installed Python. So we don't use os.execve any longer, which means
we need an actual executable in <myapp>.app/Contents/MacOS. For applets
we make a symlink to the Python executable used to build the applet,
for standalone apps we simply copy it.
- Added support for the new any_missing_maybe() feature of modulefinder.py,
which is pending as patch #643711. Its use is optional so it still works
with the existing version of modulefinder.py

consistency with the built-in open() (and every other sane open()
function, for that matter).  The two valid ways to call this open() are
now open(mode) and open(device, mode).

For backwards compatibility, retain the old open(mode) calling syntax --
this makes the error message when you call open(device) a bit confusing,
but oh well.

This is the first half of SF patch #644977.

Various edits

the results of *setting* __name__ are not so surprising.

If people can suggest more tests, that'd be grand, or is what's there
sufficient?

Convert test/README to reST

see problems with my code that I didn't see before the checkin, but:

When a subtype .mro() fails, we need to reset the type whose __bases__
are being changed, too.  Fix + test.

mro() methods.  Now any exception aborts the whole __bases__ change.

And more tests.

   of using the CCompiler.find_library_file() provided by the Distutils.
   This patch fixes it to use the Distutils method at the cost of some
   additional glue.

(The duplication resulted in the SSL module not being automatically
built on Macs; the Distutils knew that shared libraries on OS X have a
.dylib extension, but the setup.py code didn't.)

Fill in the missing details. (From Jack Jansen)

Constructor accepts optional keyword arguments after a optional items list.
Add fromkeys() as an alternate constructor from an iterable over keys.
Expand related unittests.

method for constructing new dictionaries from sequences of keys.

new-style classes.
Closes SF bug #637941.

Avoid depending on this in the test.

use regrtest, so it might be interesting.

allow the use of any resource except bsddb, give the option
"-uall,-bsddb".

  Fixed minor bug preventing documentation compilation.

It's odd that gcc on my ibook didn't complain about this.

This adds a --relative option to the bdist_dumb command that defaults
to false; if true, the .tar.gz or .zip will be assembled using relative
paths.

[ 635933 ] make some type attrs writable

but should have been!

News about the above.

[ 635933 ] make some type attrs writable

Plus a couple of extra tests beyond what's up there.

It hasn't been as carefully reviewed as it perhaps should, so all readers
are encouraged, nay exhorted, to give this a close reading.

There are still a couple of oddities related to assigning to __name__,
but I intend to solicit python-dev's opinions on these.

Mark writes in private email:

  "Modules listed in the registry was a dumb idea.  This whole scheme
  can die.  AFAIK, no one in the world uses it (including win32all
  since the last build)."

(See also SF #643711)

- requires modulefinder.py to be on sys.path
- does *not* work for Python.framework (yet), only for static builds