What's new in ZODB3 3.4? ======================== Release date: DD-MMM-2004 DB -- - There is no longer a hard limit on the number of connections that ``DB.open()`` will create. In other words, ``DB.open()`` never blocks anymore waiting for an earlier connection to close, and ``DB.open()`` always returns a connection now (while it wasn't documented, it was possible for ``DB.open()`` to return ``None`` before). ``pool_size`` continues to default to 7, but its meaning has changed: if more than ``pool_size`` connections are obtained from ``DB.open()`` and not closed, a warning is logged; if more than twice ``pool_size``, a critical problem is logged. ``pool_size`` should be set to the maximum number of connections from the ``DB`` instance you expect to have open simultaneously. In addition, if a connection obtained from ``DB.open()`` becomes unreachable without having been explicitly closed, when Python's garbage collection reclaims that connection it no longer counts against the ``pool_size`` thresholds for logging messages. The following optional arguments to ``DB.open()`` are deprecated: ``transaction``, ``waitflag``, ``force`` and ``temporary``. If one is specified, its value is ignored, and ``DeprecationWarning`` is raised. In ZODB 3.6, these optional arguments will be removed. BTrees ------ - A new family of BTree types, in the ``IFBTree`` module, map signed integers (32 bits) to C floats (also 32 bits). The intended use is to help construct search indices, where, e.g., integer word or document identifiers map to scores of some kind. This is easier than trying to work with scaled integer scores in an ``IIBTree``, and Zope3 has moved to ``IFBTrees`` for these purposes in its search code. Tools ----- New tool fsoids.py, for heavy debugging of FileStorages; shows all uses of specified oids in the entire database (e.g., suppose oid 0x345620 is missing -- did it ever exist? if so, when? who referenced it? when was the last transaction that modified an object that referenced it? which objects did it reference? what kind of object was it?). ZODB/test/testfsoids.py is a tutorial doctest. What's new in ZODB3 3.3.1a2? ============================ Release date: DD-MMM-2005 ZEO --- Repaired subtle race conditions in establishing ZEO connections, both client- and server-side. These account for intermittent cases where ZEO failed to make a connection (or reconnection), accompanied by a log message showing an error caught in ``asyncore`` and having a traceback ending with: ``UnpicklingError: invalid load key, 'Z'.`` or: ``ZRPCError: bad handshake '(K\x00K\x00U\x0fgetAuthProtocol)t.'`` or: ``error: (9, 'Bad file descriptor')`` or an ``AttributeError``. These were exacerbated when running the test suite, because of an unintended busy loop in the test scaffolding, which could starve the thread trying to make a connection. The ZEO reconnection tests may run much faster now, depending on platform, and should suffer far fewer (if any) intermittent "timed out waiting for storage to connect" failures. FileStorage ----------- - The ``.store()`` and ``.restore()`` methods didn't update the storage's belief about the largest oid in use when passed an oid larger than the largest oid the storage already knew about. Because ``.restore()`` in particular is used by ``copyTransactionsFrom()``, and by the first stage of ZRS recovery, a large database could be created that believed the only oid in use was oid 0 (the special oid reserved for the root object). In rare cases, it could go on from there assigning duplicate oids to new objects, starting over from oid 1 again. This has been repaired. A new ``set_max_oid()`` method was added to the ``BaseStorage`` class so that derived storages can update the largest oid in use in a threadsafe way. - A FileStorage's index file tried to maintain the index's largest oid as a separate piece of data, incrementally updated over the storage's lifetime. This scheme was more complicated than necessary, so was also more brittle and slower than necessary. It indirectly participated in a rare but critical bug: when a FileStorage was created via ``copyTransactionsFrom()``, the "maximum oid" saved in the index file was always 0. Use that FileStorage, and it could then create "new" oids starting over at 0 again, despite that those oids were already in use by old objects in the database. Packing a FileStorage has no reason to try to update the maximum oid in the index file either, so this kind of damage could (and did) persist even across packing. The index file's maximum-oid data is ignored now, but is still written out so that ``.index`` files can be read by older versions of ZODB. Finding the true maximum oid is done now by exploiting that the main index is really a kind of BTree (long ago, this wasn't true), and finding the largest key in a BTree is inexpensive. - A FileStorage's index file could be updated on disk even if the storage was opened in read-only mode. That bug has been repaired. - An efficient ``maxKey()`` implementation was added to class ``fsIndex``. Pickle (in-memory Connection) Cache ----------------------------------- You probably never saw this exception: ``ValueError: Can not re-register object under a different oid`` It's been changed to say what it meant: ``ValueError: A different object already has the same oid`` This happens if an attempt is made to add distinct objects to the cache that have the same oid (object identifier). ZODB should never do this, but it's possible for application code to force such an attempt. fsIndex ------- An efficient ``maxKey()`` method was implemented for the ``fsIndex`` class. This makes it possible to determine the largest oid in a ``FileStorage`` index efficiently, directly, and reliably, replacing a more delicate scheme that tried to keep track of this by saving an oid high water mark in the index file and incrementally updating it. What's new in ZODB3 3.3.1a1? ============================ Release date: 11-Jan-2005 ZEO client cache ---------------- - Collector 1536: The ``cache-size`` configuration option for ZEO clients was being ignored. Worse, the client cache size was only one megabyte, much smaller than the advertised default of 20MB. Note that the default is carried over from a time when gigabyte disks were expensive and rare; 20MB is also too small on most modern machines. - Fixed a nasty bug in cache verification. A persistent ZEO cache uses a disk file, and, when active, has some in-memory data structures too to speed operation. Invalidations processed as part of startup cache verification were reflected in the in-memory data structures, but not correctly in the disk file. So if an object revision was invalidated as part of verification, the object wasn't loaded again before the connection was closed, and the object revision remained in the cache file until the connection was closed, then the next time the cache file was opened it could believe that the stale object revision in the file was actually current. - Fixed a bug wherein an object removed from the client cache didn't properly mark the file slice it occupied as being available for reuse. ZEO --- Collector 1503: excessive logging. It was possible for a ZEO client to log "waiting for cache verification to finish" messages at a very high rate, producing gigabytes of such messages in short order. ``ClientStorage._wait_sync()`` was changed to log no more than one such message per 5 minutes. persistent ---------- Collector #1350: ZODB has a default one-thread-per-connection model, and two threads should never do operations on a single connection simultaneously. However, ZODB can't detect violations, and this happened in an early stage of Zope 2.8 development. The low-level ``ghostify()`` and ``unghostify()`` routines in ``cPerisistence.c`` were changed to give some help in detecting this when it happens. In a debug build, both abort the process if thread interference is detected. This is extreme, but impossible to overlook. In a release build, ``unghostify()`` raises ``SystemError`` if thread damage is detected; ``ghostify()`` ignores the problem in a release build (``ghostify()`` is supposed to be so simple that it "can't fail"). ConflictError ------------- New in 3.3, a ``ConflictError`` exception may attempt to insert the path to the object's class in its message. However, a ZEO server may not have access to application class implementations, and then the attempt by the server to raise ``ConflictError`` could raise ``ImportError`` instead while trying to determine the object's class path. This was confusing. The code has been changed to obtain the class path from the object's pickle, without trying to import application modules or classes. FileStorage ----------- Collector 1581: When an attempt to pack a corrupted ``Data.fs`` file was made, it was possible for the pack routine to die with a reference to an undefined global while it was trying to raise ``CorruptedError``. It raises ``CorruptedError``, as it always intended, in these cases now. Install ------- The C header file ``ring.h`` is now installed. Tools ----- - ``BTrees.check.display()`` now displays the oids (if any) of the BTree's or TreeSet's constituent objects. What's new in ZODB3 3.3? ======================== Release date: 06-Oct-2004 ZEO --- The encoding of RPC calls between server and client was being done with protocol 0 ("text mode") pickles, which could require sending four times as many bytes as necessary. Protocol 1 pickles are used now. Thanks to Andreas Jung for the diagnosis and cure. ZODB/component.xml ------------------ ``cache-size`` parameters were changed from type ``integer`` to type ``byte-size``. This allows you to specify, for example, "``cache-size 20MB``" to get a 20 megabyte cache. transaction ----------- The deprecation warning for ``Transaction.begin()`` was changed to point to the caller, instead of to ``Transaction.begin()`` itself. Connection ---------- Restored Connection's private ``_opened`` attribute. This was still referenced by ``DB.connectionDebugInfo()``, and Zope 2 calls the latter. FileStorage ----------- 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. Looked like cut-and-paste repetition of the same bug in ``FileStorage.FileIterator`` noted in the news for 3.3c1. What's new in ZODB3 3.3 release candidate 1? ============================================ Release date: 14-Sep-2004 Connection ---------- ZODB intends to raise ``ConnnectionStateError`` if an attempt is made to close a connection while modifications are pending (the connection is involved in a