couple of potential stack overflows, including bug #110615.

closes patch #101238

fork.  This solves the test_fork1 problem.  (ceval.c, signalmodule.c,
intrcheck.c)

SourceForge: [ Patch #101226 ] make threading fork-safe

Again due to the duplicate copies of test_support, the checked-in
"expected output" file actually contains verbose-mode output.

subtle breakage on Windows (the test is skipped here, but the TestSkipped
exception wasn't recognized as such, because of duplicate copies of
test_support got loaded; so the test looks like a failure under Windows
instead of a skip).
Repaired the import, but

        THIS TEST *WILL* FAIL ON OTHER SYSTEMS NOW!

Again due to the duplicate copies of test_support, the checked-in
"expected output" file actually contains verbose-mode output.  I can't
generate the *correct* non-verbose output on my system.  So, somebody
please do that.

Windows "inconsistent linkage" warnings at the same time.  I agree
with Mark Hammond that the whole DL_IMPORT/DL_EXPORT macro system
needs an overhaul; this is just an expedient hack until then.

The known bug (bogus error message when an empty file is
extracted) is fixed.

Other changes:

- The target-compile and target-optimize flags of bdist_wininst
  are gone. It is no longer possible to compile the python
  files during installation.
- The zlib module is no longer required or used by bdist_wininst.

- I moved the decompression/extraction code into a separate
  file (extract.c).

- The installer stub is now compressed by UPX (see
  http://upx.tsx.org/). This reduces the size of the exe
  (and thus the overhead of the final installer program)
  from 40 kB to 16 kB.

- The installer displays a more uptodate user wizard-like
  user interface, also containing a graphic: Just's Python Powered logo.
  (I could not convince myself to use one of the BeOpen logos).
- The installation progress bar now moves correctly.

don't assign None to any attributes of the Extension object.

Add some test cases.

file.writelines() now tries to emulate the behaviour of file.write()
as closely as possible. Due to the problems with releasing the
interpreter lock the solution isn't exactly optimal, but still better
than not supporting the file.write() semantics at all.

to be extended somewhat -- right now it only tests the .writelines()
method.

How-to document for using Python for CGI scripting. Not complete and untested, but I'm checking it in so it won't get lost.

GetDialogWindow() returns an existing window.
Added a comment that the methodchain Dialogs->Windows is suspect under Carbon/MacOSX, but we need it under Carbon. No solution yet:-(

Window objects now also have an AutoDispose funcpointer (set for our windows, cleared for foreign windows). Needed mainly for Carbon (where we don't know about the windows belonging to our dialogs).
Fixed a few calls that return an ExistingWindow.

- Renamed socket to _socket.

Don't use our own malloc any more. It made test_longexp crash due to its optimization to lock blocks into a specific size malloc requests. This loses 15% performance, but that we'll have to live with:-(

Cheaper implementation of PyOS_CheckStack: only call StackSpace once and keep a sentinel in a static variable.

language found in the environment variable, contributed by James
Henstridge.

useless.  So the test of the dcgettext() function is removed.

useless.  So the category argument on _find() is removed, as is the
dcgettext() function.

cvs tree.  It creates the directory and gettext.mo file on the fly,
from the base64 encode binary data.