Fixed some beus in the change management code that cause no content to be returned when using ZODB 3.

Fixed a bug in the copy machinery which prevented instances of ZClasses defined in the methods of another ZClass from being pasted and renamed.

how to use ZServer with PCGI.

how to use ZServer with PCGI.

Updated to reflect that ZODB 3 is now the default.

This is insane; copyright.dtml should render CREDITS.txt.

Python methods are passed as context. In this case, the method now
correctly uses the im_self of the method to continue sniffing up
the acquistion hierarchy for local roles.

templates are sometimes pickled without subclassing
Persistence.Persistent.

is missing.

Made it easier to run from the command prompt, for example to
debug.

custom_zodb module from INSTANCE_HOME.

In particular, BASE1 is always the URL of the tiop=lebel published object.
Note that when using ZPublisher, SCRIPT_NAME is '' and
BASE0 and BASE1 are the same.

Changed lookup of if URLx and BASEx to allow x to have multiple
digits.  So you can use URL999999 if your URL is ling enough. ;)

Added has_key and updated keys to work with new URL lookup mechansism.

This is necessary due to oddities in the way that propertysheets
currently manage their data. :(

Don't we always want this to default to true?

Added CreateAFactory argument when calling the real constructor.
This cased adding classes as methods to return no data.

 "It fixes a memory corruption error resulting from BadPickleGet
  exceptions in load_get, load_binget and load_long_binget.  This was
  initially reported on c.l.py as a problem with Cookie.py; see the thread
  titled "python core dump (SIGBUS) on Solaris" for more details."