transaction that hasn't been ``abort()``'ed or ``commit()``'ed). It was missing the case where the only pending modifications were made in subtransactions. This has been fixed. If an attempt to close a connection with pending subtransactions is made now:: ConnnectionStateError: Cannot close a connection with a pending subtransaction is raised. transaction ----------- - Transactions have new, backward-incompatible behavior in one respect: if a ``Transaction.commit()``, ``Transaction.commit(False)``, or ``Transaction.commit(True)`` raised an exception, prior behavior was that the transaction effectively aborted, and a new transaction began. A primary bad consequence was that, if in a sequence of subtransaction commits, one of the commits failed but the exception was suppressed, all changes up to and including the failing commit were lost, but later subtransaction commits in the sequence got no indication that something had gone wrong, nor did the final (top level) commit. This could easily lead to inconsistent data being committed, from the application's point of view. The new behavior is that a failing commit "sticks" until explicitly cleared. Now if an exception is raised by a ``commit()`` call (whether subtransaction or top level) on a Transaction object ``T``: - Pending changes are aborted, exactly as they were for a failing commit before. - But ``T`` remains the current transaction object (if ``tm`` is ``T``'s transaction manger, ``tm.get()`` continues to return ``T``). - All subsequent attempts to do ``T.commit()``, ``T.join()``, or ``T.register()`` raise the new ``TransactionFailedError`` exception. Note that if you try to modify a persistent object, that object's resource manager (usually a ``Connection`` object) will attempt to ``join()`` the failed transaction, and ``TransactionFailedError`` will be raised right away. So after a transaction or subtransaction commit fails, that must be explicitly cleared now, either by invoking ``abort()`` on the transaction object, or by invoking ``begin()`` on its transaction manager. - Some explanations of new transaction features in the 3.3a3 news were incorrect, and this news file has been retroactively edited to repair that. See news for 3.3a3 below. - If ReadConflictError was raised by an attempt to load an object with a ``_p_independent()`` method that returned false, attempting to commit the transaction failed to (re)raise ReadConflictError for that object. Note that ZODB intends to prevent committing a transaction in which a ReadConflictError occurred; this was an obscure case it missed. - Growing pains: ZODB 3.2 had a bug wherein ``Transaction.begin()`` didn't abort the current transaction if the only pending changes were in a subtransaction. In ZODB 3.3, it's intended that a transaction manager be used to effect ``begin()`` (instead of invoking ``Transaction.begin()``), and calling ``begin()`` on a transaction manager didn't have this old bug. However, ``Transaction.begin()`` still exists in 3.3, and it had a worse bug: it never aborted the transaction (not even if changes were pending outside of subtransactions). ``Transaction.begin()`` has been changed to abort the transaction. ``Transaction.begin()`` is also deprecated. Don't use it. Use ``begin()`` on the relevant transaction manager instead. For example, >>> import transaction >>> txn = transaction.begin() # start a txn using the default TM if using the default ``ThreadTransactionManager`` (see news for 3.3a3 below). In 3.3, it's intended that a single ``Transaction`` object is used for exactly one transaction. So, unlike as in 3.2, when somtimes ``Transaction`` objects were reused across transactions, but sometimes weren't, when you do ``Transaction.begin()`` in 3.3 a brand new transaction object is created. That's why this use is deprecated. Code of the form: >>> txn = transaction.get() >>> ... >>> txn.begin() >>> ... >>> txn.commit() can't work as intended in 3.3, because ``txn`` is no longer the current ``Transaction`` object the instant ``txn.begin()`` returns. BTrees ------ The BTrees __init__.py file is now just a comment. It had been trying to set up support for (long gone) "int sets", and to import an old version of Zope's Interface package, which doesn't even ship with ZODB. The latter in particular created problems, at least clashing with PythonCAD's Interface package. POSException ------------ Collector #1488 (TemporaryStorage -- going backward in time). This confusion was really due to that the detail on a ConflictError exception didn't make sense. It called the current revision "was", and the old revision "now". The detail is much more informative now. For example, if the exception said:: ConflictError: database conflict error (oid 0xcb22, serial was 0x03441422948b4399, now 0x034414228c3728d5) before, it now says:: ConflictError: database conflict error (oid 0xcb22, serial this txn started with 0x034414228c3728d5 2002-04-14 20:50:32.863000, serial currently committed 0x03441422948b4399 2002-04-14 20:50:34.815000) ConflictError ------------- The undocumented ``get_old_serial()`` and ``get_new_serial()`` methods were swapped (the first returned the new serial, and the second returned the old serial). Tools ----- ``FileStorage.FileIterator`` was confused about how to read a transaction's user and description fields, which caused several tools to display binary gibberish for these values. ``ZODB.utils.oid_repr()`` changed to add a leading "0x", and to strip leading zeroes. This is used, e.g., in the detail of a ``POSKeyError`` exception, to identify the missing oid. Before, the output was ambiguous. For example, oid 17 was displayed as 0000000000000011. As a Python integer, that's octal 9. Or was it meant to be decimal 11? Or was it meant to be hex? Now it displays as 0x11. fsrefs.py: When run with ``-v``, produced tracebacks for objects whose creation was merely undone. This was confusing. Tracebacks are now produced only if there's "a real" problem loading an oid. If the current revision of object O refers to an object P whose creation has been undone, this is now identified as a distinct case. Captured and ignored most attempts to stop it via Ctrl+C. Repaired. Now makes two passes, so that an accurate report can be given of all invalid references. ``analyze.py`` produced spurious "len of unsized object" messages when finding a data record for an object uncreation or version abort. These no longer appear. ``fsdump.py``'s ``get_pickle_metadata()`` function (which is used by several tools) was confused about what to do when the ZODB pickle started with a pickle ``GLOBAL`` opcode. It actually loaded the class then, which it intends never to do, leading to stray messages on stdout when the class wasn't available, and leading to a strange return value even when it was available (the repr of the type object was returned as "the module name", and an empty string was returned as "the class name"). This has been repaired. What's new in ZODB3 3.3 beta 2 ============================== Release date: 13-Aug-2004 Transaction Managers -------------------- Zope3-dev Collector #139: Memory leak involving buckets and connections The transaction manager internals effectively made every Connection object immortal, except for those explicitly closed. Since typical practice is not to close connections explicitly (and closing a DB happens not to close the connections to it -- although that may change), this caused massive memory leaks when many connections were opened. The transaction manager internals were reworked to use weak references instead, so that connection memory (and other registered synch objects) now get cleaned up when nothing other than the transaction manager knows about them. Storages -------- Collector #1327: FileStorage init confused by time travel If the system clock "went backwards" a long time between the times a FileStorage was closed and reopened, new transaction ids could be smaller than transaction ids already in the storage, violating a key invariant. Now transaction ids are guaranteed to be increasing even when this happens. If time appears to have run backwards at all when a FileStorage is opened, a new message saying so is logged at warning level; if time appears to have run backwards at least 30 minutes, the message is logged at critical level (and you should investigate to find and repair the true cause). Tools ----- repozo.py: Thanks to a suggestion from Toby Dickenson, backups (whether incremental or full) are first written to a temp file now, which is fsync'ed at the end, and only after that succeeds is the file renamed to YYYY-MM-DD-HH-MM-SS.ext form. In case of a system crash during a repozo backup, this at least makes it much less likely that a backup file with incomplete or incorrect data will be left behind. fsrefs.py: Fleshed out the module docstring, and repaired a bug wherein spurious error msgs could be produced after reporting a problem with an unloadable object. Test suite ---------- Collector #1397: testTimeStamp fails on FreeBSD The BSD distributions are unique in that their mktime() implementation usually ignores the input tm_isdst value. Test checkFullTimeStamp() was sensitive to this platform quirk. Reworked the way some of the ZEO tests use threads, so that unittest is more likely to notice the real cause of a failure (which usually occurs in a thread), and less likely to latch on to spurious problems resulting from the real failure. What's new in ZODB3 3.3 beta 1 ============================== Release date: 07-Jun-2004 3.3b1 is the first ZODB release built using the new zpkg tools: http://zope.org/Members/fdrake/zpkgtools/ This appears to have worked very well. The structure of the tarball release differs from previous releases because of it, and the set of installed files includes some that were not installed in previous releases. That shouldn't create problems, so let us know if it does! We'll fine-tune this for the next release. BTrees ------ Fixed bug indexing BTreeItems objects with negative indexes. This caused reverse iteration to return each item twice. Thanks to Casey Duncan for the fix. ZODB ---- Methods removed from the database (ZODB.DB.DB) class: cacheStatistics(), cacheMeanAge(), cacheMeanDeac(), and cacheMeanDeal(). These were undocumented, untested, and unused. The first always returned an empty tuple, and the rest always returned None. When trying to do recovery to a time earlier than that of the most recent full backup, repozo.py failed to find the appropriate files, erroneously claiming "No files in repository before <specified time>". This has been repaired. Collector #1330: repozo.py -R can create corrupt .fs. When looking for the backup files needed to recreate a Data.fs file, repozo could (unintentionally) include its meta .dat files in the list, or random files of any kind created by the user in the backup directory. These would then get copied verbatim into the reconstructed file, filling parts with junk. Repaired by filtering the file list to include only files with the data extensions repozo.py creates (.fs, .fsz, .deltafs, and .deltafsz). Thanks to James Henderson for the diagnosis. fsrecover.py couldn't work, because it referenced attributes that no longer existed after the MVCC changes. Repaired that, and added new tests to ensure it continues working. Collector #1309: The reference counts reported by DB.cacheExtremeDetails() for ghosts were one too small. Thanks to Dieter Maurer for the diagnosis. Collector #1208: Infinite loop in cPickleCache. If a persistent object had a __del__ method (probably not a good idea regardless, but we don't prevent it) that referenced an attribute of self, the code to deactivate objects in the cache could get into an infinite loop: ghostifying the object could lead to calling its __del__ method, the latter would load the object into cache again to satsify the attribute reference, the cache would again decide that the object should be ghostified, and so on. The infinite loop no longer occurs, but note that objects of this kind still aren't sensible (they're effectively immortal). Thanks to Toby Dickenson for suggesting a nice cure. What's new in ZODB3 3.3 alpha 3 =============================== Release date: 16-Apr-2004 transaction ----------- There is a new transaction package, which provides new interfaces for application code and for the interaction between transactions and resource managers. The top-level transaction package has functions ``commit()``, ``abort()``, ``get()``, and ``begin()``. They should be used instead of the magic ``get_transaction()`` builtin, which will be deprecated. For example: >>> get_transaction().commit() should now be written as >>> import transaction >>> transaction.commit() The new API provides explicit transaction manager objects. A transaction manager (TM) is responsible for associating resource managers with a "current" transaction. The default TM, implemented by class ``ThreadedTransactionManager``, assigns each thread its own current transaction. This default TM is available as ``transaction.manager``. The ``TransactionManager`` class assigns all threads to the same transaction, and is an explicit replacement for the ``Connection.setLocalTransaction()`` method: A transaction manager instance can be passed as the txn_mgr argument to ``DB.open()``. If you do, the connection will use the specified transaction manager instead of the default TM. The current transaction is obtained by calling ``get()`` on a TM. For example: >>> tm = transaction.TransactionManager() >>> cn = db.open(txn_mgr=tm) [...] >>> tm.get().commit() The ``setLocalTransaction()`` and ``getTransaction()`` methods of Connection are deprecated. Use an explicit TM passed via ``txn_mgr=`` to ``DB.open()`` instead. The ``setLocalTransaction()`` method still works, but it returns a TM instead of a Transaction. A TM creates Transaction objects, which are used for exactly one transaction. Transaction objects still have ``commit()``, ``abort()``, ``note()``, ``setUser()``, and ``setExtendedInfo()`` methods. Resource managers, e.g. Connection or RDB adapter, should use a Transaction's ``join()`` method instead of its ``register()`` method. An object that calls ``join()`` manages its own resources. An object that calls ``register()`` expects the TM to manage the objects. Data managers written against the ZODB 4 transaction API are now supported in ZODB 3. persistent ---------- A database can now contain persistent weak references. An object that is only reachable from persistent weak references will be removed by pack(). The persistence API now distinguishes between deactivation and invalidation. This change is intended to support objects that can't be ghosts, like persistent classes. Deactivation occurs when a user calls _p_deactivate() or when the cache evicts objects because it is full. Invalidation occurs when a transaction updates the object. An object that can't be a ghost must load new state when it is invalidated, but can ignore deactivation. Persistent objects can implement a __getnewargs__() method that will be used to provide arguments that should be passed to __new__() when instances (including ghosts) are created. An object that implements __getnewargs__() must be loaded from storage even to create a ghost. There is new support for writing hooks like __getattr__ and __getattribute__. The new hooks require that user code call special persistence methods like _p_getattr() inside their hook. See the ZODB programming guide for details. The format of serialized persistent references has changed; that is, the on-disk format for references has changed. The old format is still supported, but earlier versions of ZODB will not be able to read the new format. ZODB ---- Closing a ZODB Connection while it is registered with a transaction, e.g. has pending modifications, will raise a ConnnectionStateError. Trying to load objects from or store objects to a closed connection will also raise a ConnnectionStateError. ZODB connections are synchronized on commit, even when they didn't modify objects. This feature assumes that the thread that opened the connection is also the thread that uses it. If not, this feature will cause problems. It can be disabled by passing synch=False to open(). New broken object support. New add() method on Connection. User code should not assign the _p_jar attribute of a new persistent object directly; a deprecation warning is issued in this case. Added a get() method to Connection as a preferred synonym for __getitem__(). Several methods and/or specific optional arguments of methods have been deprecated. The cache_deactivate_after argument used by DB() and Connection() is deprecated. The DB methods getCacheDeactivateAfter(), getVersionCacheDeactivateAfter(), setCacheDeactivateAfter(), and setVersionCacheDeactivateAfter() are also deprecated. The old-style undo() method was removed from the storage API, and transactionalUndo() was renamed to undo(). The BDBStorages are no longer distributed with ZODB. Fixed a serious bug in the new pack implementation. If pack was called on the storage and passed a time earlier than a previous pack time, data could be lost. In other words, if there are any two pack calls, where the time argument passed to the second call was earlier than the first call, data loss could occur. The bug was fixed by causing the second call to raise a StorageError before performing any work. Fixed a rare bug in pack: if a pack started during a small window of time near the end of a concurrent transaction's commit, it was possible for the pack attempt to raise a spurious CorruptedError: ... transaction with checkpoint flag set exception. This did no damage to the database, or to the transaction in progress, but no pack was performed then. By popular demand, FileStorage.pack() no longer propagates a FileStorageError: The database has already been packed to a later time or no changes have been made since the last pack exception. Instead that message is logged (at INFO level), and the pack attempt simply returns then (no pack is performed). ZEO --- Fixed a bug that prevented the -m / --monitor argument from working. zdaemon ------- Added a -m / --mask option that controls the umask of the subprocess. zLOG ---- The zLOG backend has been removed. zLOG is now just a facade over the standard Python logging package. Environment variables like STUPID_LOG_FILE are no longer honored. To configure logging, you need to follow the directions in the logging package documentation. The process is currently more complicated than configured zLOG. See test.py for an example. ZConfig ------- This release of ZODB contains ZConfig 2.1. More documentation has been written. Make sure keys specified as attributes of the <default> element are converted by the appropriate key type, and are re-checked for derived sections. Refactored the ZConfig.components.logger schema components so that a schema can import just one of the "eventlog" or "logger" sections if desired. This can be helpful to avoid naming conflicts. Added a reopen() method to the logger factories. Always use an absolute pathname when opening a FileHandler. Miscellaneous ------------- The layout of the ZODB source release has changed. All the source code is contained in a src subdirectory. The primary motivation for this change was to avoid confusion caused by installing ZODB and then testing it interactively from the source directory; the interpreter would find the uncompiled ZODB package in the source directory and report an import error. A reference-counting bug was fixed, in the logic calling a modified persistent object's data manager's register() method. The primary symptom was rare assertion failures in Python's cyclic garbage collection. The Connection class's onCommitAction() method was removed. Some of the doc strings in ZODB are now written for processing by epydoc. Several new test suites were written using doctest instead of the standard unittest TestCase framework. MappingStorage now implements getTid(). ThreadedAsync: Provide a way to shutdown the servers using an exit status. The mkzeoinstance script looks for a ZODB installation, not a Zope installation. The received wisdom is that running a ZEO server without access to the appserver code avoids many mysterious problems. What's new in ZODB3 3.3 alpha 2 =============================== Release date: 06-Jan-2004 This release contains a major overhaul of the persistence machinery, including some user-visible changes. The Persistent base class is now a new-style class instead of an ExtensionClass. The change enables the use of features like properties with persistent object classes. The Persistent base class is now contained in the persistent package. The Persistence package is included for backwards compatibility. The Persistence package is used by Zope to provide special ExtensionClass-compatibility features like a non-C3 MRO and an __of__ method. ExtensionClass is not included with this release of ZODB3. If you use the Persistence package, it will print a warning and import Persistent from persistent. In short, the new persistent package