Revert rev 30607 -- Zope trunk actually does still use ZODB's _Persistence.c,
although ZODB doesn't.

Port from ZODB 3.2.

referencesf():  Use %r instead of %s format in the error message, else
the pickle shown is full of unprintable characters.

Remove unused code; Zope 2.8 already has its own copy.

Use a zdaemon tag.

Prepare to stitch in zdaemon from a tag.

Use the same Zope3 tag as Zope 2.8.

Connection (and IDataManager) grows a public .transaction_manager
attribute.

The related txn_mgr spelling of various method arguments is
now deprecated, and a "transaction_manager" spelling is added.

Change test.py to use zope.testing.testrunner, and update the
latter to Jim's development version.  Hardest part turned out
to be teaching ZODB's setup.py how to "build" this version of
zope.testing (it has a lot of packages in a test directory,
where the latter is not itself a package).

Incidentally repaired an oversight in building zope.interface
too.

Missed a socket address change.

Use ZConfig's new socket address types appropriately.

These config file keys changed:

section    key              was             is
-------    ---------------  --------------  -------------------------
zeo        address          socket-address  socket-binding-address
zeo        monitor-address  socket-address  socket-binding-address
zeoclient  server           socket-address  socket-connection-address

Stitch in ZConfig as an SVN external.
Use the same tag Zope trunk (2.8) is using.

This is probably hosed for a bit, since it's trying to delete and
re-insert ZConfig at the same time.

Collector 1788:  runzeo fails in 2.8

A patch from Mark Hammond to repair a new Windows-specific
gimmick in ZEOServer.setup_win32_signals().

ZODB grows its own zodb3.map.

checkUndoInVersion():  Add an optional `hook` argument.

This gets called "in the middle" of the test, if specified.
ZRS 1.5 uses this to pass a callback that arranges to start a
ZRS secondary then.  ZRS had its own copy of this test, but
it's a miserable & messy test, and the copy it had failed in
5 different places when using ZODB 3.4 (it had gotten way of
synch with changes since ZODB 3.2).

Also removed all traces of the bizarre _x_dostore() method.
Not sure what that was about, but if the comments were right
we don't care about ZEO 1.0 anymore.

ZEO/tests/zeoserver.py, log():  repair broken interface
between old logging code and the use of Python's logging
package.

ZODB/tests/TransactionalUndoStorage.py, _exercise_info_indices():
Jeez Louise, the new tests I added for undoInfo+undoLog work fine
in ZODB, but break the ZRS tests(!).  They close the DB "too soon",
and in one of the ZRS scenarios that leaves a recovering secondary
without a primary to recover from.

undoInfo() and undoLog() almost always returned wrong # of results.

Repaired, + new tests.

Typos.

Docs for undoLog() and undoInfo().

Irony:  While these are the only accurate docs that
exist, FileStorage doesn't actually implement it this
way.  Fixing that is the next step.

Combine the news items from the last 8 "internal
releases" -- had become impossible to follow.

An internal 3.4a8 release, to test Zope2 with ZAapplication moved.

ZApplication.py is moving to Zope trunk.

__call__() got broken about a year ago.

There are no tests for this.

Fixed some incorrect comments.

Fix obscure bug.

If a threaded transaction manager ever passed None to
the Transaction constructor's `synchronizers` argument, then
synchronizers registered later in the same transaction
were invisible to the transaction, and so their afterCompletion()
methods wouldn't get called when the transaction ended.

`synchs` can be None -- deal with it.  Reported by Gary Poster.

ISynchronizer grows a newTransaction() method, called
whenever TransactionManager.begin() is called.

Connection implements that, and changes its ISynchronizer
afterCompletion() method, to call sync() on its storage
(if the storage has such a method), and to process
invalidations in any case.

The bottom line is that storage sync() will get done "by
magic" now after top-level commit() and abort(), and after
explicit TransactionManager.begin().  This should make it
possible to deprecate Connection.sync(), although I'm not
doing that yet.  Made a small but meaningful start by
purging many sync() calls from some of the nastiest ZEO
tests -- and they still work fine.

Partial savepoint review.

Added more comments.  Changed some comments to English.
Renamed some attributes and methods for clarity.
Switched to using a Python WeakKeyDictionary instead of
rolling our own out of a Python dict and raw weakrefs.

Port from ZODB 3.2.

Added new test checkSubtxnCommitDoesntGetInvalidations to
verify that a longstanding bug in subtransaction commit is
repaired.

Jim (Fulton) discovered this in ZODB 3.4's code, while implementing
savepoint/rollback.  Same bugs had been there at least since ZODB 3.1.

Also added news about the bug.

There is no logging level named EXCEPTION.

Original checkin msgs follow:

r30203 | jim | 2005-04-27 11:55:09 -0400 (Wed, 27 Apr 2005) | 2 lines
   M /ZODB/branches/3.4/src/transaction/_transaction.py

Fixed stupid bug.

r30200 | jim | 2005-04-27 07:20:56 -0400 (Wed, 27 Apr 2005) | 39 lines
   M /ZODB/branches/3.4/src/transaction/_transaction.py

Changed the strategy for managing savepoints.  The requirements
for savepoint management are:

- All savepoints for a transaction should be invalidated when the
  transaction commits or aborts

- If a savepoint is rolled back, then all savepoints after it within
  a transaction must be invalidated.

We previously implemented these requirements by organizing transaction
savepoints into a doubly linked list.  This was overkill.  We didn't
have need for such fine-grained ordering.  This strategy had the
disadvantage that it kept all savepoints around until the transaction
ended.  Savepoints could be expensive to keep and it's possible that
some applications could keep a lot of them.

The new stragey is to:

- Keep weak references to savepoints.  We can forget about savepoints
  that the application isn't using.  Any resources used by these
  savepoints can be freed.

(We have to keep a strong reference to the last savepoint used for
  a subtransaction.)

- We assign indexes to savepoints within a transaction.  When a
  savepoint is rolled back, in addition to invalidating that
  savepoint, we also invalidate savepoints with a higher index.

A side effect of this change is that code using the savepoint API
should interfere less with code using subtransactions.  Of course, we
really need to phase out code that uses subtransactions.

It is likely that we can leverage this change in strategy to speed
creation of ZODB connection savepoints.  Creating a ZODB connection
savepoint now requires copying the savepoint storage index.  This
index could become large.  If applications aren't holding on to old
savepoints, then it is possible that we could avoid this copy.

Trimmed trailing whitespace.

Repaired NEWS format; predicted the next release will be 3.4b1.

A ``getBeforeCommitHooks()`` method has been added to Transaction. It
returns an iterable producing the registered beforeCommit hooks.