sebastian Noack, 2015/04/01 08:11:27]

I'm talking about this discussion:
http://codereview.adblockplus.org/5809999318089728/patch/5718998062727168/568...

re.match(...) is equivalent to re.search('^' + ...). And both Felix and Wladimir
were irritated by that behavior, that it is implicitly matching start but not
end of string. So we decided to don't use it anymore.

+  if match:
+    bindAddress = (match.group(1), int(match.group(2)))
+
 srv = WSGIServer(multiplex, debug=False, bindAddress=bindAddress)
 
 if __name__ == '__main__':