is recommended for non-Zope applications. The following dotted class names are now preferred over earlier names: - persistent.Persistent - persistent.list.PersistentList - persistent.mapping.PersistentMapping - persistent.TimeStamp The in-memory, per-connection object cache (pickle cache) was changed to participate in garbage collection. This should reduce the number of memory leaks, although we are still tracking a few problems. Multi-version concurrency control --------------------------------- ZODB now supports multi-version concurrency control (MVCC) for storages that support multiple revisions. FileStorage and BDBFullStorage both support MVCC. In short, MVCC means that read conflicts should almost never occur. When an object is modified in one transaction, other concurrent transactions read old revisions of the object to preserve consistency. In earlier versions of ZODB, any access of the modified object would raise a ReadConflictError. The ZODB internals changed significantly to accommodate MVCC. There are relatively few user visible changes, aside from the lack of read conflicts. It is possible to disable the MVCC feature using the mvcc keyword argument to the DB open() method, ex.: db.open(mvcc=False). ZEO --- Changed the ZEO server and control process to work with a single configuration file; this is now the default way to configure these processes. (It's still possible to use separate configuration files.) The ZEO configuration file can now include a "runner" section used by the control process and ignored by the ZEO server process itself. If present, the control process can use the same configuration file. Fixed a performance problem in the logging code for the ZEO protocol. The logging code could call repr() on arbitrarily long lists, even though it only logged the first 60 bytes; worse, it called repr() even if logging was currently disabled. Fixed to call repr() on individual elements until the limit is reached. Fixed a bug in zrpc (when using authentication) where the MAC header wasn't being read for large messages, generating errors while unpickling commands sent over the wire. Also fixed the zeopasswd.py script, added testcases and provided a more complete commandline interface. Fixed a misuse of the _map variable in zrpc Connectio objects, which are also asyncore.dispatcher objects. This allows ZEO to work with CVS Python (2.4). _map is used to indicate whether the dispatcher users the default socket_map or a custom socket_map. A recent change to asyncore caused it to use _map in its add_channel() and del_channel() methods, which presumes to be a bug fix (may get ported to 2.3). That causes our dubious use of _map to be a problem, because we also put the Connections in the global socket_map. The new asyncore won't remove it from the global socket map, because it has a custom _map. The prefix used for log messages from runzeo.py was changed from RUNSVR to RUNZEO. Miscellaneous ------------- ReadConflictError objects now have an ignore() method. Normally, a transaction that causes a read conflict can't be committed. If the exception is caught and its ignore() method called, the transaction can be committed. Application code may need this in advanced applications. What's new in ZODB3 3.3 alpha 1 =============================== Release date: 17-Jul-2003 New features of Persistence --------------------------- The Persistent base class is a regular Python type implemented in C. It should be possible to create new-style classes that inherit from Persistent, and, thus, use all the new Python features introduced in Python 2.2 and 2.3. The __changed__() method on Persistent objects is no longer supported. New features in BTrees ---------------------- BTree, Bucket, TreeSet and Set objects are now iterable objects, playing nicely with the iteration protocol introduced in Python 2.2, and can be used in any context that accepts an iterable object. As for Python dicts, the iterator constructed for BTrees and Buckets iterates over the keys. >>> from BTrees.OOBTree import OOBTree >>> b = OOBTree({"one": 1, "two": 2, "three": 3, "four": 4}) >>> for key in b: # iterates over the keys ... print key four one three two >>> list(enumerate(b)) [(0, 'four'), (1, 'one'), (2, 'three'), (3, 'two')] >>> i = iter(b) >>> i.next() 'four' >>> i.next() 'one' >>> i.next() 'three' >>> i.next() 'two' >>> As for Python dicts in 2.2, BTree and Bucket objects have new .iterkeys(), .iteritems(), and .itervalues() methods. TreeSet and Set objects have a new .iterkeys() method. Unlike as for Python dicts, these new methods accept optional min and max arguments to effect range searches. While Bucket.keys() produces a list, Bucket.iterkeys() produces an iterator, and similarly for Bucket values() versus itervalues(), Bucket items() versus iteritems(), and Set keys() versus iterkeys(). The iter{keys,values,items} methods of BTrees and the iterkeys() method of Treesets also produce iterators, while their keys() (etc) methods continue to produce BTreeItems objects (a form of "lazy" iterator that predates Python 2.2's iteration protocol). >>> sum(b.itervalues()) 10 >>> zip(b.itervalues(), b.iterkeys()) [(4, 'four'), (1, 'one'), (3, 'three'), (2, 'two')] >>> BTree, Bucket, TreeSet and Set objects also implement the __contains__ method new in Python 2.2, which means that testing for key membership can be done directly now via the "in" and "not in" operators: >>> "won" in b False >>> "won" not in b True >>> "one" in b True >>> All old and new range-search methods now accept keyword arguments, and new optional excludemin and excludemax keyword arguments. The new keyword arguments allow doing a range search that's exclusive at one or both ends (doesn't include min, and/or doesn't include max). >>> list(b.keys()) ['four', 'one', 'three', 'two'] >>> list(b.keys(max='three')) ['four', 'one', 'three'] >>> list(b.keys(max='three', excludemax=True)) ['four', 'one'] >>> Other improvements ------------------ The exceptions generated by write conflicts now contain the name of the conflicted object's class. This feature requires support for the storage. All the standard storages support it. What's new in ZODB3 3.2 ======================== Release date: 08-Oct-2003 Nothing has changed since release candidate 1. What's new in ZODB3 3.2 release candidate 1 =========================================== Release date: 01-Oct-2003 Added a summary to the Doc directory. There are several new documents in the 3.2 release, including "Using zdctl and zdrun to manage server processes" and "Running a ZEO Server HOWTO." Fixed ZEO's protocol negotiation mechanism so that a client ZODB 3.1 can talk to a ZODB 3.2 server. Fixed a memory leak in the ZEO server. The server was leaking a few KB of memory per connection. Fixed a memory leak in the ZODB object cache (cPickleCache). The cache did not release two references to its Connection, causing a large cycle of objects to leak when a database was closed. Fixed a bug in the ZEO code that caused it to leak socket objects on Windows. Specifically, fix the trigger mechanism so that both sockets created for a trigger are closed. Fixed a bug in the ZEO storage server that caused it to leave temp files behind. The CommitLog class contains a temp file, but it was not closing the file. Changed the order of setuid() and setgid() calls in zdrun, so that setgid() is called first. Added a timeout to the ZEO test suite that prevents hangs. The test suite creates ZEO servers with randomly assigned ports. If the port happens to be in use, the test suite would hang because the ZEO client would never stop trying to connect. The fix will cause the test to fail after a minute, but should prevent the test runner from hanging. The logging package was updated to include the latest version of the logging package from Python CVS. Note that this package is only installed for Python 2.2. In later versions of Python, it is available in the Python standard library. The ZEO1 directory was removed from the source distribution. ZEO1 is not supported, and we never intended to include it in the release. What's new in ZODB3 3.2 beta 3 ============================== Release date: 23-Sep-2003 Note: The changes listed for this release include changes also made in ZODB 3.1.x releases and ported to the 3.2 release. This version of ZODB 3.2 is not compatible with Python 2.1. Early versions were explicitly designed to be compatible with Zope 2.6. That plan has been dropped, because Zope 2.7 is already in beta release. Several of the classes in ZEO and ZODB now inherit from object, making them new-style classes. The primary motivation for the change was to make it easier to debug memory leaks. We don't expect any behavior to change as a result. A new feature to allow removal of connection pools for versions was ported from Zope 2.6. This feature is needed by Zope to avoid denial of service attacks that allow a client to create an arbitrary number of version pools. Fixed several critical ZEO bugs. - If several client transactions were blocked waiting for the storage and one of the blocked clients disconnected, the server would attempt to restart one of the other waiting clients. Since the disconnected client did not have the storage lock, this could lead to deadlock. It could also cause the assertion "self._client is None" to fail. - If a storage server fails or times out between the vote and the finish, the ZEO cache could get populated with objects that didn't make it to the storage server. - If a client loses its connection to the server near the end of a transaction, it is now guaranteed to get a ClientDisconnected error even if it reconnects before the transaction finishes. This is necessary because the server will always abort the transaction. In some cases, the client would never see an error for the aborted transaction. - In tpc_finish(), reordered the calls so that the server's tpc_finish() is called (and must succeed) before we update the ZEO client cache. - The storage name is now prepended to the sort key, to ensure a unique global sort order if storages are named uniquely. This can prevent deadlock in some unusual cases. Fixed several serious flaws in the implementation of the ZEO authentication protocol. - The smac layer would accept a message without a MAC even after the session key was established. - The client never initialized its session key, so it never checked incoming messages or created MACs for outgoing messags. - The smac layer used a single HMAC instance for sending and receiving messages. This approach could only