- Enforced usage of bytes for '_p_serial' of persistent objects (fixes compatibility with recent persistent releases).
- Fixed: An elaborate test for trvial logic corrupted module state in a
- way that made other tests fail spuriously.
- An unneeded left-over setting in setup.py caused installation with pip to fail.
- The
persistentandBTreespackages are now released as separate distributions, on which ZODB now depends. - ZODB no longer depends on zope.event. It now uses ZODB.event, which uses zope.event if it is installed. You can override ZODB.event.notify to provide your own event handling, although zope.event is recommended.
- BTrees allowed object keys with insane comparison. (Comparison inherited from object, which compares based on in-process address.) Now BTrees raise TypeError if an attempt is made to save a key with comparison inherited from object. (This doesn't apply to old-style class instances.)
Ensured that the export file and index file created by
repozoshare the same timestamp.Pinned the
transactionandmanueldependencies to Python 2.5- compatible versions when installing under Python 2.5.
- Conflict resolution failed when state included cross-database persistent references with classes that couldn't be imported.
- Conflict resolution failed when state included persistent references with classes that couldn't be imported.
"activity monitor not updated for subconnections when connection returned to pool"
"Blob temp file get's removed before it should", https://bugs.launchpad.net/zodb/+bug/595378
A way this to happen is that a transaction is aborted after the commit process has started. I don't know how this would happen in the wild.
In 3.10.3, the ZEO tpc_abort call to the server is changed to be synchronous, which should address this case. Maybe there's another case.
- Improved ZEO client cache implementation to make it less likely to evict objects that are being used.
- Small (possibly negligable) reduction in CPU in ZEO storage servers to service object loads and in networking code.
3.10 introduced an optimization to try to address BTree conflict errors arrising for basing BTree keys on object ids. The optimization caused object ids allocated in aborted transactions to be reused. Unfortunately, this optimzation led to some rather severe failures in some applications. The symptom is a conflict error in which one of the serials mentioned is zero. This optimization has been removed.
See (for example): https://bugs.launchpad.net/zodb/+bug/665452
ZEO server transaction timeouts weren't logged as critical.
When a transaction rolled back a savepoint after adding objects and subsequently added more objects and committed, an error could be raised "ValueError: A different object already has the same oid" causing the transaction to fail. Worse, this could leave a database in a state where subsequent transactions in the same process would fail.
Unix domain sockets didn't work for ZEO (since the addition of IPv6 support). https://bugs.launchpad.net/zodb/+bug/663259
Removed a missfeature that can cause performance problems when using an external garbage collector with ZEO. When objects were deleted from a storage, invalidations were sent to clients. This makes no sense. It's wildly unlikely that the other connections/clients have copies of the garbage. In normal storage garbage collection, we don't send invalidations. There's no reason to send them when an external garbage collector is used.
ZEO client cache simulation misshandled invalidations causing incorrect statistics and errors.
There are a number of performance enhancements for ZEO storage servers.
FileStorage indexes use a new format. They are saved and loaded much faster and take less space. Old indexes can still be read, but new indexes won't be readable by older versions of ZODB.
The API for undoing multiple transactions has changed. To undo multiple transactions in a single transaction, pass a list of transaction identifiers to a database's undoMultiple method. Calling a database's undo method multiple times in the same transaction now raises an exception.
The ZEO protocol for undo has changed. The only user-visible consequence of this is that when ZODB 3.10 ZEO servers won't support undo for older clients.
The storage API (IStorage) has been tightened. Now, storages should raise a StorageTransactionError when invalid transactions are passed to tpc_begin, tpc_vote, or tpc_finish.
ZEO clients (
ClientStorageinstances) now work in forked processes, including those created viamultiprocessing.Processinstances.Broken objects now provide the IBroken interface.
As a convenience, you can now pass an integer port as an address to the ZEO ClientStorage constructor.
As a convenience, there's a new
clientfunction in the ZEO package for constructing a ClientStorage instance. It takes the same arguments as the ClientStorage constructor.DemoStorages now accept constructor athuments, close_base_on_close and close_changes_on_close, to control whether underlying storages are closed when the DemoStorage is closed.
Removed the dependency on zope.proxy.
Removed support for the _p_independent mini framework, which was made moot by the introduction of multi-version concurrency control several years ago.
Added support for the transaction retry convenience (transaction-manager attempts method) introduced in the
transaction1.1.0 release.Enhanced the database opening conveniences:
- You can now pass storage keyword arguments to ZODB.DB and ZODB.connection.
- You can now pass None (rather than a storage or file name) to get a database with a mapping storage.
Databases now warn when committing very large records (> 16MB). This is to try to warn people of likely design mistakes. There is a new option (large_record_size/large-record-size) to control the record size at which the warning is issued.
Added support for wrapper storages that transform pickle data. Applications for this include compression and encryption. An example wrapper storage implementation, ZODB.tests.hexstorage, was included for testing.
