I've made considerable changes to Michael's code, specifically to use
the select() system call directly and to store the timeout as a C
double instead of a Python object; internally, -1.0 (or anything
negative) represents the None from the API.

I'm not 100% sure that all corner cases are covered correctly, so
please keep an eye on this.  Next I'm going to try it Windows before
Tim complains.

No way is this a bugfix candidate. :-)

otherwise this code could segfault:

  from socket import socket
  s = socket.__new__(socket)
  s.recv(100)

Straightforward fix.  Will backport to 2.2.  If there's ever a new 2.1
release, this could be backported there too (since it's an issue with
anything that's got both a __reduce__ and a __setstate__).

Straightforward fix.  Will backport to 2.2.  If there's ever a new 2.1
release, this could be backported there too (since it's an issue with
anything that's got both a __reduce__ and a __setstate__).

This uses cgi.parse_header() in Checker.checkforhtml(), so that
webchecker recognises the mime type text/html even if options
are specified.

Made conversion failure error messages consistent between types.
Added related unittests.

The specific warning is that clean didn't find a directory that should
be removed if it exists.

out how to just activate my code on platforms where I know it works
(currently only linux/x86).

equivalent attributes in a few more spots.

This closes SF patch http://www.python.org/sf/562373

Also, added more regression tests to cover the new type and test its
conformity with range().

Change default for get() back to None.
Will backport to 2.2.1.

Closes SF patch # 560250.

Bugfix candidate IMO.

Fix a bunch of multiline string constants that used +.

Call ensure_string() with one arg too, since the second value passed
was the default.

Make all returns explicit and rearrange logic to avoid extra
indentation.

mode is not computed in dry_run mode, so it can't be included in the
log message.