applications to know less about how this is implemented

The comment indicated that end could be None, and I've fleshed out the
comment with more detail.

Move setLastTid() call into block where lock is held.

Remove some unnecessary pack() calls.

abnormalization.

rather than as a (probably) unprintable character.

improve the code coverage for loadBefore() in the BDBFullStorage.

extension

This eliminates a bunch of checks that weren't exercised by the test
suite.  Not sure if they were superstition or not, but I'm willing to
risk it.

with Python's newer iteration and __contains__ protocols; optional
Boolean args to exclude endpoints on range searches).

Adding a simple test for the zeopasswd script. Modifies zeopasswd.main()                    slightly to accept a Database class; this is used only because                              auth_plaintext isn't an "officially supported" authentication protocol.

consistency's sake, since we use a 0 for the state flag as well)

authentication database details via commandline parameters (in lieu of
the default ZConfig). Also fixed some minor bugs -- missing import,
missing realm, etc.

For no extra charge did some usage cleanups and added a check to
Database to make sure the realm is consistent.

the checkin!

The presence or absence of a mac bit is really part of the persistent
state of SizedMessageAsyncConnection.  If state is True when
handle_read() is entered, we get into the code to check the mac and
process the message without testing whether a mac is present.

I wish there was a test for this, but the code it's very difficult to
setup a test that calls handle_read() in just the right state.  A good
project, but not one I'm up for now.

ancient pickles refering to BoboPOS. Waaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!

Got rid of the list class.

Added code to support old pickles. This error was causing
attribute errors on 'data'.

If you want a persistent list, use persistent.list

Apparent fix for the bug Sidnei reported today.  He didn't have a
simple test case, rather reported that starting Zope with Formulator
caused a crash.  I think this will fix the problem.

We had to do the same thing in Python code in ZODB4, but weren't
worrying about performance there.

- Renamed updateCodeTimestamp() to resetCaches().

- Added a test of resetCaches().

- Removed a spurious comment about chaining connections together.

Patch from Dieter Maurer.