now (and CMF 1.5+ does not work that way anyway).

Merged branches/tiran-zpt-pydefer r29702:30399 into trunk (including manual merge because I renamed the original branch)

- A quote too much

After consulting with Mark Hammond, removed the explicit calls to SvcStop().

That method gets invoked automatically by the services framework when a
stop request is generated.  Calling it explicitly too caused the onStop()
method to get called multiple times.  At least ZRS's Service subclasses
have onStop() methods that can't be called multiple times without raising
exceptions (they shut things down, and stuff like sockets go away the
first time onStop() gets called).

* Implement a file handler with 'rotate'.  This closes the file, attempts
  a rename to {filename}.last, then reopens the original name.  This
  then makes the log available for another process to perform the rotation
  logic on. This log handler is only installed on the Windows
  platform, where you can't rename an open file.

|- Add version_txt target
|- Call version_txt target from install target
|

Major service enhancements.  Service cleanly shuts down child, and if child
fails the tail of the process output (which generally contains a traceback)
is
written to the event log.

Minor tweaks to the Windows build 'clean' process and documentation tweaks.

Don't kill the service if we can't write to the event log

* Register the new signal handlers on Windows

  See comments in WinSignalHandler.py for details.
* As Windows can not rename an open file, when the 'reopen' signal is
  received try and call a 'rotate' method rather than a reopen.  Rotation
  itself is implemented in zLOG.LogHandler

containers. This is necessary to satisfy expectations of python-script
security machinery.

Added a special case for old ZClass records.

on whether a class has an __of__ definition.

- Collector #1750: StructuredText: fixed handling of image URLs
  with query string                                            i
- some more unittests

Keep defer expression like it was coded by Evan. Added new expression type lazy: that works as a lazy initialization expression. Also added some doc strings

branch but no to the trunk

changing the existing behaviour (at least the original
unittests pass).

originally derived from this test.

at the end even if there were test failures and/or errors, and
didn't display any failure/error tracebacks at the end.  Now it
reports failures/errors at the end the same as when -p isn't
passed.

Two of the ZClasses tests are disabled here, because they
run afoul of Zope trunk's security machinery.  Jim needs to look
at them.  A blurb also needs to be added to CHANGES.txt.

Here are checkin msgs from the branch relating to Zope code
(the msgs relating to ZODB code were reproduced in the
ZODB 3.4 branch checkin):

    ------------------------------------------------------------------------
    r29869 | jim | 2005-04-04 07:04:21 -0400 (Mon, 04 Apr 2005) | 6 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.txt

    Added configuration of class factory.

    Use explicit tm for second connection rather than separate thread.

    Added copy test

    ------------------------------------------------------------------------
    r29868 | jim | 2005-04-04 07:03:57 -0400 (Mon, 04 Apr 2005) | 2 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/_pmc.txt

    Added configuration of class factory.

    ------------------------------------------------------------------------
    r29776 | jim | 2005-04-01 06:24:31 -0500 (Fri, 01 Apr 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/OFS/tests/testProductInit.py

    Added clean-up code to remove non-exixtent directoris from products
    path.

    ------------------------------------------------------------------------
    r29775 | jim | 2005-04-01 06:24:29 -0500 (Fri, 01 Apr 2005) | 2 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/tests.py

    Added missing abourt to tear-down code

    ------------------------------------------------------------------------
    r29772 | jim | 2005-04-01 06:24:22 -0500 (Fri, 01 Apr 2005) | 2 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/App/ApplicationManager.py
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/Basic.py
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.py
       M /Zope/branches/jim-fix-zclasses/lib/python/Zope2/App/startup.py

    Removed BoboPOS compatibility code.

    ------------------------------------------------------------------------
    r29201 | jim | 2005-02-18 07:18:05 -0500 (Fri, 18 Feb 2005) | 3 lines
    Changed paths:
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/27.fs
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/27.txt
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/tests.py

    Added a test for reading ZClasses and their instances from Zope 2.7
    databases.