work if client and server were guaranteed to process all messages in the same total order, which could only happen in simple scenarios like unit tests. Fixed a bug in ExtensionClass when comparing ExtensionClass instances. The code could raise RuntimeWarning under Python 2.3, and produce incorrect results on 64-bit platforms. Fixed bug in BDBStorage that could lead to DBRunRecoveryErrors when a transaction was aborted after performing operations like commit version or undo that create new references to existing pickles. Fixed a bug in Connection.py that caused it to fail with an AttributeError if close() was called after the database was closed. The test suite leaves fewer log files behind, although it still leaves a lot of junk. The test.py script puts each tests temp files in a separate directory, so it is easier to see which tests are causing problems. Unfortunately, it is still to tedious to figure out why the identified tests are leaving files behind. This release contains the latest and greatest version of the BDBStorage. This storage has still not seen testing in a production environment, but it represents the current best design and most recent code culled from various branches where development has occurred. The Tools directory contains a number of small improvements, a few new tools, and README.txt that catalogs the tools. Many of the tools are installed by setup.py; those scripts will now have a #! line set automatically on Unix. Fixed bugs in Tools/repozo.py, including a timing-dependent one that could cause the following invocation of repozo to do a full backup when an incremental backup would have sufficed. A pair of new scripts from Jim Fulton can be used to synthesize workloads and measure ZEO performance: see zodbload.py and zeoserverlog.py in the Tools directory. Note that these require Zope. Tools/checkbtrees.py was strengthened in two ways: - In addition to running the _check() method on each BTree B found, BTrees.check.check(B) is also run. The check() function was written after checkbtrees.py, and identifies kinds of damage B._check() cannot find. - Cycles in the object graph no longer lead to unbounded output. Note that preventing this requires remembering the oid of each persistent object found, which increases the memory needed by the script. What's new in ZODB3 3.2 beta 2 ============================== Release date: 16-Jun-2003 Fixed critical race conditions in ZEO's cache consistency code that could cause invalidations to be lost or stale data to be written to the cache. These bugs can lead to data loss or data corruption. These bugs are relatively unlikely to be provoked in sites with few conflicts, but the possibility of failure existed any time an object was loaded and stored concurrently. Fixed a bug in conflict resolution that failed to ghostify an object if it was involved in a conflict. (This code may be redundant, but it has been fixed regardless.) The ZEO server was fixed so that it does not perform any I/O until all of a transactions' invalidations are queued. If it performs I/O in the middle of sending invalidations, it would be possible to overlap a load from a client with the invalidation being sent to it. The ZEO cache now handles invalidations atomically. This is the same sort of bug that is described in the 3.1.2b1 section below, but it affects the ZEO cache. Fixed several serious bugs in fsrecover that caused it to fail catastrophically in certain cases because it thought it had found a checkpoint (status "c") record when it was in the middle of the file. Two new features snuck into this beta release. The ZODB.transact module provides a helper function that converts a regular function or method into a transactional one. The ZEO client cache now supports Adaptable Persistence (APE). The cache used to expect that all OIDs were eight bytes long. What's new in ZODB3 3.2 beta 1 ============================== Release date: 30-May-2003 ZODB ---- Invalidations are now processed atomically. Each transaction will see all the changes caused by an earlier transaction or none of them. Before this patch, it was possible for a transaction to see invalid data because it saw only a subset of the invalidations. This is the most likely cause of reported BTrees corruption, where keys were stored in the wrong bucket. When a BTree bucket splits, the bucket and the bucket's parent are both modified. If a transaction sees the invalidation for the bucket but not the parent, the BTree in memory will be internally inconsistent and keys can be put in the wrong bucket. The atomic invalidation fix prevents this problem. A number of minor reference count fixes in the object cache were fixed. That's the cPickleCache.c file. It was possible for a transaction that failed in tpc_finish() to lose the traceback that caused the failure. The transaction code was fixed to report the original error as well as any errors that occur while trying to recover from the original error. The "other" argument to copyTransactionsFrom() only needs to have an .iterator() method. For convenience, change FileStorage's and BDBFullStorage's iterator to have this method, which just returns self. Mount points are now visible from mounted objects. Fixed memory leak involving database connections and caches. When a connection or database was closed, the cache and database leaked, because of a circular reference involving the cache. Fixed the cache to explicitly clear out its contents when its connection is closed. The ZODB cache has fewer methods. It used to expose methods that could mutate the dictionary, which allowed users to violate internal invariants. ZConfig ------- It is now possible to configure ZODB databases and storages and ZEO servers using ZConfig. ZEO & zdaemon ------------- ZEO now supports authenticated client connections. The default authentication protocol uses a hash-based challenge-response protocol to prove identity and establish a session key for message authentication. The architecture is pluggable to allow third-parties to developer better authentication protocols. There is a new HOWTO for running a ZEO server. The draft in this release is incomplete, but provides more guidance than previous releases. See the file Doc/ZEO/howto.txt. The ZEO storage server's transaction timeout feature was refactored and made slightly more rebust. A new ZEO utility script, ZEO/mkzeoinst.py, was added. This creates a standard directory structure and writes a configuration file with mostly default values, and a bootstrap script that can be used to manage and monitor the server using zdctl.py (see below). Much work was done to improve zdaemon's zdctl.py and zdrun.py scripts. (In the alpha 1 release, zdrun.py was called zdaemon.py, but installing it in <prefix>/bin caused much breakage due to the name conflict with the zdaemon package.) Together with the new mkzeoinst.py script, this makes controlling a ZEO server a breeze. A ZEO client will not read from its cache during cache verification. This fix was necessary to prevent the client from reading inconsistent data. The isReadOnly() method of a ZEO client was fixed to return the false when the client is connected to a read-only fallback server. The sync() method of ClientStorage and the pending() method of a zrpc connection now do both input and output. The short_repr() function used to generate log messages was fixed so that it does not blow up creating a repr of very long tuples. Storages -------- FileStorage has a new pack() implementation that fixes several reported problems that could lead to data loss. Two small bugs were fixed in DemoStorage. undoLog() did not handle its arguments correctly and pack() could accidentally delete objects created in versions. Fixed trivial bug in fsrecover that prevented it from working at all. FileStorage will use fsync() on Windows starting with Python 2.2.3. FileStorage's commit version was fixed. It used to stop after the first object, leaving all the other objects in the version. BTrees ------ Trying to store an object of a non-integer type into an IIBTree or OIBTree could leave the bucket in a variety of insane states. For example, trying b[obj] = "I'm a string, not an integer" where b is an OIBTree. This manifested as a refcount leak in the test suite, but could have been much worse (most likely in real life is that a seemingly arbitrary existing key would "go missing"). When deleting the first child of a BTree node with more than one child, a reference to the second child leaked. This could cause the entire bucket chain to leak (not be collected as garbage despite not being referenced anymore). Other minor BTree leak scenarios were also fixed. Tools ----- New tool zeoqueue.py for parsing ZEO log files, looking for blocked transactions. New tool repozo.py (originally by Anthony Baxter) for performing incremental backups of Data.fs files. The fsrecover.py script now does a better job of recovering from errors the occur in the middle of a transaction record. Fixed several bugs that caused partial or total failures in earlier versions. What's new in ZODB3 3.2 alpha 1 =============================== Release date: 17-Jan-2003 Most of the changes in this release are performance and stability improvements to ZEO. A major packaging change is that there won't be a separate ZEO release. The new ZConfig is a noteworthy addtion (see below). ZODB ---- An experimental new transaction API was added. The Connection class has a new method, setLocalTransaction(). ZODB applications can call this method to bind transactions to connections rather than threads. This is especially useful for GUI applications, which often have only one thread but multiple independent activities within that thread (generally one per window). Thanks to Christian Reis for championing this feature. Applications that take advantage of this feature should not use the get_transaction() function. Until now, ZODB itself sometimes assumed get_transaction() was the only way to get the transaction. Minor corrections have been added. The ZODB test suite, on the other hand, can continue to use get_transaction(), since it is free to assume that transactions are bound to threads. ZEO --- There is a new recommended script for starting a storage server. We recommend using ZEO/runzeo.py instead of ZEO/start.py. The start.py script is still available in this release, but it will no longer be maintained and will eventually be removed. There is a new zdaemon implementation. This version is a separate script that runs an arbitrary daemon. To run the ZEO server as a