It is important that storage implementations not assume that storages contain pickles. Renamed IStorageDB to IStorageWrapper and expanded it to provide methods for transforming and untransforming data records. Storages implementations should use these methods to get pickle data from stored records.
Deprecated ZODB.interfaces.StorageStopIteration. Storage iterator implementations should just raise StopIteration, which means they can now be implemented as generators.
The filestorage packer configuration option noe accepts values of the form
modname:expression, allowing the use of packer factories with options.Added a new API that allows applications to make sure that current data are read. For example, with:
self._p_jar.readCurrent(ob)
A conflict error will be raised if the version of ob read by the transaction isn't current when the transaction is committed.
Normally, ZODB only assures that objects read are consistent, but not necessarily up to date. Checking whether an object is up to date is important when information read from one object is used to update another.
BTrees are an important case of reading one object to update another. Internal nodes are read to decide which leave notes are updated when a BTree is updated. BTrees now use this new API to make sure that internal nodes are up to date on updates.
When transactions are aborted, new object ids allocated during the transaction are saved and used in subsequent transactions. This can help in situations where object ids are used as BTree keys and the sequential allocation of object ids leads to conflict errors.
ZEO servers now support a server_status method for for getting information on the number of clients, lock requests and general statistics.
ZEO clients now support a client_label constructor argument and client-label configuration-file option to specify a label for a client in server logs. This makes it easier to identify specific clients corresponding to server log entries, especially when there are multiple clients originating from the same machine.
Improved ZEO server commit lock logging. Now, locking activity is logged at the debug level until the number of waiting lock requests gets above 3. Log at the critical level when the number of waiting lock requests gets above 9.
The file-storage backup script, repozo, will now create a backup index file if an output file name is given via the --output/-o option.
Added a '--kill-old-on-full' argument to the repozo backup options: if passed, remove any older full or incremental backup files from the repository after doing a full backup. (https://bugs.launchpad.net/zope2/+bug/143158)
The mkzeoinst script has been moved to a separate project:
and is no-longer included with ZODB.
Removed untested unsupported dbmstorage fossile.
ZEO servers no longer log their pids in every log message. It's just not interesting. :)
When a pool timeout was specified for a database and old connections were removed due to timing out, an error occured due to a bug in the connection cleanup logic.
When multi-database connections were no longer used and cleaned up, their subconnections weren't cleaned up properly.
ZEO didn't work with IPv6 addrsses. Added IPv6 support contributed by Martin v. Löwis.
A file storage bug could cause ZEO clients to have incorrect information about current object revisions after reconnecting to a database server.
Updated the 'repozo --kill-old-on-full' option to remove any '.index' files corresponding to backups being removed.
ZEO extension methods failed when a client reconnected to a storage. (https://bugs.launchpad.net/zodb/+bug/143344)
Clarified the return Value for lastTransaction in the case when there aren't any transactions. Now a string of 8 nulls (aka "z64") is specified.
Setting _p_changed on a blob wo actually writing anything caused an error. (https://bugs.launchpad.net/zodb/+bug/440234)
The verbose mode of the fstest was broken. (https://bugs.launchpad.net/zodb/+bug/475996)
Object ids created in a savepoint that is rolled back wren't being reused. (https://bugs.launchpad.net/zodb/+bug/588389)
Database connections didn't invalidate cache entries when conflict errors were raised in response to checkCurrentSerialInTransaction errors. Normally, this shouldn't be a problem, since there should be pending invalidations for these oids which will cause the object to be invalidated. There have been issues with ZEO persistent cache management that have caused out of date data to remain in the cache. (It's possible that the last of these were addressed in the 3.10.0b5.) Invalidating read data when there is a conflict error provides some extra insurance.
The interface, ZODB.interfaces.IStorage was incorrect. The store method should never return a sequence of oid and serial pairs.
When a demo storage push method was used to create a new demo storage and the new storage was closed, the original was (incorrectly) closed.
There were numerous bugs in the ZEO cache tracing and analysis code. Cache simulation, while not perfect, seems to be much more accurate now than it was before.
The ZEO cache trace statistics and simulation scripts have been given more descriptive names and moved to the ZEO scripts package.
BTree sets and tree sets didn't correctly check values passed to update or to constructors, causing Python to exit under certain circumstances.
Fixed bug in copying a BTrees.Length instance. (https://bugs.launchpad.net/zodb/+bug/516653)
Fixed a serious bug that caused cache failures when run with Python optimization turned on.
When using using a ClientStorage in a Storage server, there was a threading bug that caused clients to get disconnected.
On Mac OS X, clients that connected and disconnected quickly could cause a ZEO server to stop accepting connections, due to a failure to catch errors in the initial part of the connection process.
The failure to properly handle exceptions while accepting connections is potentially problematic on other platforms.
Object state management wasn't done correctly when classes implemented custom _p_deavtivate methods. (https://bugs.launchpad.net/zodb/+bug/185066)