    ------------------------------------------------------------------------
    r29149 | jim | 2005-02-15 07:24:49 -0500 (Tue, 15 Feb 2005) | 2 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.txt
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/tests.py

    Made the tests a bit more thorough.

    ------------------------------------------------------------------------
    r29148 | jim | 2005-02-15 07:24:44 -0500 (Tue, 15 Feb 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/OFS/tests/testAppInitializer.py

    Made the test cleanup more robust by cleaning up any non-existent
    directories found in Products.__path__.

    ------------------------------------------------------------------------
    r29147 | jim | 2005-02-15 07:24:37 -0500 (Tue, 15 Feb 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ExtensionClass/ExtensionClass.h
       M /Zope/branches/jim-fix-zclasses/lib/python/ExtensionClass/_ExtensionClass.c

    Fixed code that assumed that there weren't subclasses of the
    ExtensionClass meta class.

    ------------------------------------------------------------------------
    r29136 | jim | 2005-02-13 11:37:33 -0500 (Sun, 13 Feb 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.py
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.txt
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/_pmc.py
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/tests.py

    Integrated persistent metaclass with ZClasses and got basic ZClass
    test to pass.

    ------------------------------------------------------------------------
    r29135 | jim | 2005-02-13 11:15:15 -0500 (Sun, 13 Feb 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ExtensionClass/_ExtensionClass.c
       M /Zope/branches/jim-fix-zclasses/lib/python/ExtensionClass/tests.py

    Fixed bug in Base __getattro__ that caused __of__ to be missed in
    instances of instances of subclasses (meta classes) of ExtensionClass.

    ------------------------------------------------------------------------
    r29072 | jim | 2005-02-07 07:36:07 -0500 (Mon, 07 Feb 2005) | 5 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/Property.py
       M /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.py

    Fixed some code for registering class changes to avoid the
    registration of data managers that are None.  This code will,
    eventually, go away, since new persistent classes will
    take care of their own registration.

    ------------------------------------------------------------------------
    r29070 | jim | 2005-02-07 07:36:03 -0500 (Mon, 07 Feb 2005) | 3 lines
    Changed paths:
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/ZClass.txt

    Created a basic ZClass test. It still isn't used, because the
    persistent meta class hasn't been integrated yet.

    ------------------------------------------------------------------------
    r29069 | jim | 2005-02-07 07:36:01 -0500 (Mon, 07 Feb 2005) | 3 lines
    Changed paths:
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/_pmc.py
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/_pmc.txt
       A /Zope/branches/jim-fix-zclasses/lib/python/ZClasses/tests.py

    Created a meta class for persistent classes.  This is based very
    loosly on the experimental persistent class code from Zope 3.

    ------------------------------------------------------------------------
    r29068 | jim | 2005-02-07 07:35:58 -0500 (Mon, 07 Feb 2005) | 2 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/ExtensionClass/_ExtensionClass.c

    Made the ExtensionClass meta class subclassible.

    ------------------------------------------------------------------------
    r29065 | jim | 2005-02-07 07:35:52 -0500 (Mon, 07 Feb 2005) | 3 lines
    Changed paths:
       M /Zope/branches/jim-fix-zclasses/lib/python/OFS/tests/testAppInitializer.py

    Added missing cleanup code to return Products.__path__ to it's
    original state.

      - Collector #1721: Fixed handling of an empty indexed_attrs parameter

- Forward port missing change from 2.7 branch. Humm, 1 year, 22 weeks ago, according to: http://thread.gmane.org/gmane.comp.web.zope.cvs/11080

tuple if the class has a non-empty module string.

There's really no point, because created objects will be unreachable
after invalidating old objects.  Also, if a created object is
non-ghostifiable, it will try to load it's state again, and then
either it would fail or it would load non-committed state.

persistent-class support machinery.

Use explicit tm for second connection rather than separate thread.

Added copy test