daemon, you would run "zdrun.py runzeo.py". There is also a simple shell, zdctl.py, that can be used to manage a daemon. Try "zdctl.py -p runzeo.py". There is a new version of the ZEO protocol in this release and a first stab at protocol negotiation. (It's a first stab because the protocol checking supporting in ZODB 3.1 was too primitive to support anything better.) A ZODB 3.2 ZEO client can talk to an old server, but a ZODB 3.2 server can't talk to an old client. It's safe to upgrade all the clients first and upgrade the server last. The ZEO client cache format changed, so you'll need to delete persistent caches before restarting clients. The ZEO cache verification protocol was revised to require many fewer messages in cases where a client or server restarts quickly. The performance of full cache verification has improved dramatically. Measurements from Jim were somewhere in 2x-5x. The implementation was fixed to use the very-fast getSerial() method on the storage instead of the comparatively slow load(). The ZEO server has an optional timeout feature that will abort a connection that does not commit within a certain amount of time. The timeout works by closing the socket the client is using, causing both client and server to abort the transaction and continue. This is a drastic step, but can be useful to prevent a hung client or other bug from blocking a server indefinitely. A bug was fixed in the ZEO protocol that allowed clients to read stale cache data while cache verification was being performed. The fixed version prevents the client from using the storage until after verification completes. The ZEO server has an experimental monitoring interface that reports usage statistics for the storage server including number of connected clients and number of transactions active and committed. It can be enabled by passing the -m flag to runsvr.py. The ZEO ClientStorage no longer supports the environment variables CLIENT_HOME, INSTANCE_HOME, or ZEO_CLIENT. The ZEO1 package is still included with this release, but there is no longer an option to install it. BTrees ------ The BTrees package now has a check module that inspects a BTree to check internal invariants. Bugs in older versions of the code code leave a BTree in an inconsistent state. Calling BTrees.check.check() on a BTree object should verify its consistency. (See the NEWS section for 3.1 beta 1 below to for the old BTrees bugs.) Fixed a rare conflict resolution problem in the BTrees that could cause an segfault when the conflict resolution resulted in any empty bucket. Installation ------------ The distutils setup now installs several Python scripts. The runzeo.py and zdrun.py scripts mentioned above and several fsXXX.py scripts from the Tools directory. The test.py script does not run all the ZEO tests by default, because the ZEO tests take a long time to run. Use --all to run all the tests. Otherwise a subset of the tests, mostly using MappingStorage, are run. Storages -------- There are two new storages based on Sleepycat's BerkeleyDB in the BDBStorage package. Barry will have to write more here, because I don't know how different they are from the old bsddb3Storage storages. See Doc/BDBStorage.txt for more information. It now takes less time to open an existing FileStorage. The FileStorage uses a BTree-based index that is faster to pickle and unpickle. It also saves the index periodically so that subsequent opens will go fast even if the storage was not closed cleanly. Misc ---- The new ZConfig package, which will be used by Zope and ZODB, is included. ZConfig provides a configuration syntax, similar to Apache's syntax. The package can be used to configure the ZEO server and ZODB databases. See the module ZODB.config for functions to open the database from configuration. See ZConfig/doc for more info. The zLOG package now uses the logging package by Vinay Sajip, which will be included in Python 2.3. The Sync extension was removed from ExtensionClass, because it was not used by ZODB. What's new in ZODB3 3.1.4? ========================== Release date: 11-Sep-2003 A new feature to allow removal of connection pools for versions was ported from Zope 2.6. This feature is needed by Zope to avoid denial of service attacks that allow a client to create an arbitrary number of version pools. A pair of new scripts from Jim Fulton can be used to synthesize workloads and measure ZEO performance: see zodbload.py and zeoserverlog.py in the Tools directory. Note that these require Zope. Tools/checkbtrees.py was strengthened in two ways: - In addition to running the _check() method on each BTree B found, BTrees.check.check(B) is also run. The check() function was written after checkbtrees.py, and identifies kinds of damage B._check() cannot find. - Cycles in the object graph no longer lead to unbounded output. Note that preventing this requires remembering the oid of each persistent object found, which increases the memory needed by the script. What's new in ZODB3 3.1.3? ========================== Release date: 18-Aug-2003 Fixed several critical ZEO bugs. - If a storage server fails or times out between the vote and the finish, the ZEO cache could get populated with objects that didn't make it to the storage server. - If a client loses its connection to the server near the end of a transaction, it is now guaranteed to get a ClientDisconnected error even if it reconnects before the transaction finishes. This is necessary because the server will always abort the transaction. In some cases, the client would never see an error for the aborted transaction. - In tpc_finish(), reordered the calls so that the server's tpc_finish() is called (and must succeed) before we update the ZEO client cache. - The storage name is now prepended to the sort key, to ensure a unique global sort order if storages are named uniquely. This can prevent deadlock in some unusual cases. A variety of fixes and improvements to Berkeley storage (aka BDBStorage) were back-ported from ZODB 4. This release now contains the most current version of the Berkeley storage code. Many tests have been back-ported, but not all. Modified the Windows tests to wait longer at the end of ZEO tests for the server to shut down. Before Python 2.3, there is no waitpid() on Windows, and, thus, no way to know if the server has shut down. The change makes the Windows ZEO tests much less likely to fail or hang, at the cost of increasing the time needed to run the tests. Fixed a bug in ExtensionClass when comparing ExtensionClass instances. The code could raise RuntimeWarning under Python 2.3, and produce incorrect results on 64-bit platforms. Fixed bugs in Tools/repozo.py, including a timing-dependent one that could cause the following invocation of repozo to do a full backup when an incremental backup would have sufficed. Added Tools/README.txt that explains what each of the scripts in the Tools directory does. There were many small changes and improvements to the test suite. What's new in ZODB3 3.1.2 final? ================================ Fixed bug in FileStorage pack that caused it to fail if it encountered an old undo record (status "u"). Fixed several bugs in FileStorage pack that could cause OverflowErrors for storages > 2 GB. Fixed memory leak in TimeStamp.laterThan() that only occurred when it had to create a new TimeStamp. Fixed two BTree bugs that were fixed on the head a while ago: - bug in fsBTree that would cause byValue searches to end early. (fsBTrees are never used this way, but it was still a bug.) - bug that lead to segfault if BTree was mutated via deletion while it was being iterated over. What's new in ZODB3 3.1.2 beta 2? ================================= Fixed critical race conditions in ZEO's cache consistency code that could cause invalidations to be lost or stale data to be written to the cache. These bugs can lead to data loss or data corruption. These bugs are relatively unlikely to be provoked in sites with few conflicts, but the possibility of failure existed any time an object was loaded and stored concurrently. Fixed a bug in conflict resolution that failed to ghostify an object if it was involved in a conflict. (This code may be redundant, but it has been fixed regardless.) The ZEO server was fixed so that it does not perform any I/O until all of a transactions' invalidations are queued. If it performs I/O in the middle of sending invalidations, it would be possible to overlap a load from a client with the invalidation being sent to it. The ZEO cache now handles invalidations atomically. This is the same sort of bug that is described in the 3.1.2b1 section below, but it affects the ZEO cache. Fixed several serious bugs in fsrecover that caused it to fail catastrophically in certain cases because it thought it had found a checkpoint (status "c") record when it was in the middle of the file. What's new in ZODB3 3.1.2 beta 1? ================================= ZODB ---- Invalidations are now processed atomically. Each transaction will see all the changes caused by an earlier transaction or none of them. Before this patch, it was possible for a transaction to see invalid data because it saw only a subset of the invalidations. This is the most likely cause of reported BTrees corruption, where keys were stored in the wrong bucket. When a BTree bucket splits, the bucket and the bucket's parent are both modified. If a transaction sees the invalidation for the bucket but not the parent, the BTree in memory will be internally inconsistent and keys can be put in the wrong bucket. The atomic invalidation fix prevents this problem. A number of minor reference count fixes in the object cache were fixed. That's the cPickleCache.c file. It was possible for a transaction that failed in tpc_finish() to lose the traceback that caused the failure. The transaction code was fixed to report the original error as well as any errors that occur while trying to recover from the original error. ZEO --- A ZEO client will not read from its cache during cache verification. This fix was necessary to prevent the client from reading inconsistent data. The isReadOnly() method of a ZEO client was fixed to return the false when the client is connected to a read-only fallback server. The sync() method of ClientStorage and the pending() method of a zrpc connection now do both input and output. The short_repr() function used to generate log messages was fixed so that it does not blow up creating a repr of very long tuples. Storages -------- FileStorage has a new pack() implementation that fixes several reported