Moved test-centric windows specific fix into the class itself to assure this kind of issue doesn't popup for anyone

Author: Byron

lib/git/remote.py

@@ -25,6 +25,7 @@
 						join,
 						)
 import re
+import os
 
 __all__ = ('RemoteProgress', 'PushInfo', 'FetchInfo', 'Remote')
 
@@ -418,6 +419,16 @@ def __init__(self, repo, name):
 		self.repo = repo
 		self.name = name
 
+		if os.name == 'nt':
+			# some oddity: on windows, python 2.5, it for some reason does not realize
+			# that it has the config_writer property, but instead calls __getattr__
+			# which will not yield the expected results. 'pinging' the members
+			# with a dir call creates the config_writer property that we require 
+			# ... bugs like these make me wonder wheter python really wants to be used
+			# for production. It doesn't happen on linux though.
+			dir(self)
+		# END windows special handling
+		
 	def __getattr__(self, attr):
 		"""Allows to call this instance like 
 		remote.special( *args, **kwargs) to call git-remote special self.name"""

test/testlib/helper.py

@@ -188,13 +188,6 @@ def remote_repo_creator(self):
 			d_remote.fetch()
 			remote_repo_url = "git://localhost%s" % remote_repo_dir
 
-			# some oddity: on windows, python 2.5, it for some reason does not realize
-			# that it has the config_writer property, but instead calls __getattr__
-			# which will not yield the expected results. 'pinging' the members
-			# with a dir call creates the config_writer property that we require 
-			# ... bugs like these make me wonder wheter python really wants to be used
-			# for production. It doesn't happen on linux though.
-			dir(d_remote)
 			d_remote.config_writer.set('url', remote_repo_url)
 
 			# try to list remotes to diagnoes whether the server is up