Took Nick Coghlan's advice about importing warnings globally in logging, to avoid the possibility of race conditions: "This could deadlock if a thread spawned as a side effect of importing a module happens to trigger a warning. warnings is pulled into sys.modules as part of the interpreter startup - having a global 'import warnings' shouldn't have any real effect on logging's import time."

process on success or raise an exception on error.

support unusual filenames (such as those containing semi-colons) in
Content-Disposition headers.

Issue #4384: Added logging integration with warnings module using captureWarnings(). This change includes a NullHandler which does nothing; it will be of use to library developers who want to avoid the "No handlers could be found for logger XXX" message which can appear if the library user doesn't configure logging.

whether any thread has tracing turned on, which saves one load instruction in
the fast_next_opcode path in PyEval_EvalFrameEx().  See issue 4477.

Converted a C99 style comment to a C89 style comment (found by MAL).

"add-and-remove-programs".

This change addresses part of issue 4336.

Change endheaders() to take an optional message_body argument
that is sent along with the headers.  Change xmlrpclib and
httplib's other methods to use this new interface.

It is more efficient to make a single send() call, which should
get the entire client request into one packet (assuming it is
smaller than the MTU) and will avoid the long pause for delayed
ack following timeout.

Also:
- Add a comment about the buffer size for makefile().
- Extract _set_content_length() method and fix whitespace issues there.

Otherwise, you get a warning about an undefined variable.

Merged revisions 67384,67386-67387,67389-67390,67392,67399-67400,67403-67405,67426 via svnmerge from
svn+ssh://pythondev@svn.python.org/sandbox/trunk/2to3/lib2to3

........
  r67384 | benjamin.peterson | 2008-11-25 16:13:31 -0600 (Tue, 25 Nov 2008) | 4 lines

  don't duplicate calls to start_tree()

  RefactoringTool.pre_order values now holds a list of the fixers while pre_order_mapping holds the dict.
........
  r67386 | benjamin.peterson | 2008-11-25 16:44:52 -0600 (Tue, 25 Nov 2008) | 1 line

  #4423 fix_imports was still replacing usage of a module if attributes were being used
........
  r67387 | benjamin.peterson | 2008-11-25 16:47:54 -0600 (Tue, 25 Nov 2008) | 1 line

  fix broken test
........
  r67389 | benjamin.peterson | 2008-11-25 17:13:17 -0600 (Tue, 25 Nov 2008) | 1 line

  remove compatibility code; we only cater to 2.5+
........
  r67390 | benjamin.peterson | 2008-11-25 22:03:36 -0600 (Tue, 25 Nov 2008) | 1 line

  fix #3994; the usage of changed imports was fixed in nested cases
........
  r67392 | benjamin.peterson | 2008-11-26 11:11:40 -0600 (Wed, 26 Nov 2008) | 1 line

  simpilfy and comment fix_imports
........
  r67399 | benjamin.peterson | 2008-11-26 11:47:03 -0600 (Wed, 26 Nov 2008) | 1 line

  remove more compatibility code
........
  r67400 | benjamin.peterson | 2008-11-26 12:07:41 -0600 (Wed, 26 Nov 2008) | 1 line

  set svn:ignore
........
  r67403 | benjamin.peterson | 2008-11-26 13:11:11 -0600 (Wed, 26 Nov 2008) | 1 line

  wrap import
........
  r67404 | benjamin.peterson | 2008-11-26 13:29:49 -0600 (Wed, 26 Nov 2008) | 1 line

  build the fix_imports pattern in compile_pattern, so MAPPING can be changed and reflected in the pattern
........
  r67405 | benjamin.peterson | 2008-11-26 14:01:24 -0600 (Wed, 26 Nov 2008) | 1 line

  stop ugly messages about runtime errors being from printed
........
  r67426 | benjamin.peterson | 2008-11-28 16:01:40 -0600 (Fri, 28 Nov 2008) | 5 lines

  don't replace a module name if it is in the middle of a attribute lookup

  This fix also stops module names from being replaced if they are not in an attribute lookup.
........

Retain copyright of processing examples. This was requested by a Debian maintainer during packaging of the multiprocessing package for 2.4/2.5