problems that could lead to data loss. Two small bugs were fixed in DemoStorage. undoLog() did not handle its arguments correctly and pack() could accidentally delete objects created in versions. Fixed trivial bug in fsrecover that prevented it from working at all. FileStorage will use fsync() on Windows starting with Python 2.2.3. FileStorage's commit version was fixed. It used to stop after the first object, leaving all the other objects in the version. BTrees ------ Trying to store an object of a non-integer type into an IIBTree or OIBTree could leave the bucket in a variety of insane states. For example, trying b[obj] = "I'm a string, not an integer" where b is an OIBTree. This manifested as a refcount leak in the test suite, but could have been much worse (most likely in real life is that a seemingly arbitrary existing key would "go missing"). When deleting the first child of a BTree node with more than one child, a reference to the second child leaked. This could cause the entire bucket chain to leak (not be collected as garbage despite not being referenced anymore). Other minor BTree leak scenarios were also fixed. Other ----- Comparing a Missing.Value object to a C type that provide its own comparison operation could lead to a segfault when the Missing.Value was on the right-hand side of the comparison operator. The Missing class was fixed so that its coercion and comparison operations are safe. Tools ----- Four tools are now installed by setup.py: fsdump.py, fstest.py, repozo.py, and zeopack.py. What's new in ZODB3 3.1.1 final? ================================ Release date: 11-Feb-2003 Tools ----- Updated repozo.py tool What's new in ZODB3 3.1.1 beta 2? ================================= Release date: 03-Feb-2003 The Transaction "hosed" feature is disabled in this release. If a transaction fails during the tpc_finish() it is not possible, in general, to know whether the storage is in a consistent state. For example, a ZEO server may commit the data and then fail before sending confirmation of the commit to the client. If multiple storages are involved in a transaction, the problem is exacerbated: One storage may commit the data while another fails to commit. In previous versions of ZODB, the database would set a global "hosed" flag that prevented any other transaction from committing until an administrator could check the status of the various failed storages and ensure that the database is in a consistent state. This approach favors data consistency over availability. The new approach is to log a panic but continue. In practice, availability seems to be more important than consistency. The failure mode is exceedingly rare in either case. The BTrees-based fsIndex for FileStorage is enabled. This version of the index is faster to load and store via pickle and uses less memory to store keys. We had intended to enable this feature in an earlier release, but failed to actually do it; thus, it's getting enabled as a bug fix now. Two rare bugs were fixed in BTrees conflict resolution. The most probable symptom of the bug would have been a segfault. The bugs were found via synthetic stress tests rather than bug reports. A value-based consistency checker for BTrees was added. See the module BTrees.check for the checker and other utilities for working with BTrees. A new script called repozo.py was added. This script, originally written by Anthony Baxter, provides an incremental backup scheme for FileStorage based storages. zeopack.py has been fixed to use a read-only connection. Various small autopack-related race conditions have been fixed in the Berkeley storage implementations. There have been some table changes to the Berkeley storages so any storage you created in 3.1.1b1 may not work. Part of these changes was to add a storage version number to the schema so these types of incompatible changes can be avoided in the future. Removed the chance of bogus warnings in the FileStorage iterator. ZEO --- The ZEO version number was bumped to 2.0.2 on account of the following minor feature additions. The performance of full cache verification has improved dramatically. Measurements from Jim were somewhere in 2x-5x. The implementation was fixed to use the very-fast getSerial() method on the storage instead of the comparatively slow load(). The ZEO server has an optional timeout feature that will abort a connection that does not commit within a certain amount of time. The timeout works by closing the socket the client is using, causing both client and server to abort the transaction and continue. This is a drastic step, but can be useful to prevent a hung client or other bug from blocking a server indefinitely. If a client was disconnected during a transaction, the tpc_abort() call did not properly reset the internal state about the transaction. The bug caused the next transaction to fail in its tpc_finish(). Also, any ClientDisconnected exceptions raised during tpc_abort() are ignored. ZEO logging has been improved by adding more logging for important events, and changing the logging level for existing messages to a more appropriate level (usually lower). What's new in ZODB3 3.1.1 beta 1? ================================= Release date: 10-Dev-2002 It was possible for earlier versions of ZODB to deadlock when using multiple storages. If multiple transactions committed concurrently and both transactions involved two or more shared storages, deadlock was possible. This problem has been fixed by introducing a sortKey() method to the transaction and storage APIs that is used to define an ordering on transaction participants. This solution will prevent deadlocks provided that all transaction participants that use locks define a valid sortKey() method. A warning is raised if a participant does not define sortKey(). For backwards compatibility, BaseStorage provides a sortKey() that uses __name__. Added code to ThreadedAsync/LoopCallback.py to work around a bug in asyncore.py: a handled signal can cause unwanted reads to happen. A bug in FileStorage related to object uncreation was fixed. If an a transaction that created an object was undone, FileStorage could write a bogus data record header that could lead to strange errors if the object was loaded. An attempt to load an uncreated object now raises KeyError, as expected. The restore() implementation in FileStorage wrote incorrect backpointers for a few corner cases involving versions and undo. It also failed if the backpointer pointed to a record that was before the pack time. These specific bugs have been fixed and new test cases were added to cover them. A bug was fixed in conflict resolution that raised a NameError when a class involved in a conflict could not be loaded. The bug did not affect correctness, but prevent ZODB from caching the fact that the class was unloadable. A related bug prevented spurious AttributeErrors when a class could not be loaded. It was also fixed. The script Tools/zeopack.py was fixed to work with ZEO 2. It was untested and had two silly bugs. Some C extensions included standard header files before including Python.h, which is not allowed. They now include Python.h first, which eliminates compiler warnings in certain configurations. The BerkeleyDB based storages have been merged from the trunk, providing a much more robust version of the storages. They are not backwards compatible with the old storages, but the decision was made to update them in this micro release because the old storages did not work for all practical purposes. For details, see Doc/BDBStorage.txt. What's new in ZODB3 3.1 final? =============================== Release date: 28-Oct-2002 If an error occurs during conflict resolution, the store will silently catch the error, log it, and continue as if the conflict was unresolvable. ZODB used to behave this way, and the change to catch only ConflictError was causing problems in deployed systems. There are a lot of legitimate errors that should be caught, but it's too close to the final release to make the substantial changes needed to correct this. What's new in ZODB3 3.1 beta 3? =============================== Release date: 21-Oct-2002 A small extension was made to the iterator protocol. The Record objects, which are returned by the per-transaction iterators, contain a new `data_txn` attribute. It is None, unless the data contained in the record is a logical copy of an earlier transaction's data. For example, when transactional undo modifies an object, it creates a logical copy of the earlier transaction's data. Note that this provide a stronger statement about consistency than whether the data in two records is the same; it's possible for two different updates to an object to coincidentally have the same data. The restore() method was extended to take the data_txn attribute mentioned above as an argument. FileStorage uses the new argument to write a backpointer if possible. A few bugs were fixed. The setattr slot of the cPersistence C API was being initialized to NULL. The proper initialization was restored, preventing crashes in some applications with C extensions that used persistence. The return value of TimeStamp's __cmp__ method was clipped to return only 1, 0, -1. The restore() method was fixed to write a valid backpointer if the update being restored is in a version. Several bugs and improvements were made to zdaemon, which can be used to run the ZEO server. The parent now forwards signals to the child as intended. Pidfile handling was improved and the trailing newline was omitted. What's new in ZODB3 3.1 beta 2? =============================== Release date: 4-Oct-2002 A few bugs have been fixed, some that were found with the help of Neal Norwitz's PyChecker. The zeoup.py tool has been fixed to allow connecting to a read-only storage, when the --nowrite option is given. Casey Duncan fixed a few bugs in the recent changes to undoLog(). The fstest.py script no longer checks that each object modified in a transaction has a serial number that matches the transaction id. This invariant is no longer maintained; several new features in the 3.1 release depend on it. The ZopeUndo package was added. If ZODB3 is being used to run a ZEO server that will be used with Zope, it is usually best if the server and the Zope client don't share any software. The Zope undo framework, however, requires that a Prefix object be passed between client and server. To support this use, ZopeUndo was created to hold the Prefix object. Many bugs were fixed in ZEO, and a couple of features added. See `ZEO-NEWS.txt` for details. The