Merge revs 27788 and 27795 from Zope trunk.

See Collector #1350.

ghostify() and unghostify() detect one form of thread insanity
now.  In a debug build, they abort the process if it happens.
In a release build, unghostify() raises SystemError, and
ghostify() ignores it (ghostify() can't raise an exception).

Use proto 1 pickles in ZEO/zrpc/Marshaller.encode().

This rehabilitates the good part of Andreas's change; adds
news; adds comments; and adds more comments about the
sequence of undocumented cPickle gimmicks this relies on.

Merge rev 27722 from Zope trunk.

Collector #1517: History tab for ZPT does not work

FileStorage.history() was reading the user, description,
and extension fields out of the object pickle, due to
starting the read at a wrong location.

Merge rev 27712 from Zope trunk.

Restore the Connection._opened attribute, actually used by
DB.connectionDebugInfo().

Transaction.begin().

When raising DeprecationWarning, point the warning machinery at
the caller, not at Transaction.begin() itself.  As is (voice
of experience <wink>), these warnings are darned near useless to
track down code that needs to be changed on Zope trunk.  With the
change, the warnings point directly at the deprecated uses.

Check in stuff mistakenly committed from the Zope trunk.

Rev 26842 changed cache-size parameters to type byte-size.
That's a good change, but it was checked in from a wrong place
(from the Zope trunk's copy of ZODB).  Since we want Zope 2.8 to
be usable with ZODB 3.3 too, I need to slam this change into
3.3 now.

Alas, I don't understand how ZODB+ZEO got stitched into the
Zope trunk (I do understand how it got stitched into the
Zope3 trunk, and into the X3 branch), and it looks like that
was left in a poor state.

New test checkReadConflictErrorClearedDuringAbort().

This confirms that 3.3 didn't have the 3.2.4c1 bug wherein
an explict transaction abort() failed to clear Connections'
memory of ReadConflictErrors.

Act as if a 3.3c1 release were being made.  Maybe it will
be.  Besides "the usual" release fiddling, repaired many
out-of-date pieces of the ZODB programming guide; I'm sure
many remain, though.

A test here didn't clean up after a failing commit().

Removed obsolete comment.

As discussed on zodb-dev, failing commit "sticks" now.

After a commit fails (raises an exception), all subsequent attempts
to commit, join, or register with the transaction now raise the new
TransactionFailedError.  The failed transaction must be explicitly
discarded now, via abort() on the transaction or begin() on its
transaction manager.

Officially deprecate Transaction.begin().

Forward port from Zope 2.7 branch.

The ConflictError.get_{old,new}_serial() methods each did what the other
was supposed to do.

Port from Zope 2.7 branch.

Collector #1488 (TemporaryStorage -- going backward in time).

This confusion was really due to that the detail on a ConflictError
exception didn't make sense.

_handle_independent():  Failed to record that a ReadConflictError
was raised for an object with a _p_independent() method that
returned false.

Raise ConnectionStateError if an attempt to close a connection
is made while the connection has a pending subtransaction.

Transaction.begin() didn't do anything.

begin() is supposed to abort the current transaction, but
Transaction.begin() did not.  Calling begin() on a transaction
*manager* worked fine, and is the intended way to do a begin()
in 3.3.  But calling begin() on a Transaction object is still
very easy to do (e.g., the older get_transaction().begin()
spelling still works), and shouldn't be a subtle disaster.

Empty BTrees/__init__.py; isn't doing anything except
creating problems.

It's possible that an oid is referenced despite never being
defined (well, it's not *supposed* to be possible ...).

automated test.  There were so many oddities getting that test
to pass, and so much consequent fiddling of other code, that I
don't intend to merge this to the 3.3 branch.  Also moved to
current Python CVS doctest.py, because the test really needs
the +ELLIPSIS option (tids vary across runs).

and extension fields from a transaction was fatally confused,
usually reading them out of the object pickle by mistake.  This
caused several tools to display binary gibberish.

Forward port from Zope 2.7 branch.

analyze_rec():  This produced spurious "len of unsized object" messages
when a data record had a backpointer instead of a pickle.  Repaired.

analyze(), analyze_trans():  Simplified overly elaborate iteration code.

Forward port from Zope 2.7 branch.

Many changes to fsrefs.py, based on problems I hit using it in real life.