ZODB guide included in the Doc directory has been updated. It is still incomplete, but most of the references to old ZODB packages have been removed. There is a new section that briefly explains how to use BTrees. The zeoup.py tool connects using a read-only connection when --nowrite is specifified. This feature is useful for checking on read-only ZEO servers. What's new in ZODB3 3.1 beta 1? =============================== Release date: 12-Sep-2002 We've changed the name and version number of the project, but it's still the same old ZODB. There have been a lot of changes since the last release. New ZODB cache -------------- Toby Dickenson implemented a new Connection cache for ZODB. The cache is responsible for pointer swizzling (translating between oids and Python objects) and for keeping recently used objects in memory. The new cache is a big improvement over the old cache. It strictly honors its size limit, where size is specified in number of objects, and it evicts objects in least recently used (LRU) order. Users should take care when setting the cache size, which has a default value of 400 objects. The old version of the cache often held many more objects than its specified size. An application may not perform as well with a small cache size, because the cache no longer exceeds the limit. Storages -------- The index used by FileStorage was reimplemented using a custom BTrees object. The index maps oids to file offsets, and is kept in memory at all times. The new index uses about 1/4 the memory of the old, dictionary-based index. See the module ZODB.fsIndex for details. A security flaw was corrected in transactionalUndo(). The transaction ids returned by undoLog() and used for transactionalUndo() contained a file offset. An attacker could construct a pickle with a bogus transaction record in its binary data, deduce the position of the pickle in the file from the undo log, then submit an undo with a bogus file position that caused the pickle to get written as a regular data record. The implementation was fixed so that file offsets are not included in the transaction ids. Several storages now have an explicit read-only mode. For example, passing the keyword argument read_only=1 to FileStorage will make it read-only. If a write operation is performed on a read-only storage, a ReadOnlyError will be raised. The storage API was extended with new methods that support the Zope Replication Service (ZRS), a proprietary Zope Corp product. We expect these methods to be generally useful. The methods are: - restore(oid, serialno, data, version, transaction) Perform a store without doing consistency checks. A client can use this method to provide a new current revision of an object. The ``serialno`` argument is the new serialno to use for the object, not the serialno of the previous revision. - lastTransaction() Returns the transaction id of the last committed transaction. - lastSerial(oid) Return the current serialno for ``oid`` or None. - iterator(start=None, stop=None) The iterator method isn't new, but the optional ``start`` and ``stop`` arguments are. These arguments can be used to specify the range of the iterator -- an inclusive range [start, stop]. FileStorage is now more cautious about creating a new file when it believes a file does not exist. This change is a workaround for bug in Python versions upto and including 2.1.3. If the interpreter was builtin without large file support but the platform had it, os.path.exists() would return false for large files. The fix is to try to open the file first, and decide whether to create a new file based on errno. The undoLog() and undoInfo() methods of FileStorage can run concurrently with other methods. The internal storage lock is released periodically to give other threads a chance to run. This should increase responsiveness of ZEO clients when used with ZEO 2. New serial numbers are assigned consistently for abortVersion() and commitVersion(). When a version is committed, the non-version data gets a new serial number. When a version is aborted, the serial number for non-version data does not change. This means that the abortVersion() transaction record has the unique property that its transaction id is not the serial number of the data records. Berkeley Storages ----------------- Berkeley storage constructors now take an optional `config` argument, which is an instance whose attributes can be used to configure such BerkeleyDB policies as an automatic checkpointing interval, lock table sizing, and the log directory. See bsddb3Storage/BerkeleyBase.py for details. A getSize() method has been added to all Berkeley storages. Berkeley storages open their environments with the DB_THREAD flag. Some performance optimizations have been implemented in Full storage, including the addition of a helper C extension when used with Python 2.2. More performance improvements will be added for the ZODB 3.1 final release. A new experimental Autopack storage was added which keeps only a certain amount of old revision information. The concepts in this storage will be folded into Full and Autopack will likely go away in ZODB 3.1 final. ZODB 3.1 final will also have much improved Minimal and Full storages, which eliminate Berkeley lock exhaustion problems, reduce memory use, and improve performance. It is recommended that you use BerkeleyDB 4.0.14 and PyBSDDB 3.4.0 with the Berkeley storages. See bsddb3Storage/README.txt for details. BTrees ------ BTrees no longer ignore exceptions raised when two keys are compared. Tim Peters fixed several endcase bugs in the BTrees code. Most importantly, after a mix of inserts and deletes in a BTree or TreeSet, it was possible (but unlikely) for the internal state of the object to become inconsistent. Symptoms then varied; most often this manifested as a mysterious failure to find a key that you knew was present, or that tree.keys() would yield an object that disgreed with the tree about how many keys there were. If you suspect such a problem, BTrees and TreeSets now support a ._check() method, which does a thorough job of examining the internal tree pointers for consistency. It raises AssertionError if it finds any problems, else returns None. If ._check() raises an exception, the object is damaged, and rebuilding the object is the best solution. All known ways for a BTree or TreeSet object to become internally inconsistent have been repaired. Other fixes include: - Many fixes for range search endcases, including the "range search bug:" If the smallest key S in a bucket in a BTree was deleted, doing a range search on the BTree with S on the high end could claim that the range was empty even when it wasn't. - Zope Collector #419: repaired off-by-1 errors and IndexErrors when slicing BTree-based data structures. For example, an_IIBTree.items()[0:0] had length 1 (should be empty) if the tree wasn't empty. - The BTree module functions weightedIntersection() and weightedUnion() now treat negative weights as documented. It's hard to explain what their effects were before this fix, as the sign bits were getting confused with an internal distinction between whether the result should be a set or a mapping. ZEO ---- For news about ZEO2, see the file ZEO-NEWS.txt. This version of ZODB ships with two different versions of ZEO. It includes ZEO 2.0 beta 1, the recommended new version. (ZEO 2 will reach final release before ZODB3.) The ZEO 2.0 protocol is not compatible with ZEO 1.0, so we have also included ZEO 1.0 to support people already using ZEO 1.0. Other features -------------- When a ConflictError is raised, the exception object now has a sensible structure, thanks to a patch from Greg Ward. The exception now uses the following standard attributes: oid, class_name, message, serials. See the ZODB.POSException.ConflictError doc string for details. It is now easier to customize the registration of persistent objects with a transaction. The low-level persistence mechanism in cPersistence.c registers with the object's jar instead of with the current transaction. The jar (Connection) then registers with the transaction. This redirection would allow specialized Connections to change the default policy on how the transaction manager is selected without hacking the Transaction module. Empty transactions can be committed without interacting with the storage. It is possible for registration to occur unintentionally and for a persistent object to compensate by making itself as unchanged. When this happens, it's possible to commit a transaction with no modified objects. The change allows such transactions to finish even on a read-only storage. Two new tools were added to the Tools directory. The ``analyze.py`` script, based on a tool by Matt Kromer, prints a summary of space usage in a FileStorage Data.fs. The ``checkbtrees.py`` script scans a FileStorage Data.fs. When it finds a BTrees object, it loads the object and calls the ``_check`` method. It prints warning messages for any corrupt BTrees objects found. Documentation ------------- The user's guide included with this release is still woefully out of date. Other bugs fixed ---------------- If an exception occurs inside an _p_deactivate() method, a traceback is printed on stderr. Previous versions of ZODB silently cleared the exception. ExtensionClass and ZODB now work correctly with a Python debug build. All C code has been fixed to use a consistent set of functions from the Python memory API. This allows ZODB to be used in conjunction with pymalloc, the default allocator in Python 2.3. zdaemon, which can be used to run a ZEO server, more clearly reports the exit status of its child processes. The ZEO server will reinitialize zLOG when it receives a SIGHUP. This allows log file rotation without restarting the server. What's new in StandaloneZODB 1.0 final? ======================================= Release date: 08-Feb-2002 All copyright notices have been updated to reflect the fact that the ZPL 2.0 covers this release. Added a cleanroom PersistentList.py implementation, which multiply inherits from UserDict and Persistent. Some improvements in setup.py and test.py for sites that don't have the Berkeley libraries installed. A new program, zeoup.py was added which simply verifies that a ZEO server is reachable. Also, a new program zeopack.py was added which connects to a ZEO server and packs it. What's new in StandaloneZODB 1.0 c1? ==================================== Release Date: 25-Jan-2002 This was the first public release of the StandaloneZODB from Zope Corporation. Everything's new! :)