- repozo: fix restoring on stdout.
- repozo: prevent an incorrect "option ignored" warning when running backup or verify.
- Add support for Python 3.12.
- Drop support for Python 2.7, 3.5, 3.6.
- Fix sorting issue in
scripts/space.py
. - FileStorage: fix a rare data corruption when using restore after multiple undos. For details see #395.
- Fix exit code of
repozo
script in case of verification error. For details see #396.
- Fix
racetest
problems. For details see #376. - Fix
--with-verify
argument in script repozo--recover
. For details see #381.
- Add support for Python 3.11.
- Expand and refactor tests for race conditions.
- Fix
TypeError: can't concat str to bytes
when running fsoids.py script with Python 3. See issue 350. - Readd transaction size information to
fsdump
output; adapt fsstats tofsdump
's exchanged order forsize
andclass
information in data records; (fixes #354 <https://github.com/zopefoundation/ZODB/issues/354>_). Make ``fsdump` callable via Python's-m
command line option. - Fix UnboundLocalError when running fsoids.py script. See issue 285.
- Rework
fsrefs
script to work significantly faster by optimizing how it does IO. See PR 340. - Require Python 3 to build the documentation.
- Fix deprecation warnings occurring on Python 3.10.
- Add support for Python 3.9 and 3.10.
- Fix race with invalidations when starting a new transaction. The bug affected Storage implementations that rely on mvccadapter, and could result in data corruption (oid loaded at wrong serial after a concurrent commit). See issue 290. As mentionned in pull request #307, interfaces are clarified about the fact that storage implementations must update at a precise moment the value that is returned by lastTransaction(): just after invalidate() or tpc_finish callback.
- Improve volatile attribute
_v_
documentation. - Make repozo's recover mode atomic by recovering the backup in a temporary file which is then moved to the expected output file.
- Add a new option to repozo in recover mode which allows to verify backups integrity on the fly.
- Drop support for Python 3.4.
- Add support for Python 3.8.
- Fix
DB.undo()
andDB.undoMultiple()
to close the storage they open behind the scenes when the transaction is committed or rolled back. See issue 268. - Make TransactionMetaData in charge of (de)serializing extension data.
A new
extension_bytes
attribute converts automatically fromextension
, or vice-versa. During storage iteration,extension_bytes
holds bytes as they are stored (i.e. no deserialization happens). See issue 207. - Make a connection's savepoint storage implement its own
(approximate)
getSize
method instead of relying on the original storage. Previously, this produced confusing DEBUG logging. See issue 282. - Fix tests with transaction 3.0.
- Fix inconsistent resolution order with zope.interface v5.
- Remove
ConnectionPool.map()
. Instead,ConnectionPool
is now iterable. See PR 280.
- Fix KeyError on releasing resources of a Connection when closing the DB.
This requires at least version 2.4 of the
transaction
package. See issue 208.
- Add support for Python 3.7.
- Bump the dependency on zodbpickle to at least 1.0.1. This is required to avoid a memory leak on Python 2.7. See issue 203.
- Bump the dependency on persistent to at least 4.4.0.
- Make the internal support functions for dealing with OIDs (
p64
andu64
) somewhat faster and raise more informative exceptions on certain types of bad input. See issue 216. - Remove support for
python setup.py test
. It hadn't been working for some time. See issue #218. - Make the tests run faster by avoiding calls to
time.sleep()
.
ZODB now uses pickle protocol 3 for both Python 2 and Python 3.
(Previously, protocol 2 was used for Python 2.)
The zodbpickle package provides a zodbpickle.binary string type that should be used in Python 2 to cause binary strings to be saved in a pickle binary format, so they can be loaded correctly in Python 3. Pickle protocol 3 is needed for this to work correctly.
Object identifiers in persistent references are saved as zodbpickle.binary strings in Python 2, so that they are loaded correctly in Python 3.
If an object is missing from the index while packing a
FileStorage
, report its fulloid
.Storage imports are a bit faster.
Storages can be important from non-seekable sources, like file-wrapped pipes.
Add support for Python 3.6.
Drop support for Python 3.3.
Ensure that the
HistoricalStorageAdapter
forwards therelease
method to its base instance. See issue 78.Use a higher pickle protocol (2) for serializing objects on Python 2; previously protocol 1 was used. This is much more efficient for new-style classes (all persistent objects are new-style), at the cost of being very slightly less efficient for old-style classes.
Note
On Python 2, this will now allow open
file
objects (but not open blobs or sockets) to be pickled (loading the object will result in a closed file); previously this would result in aTypeError
. Doing so is not recommended as they cannot be loaded in Python 3.See issue 179.
DB.close
now explicitly frees internal resources. This is helpful to avoid false positives in tests that check for leaks.- Optimize getting the path to a blob file. See issue 161.
- All classes are new-style classes on Python 2 (they were already new-style on Python 3). This improves performance on PyPy. See issue 160.
- Fix an import error. See issue 158.
- Fixed: A blob misfeature set blob permissions so that blobs and blob
directories were only readable by the database process owner, rather
than honoring user-controlled permissions (e.g.
umask
). See issue 155.
- Fixed: When opening FileStorages in read-only mode, non-existent files were silently created. Creating a read-only file-storage against a non-existent file errors.
- Call new afterCompletion API on storages to allow them to free resources after transaction complete. See issue 147.
- Take advantage of the new transaction-manager explicit mode to avoid starting transactions unnecessarily when transactions end.
Connection.new_oid
delegates to its storage, not the DB. This is helpful for improving concurrency in MVCC storages like RelStorage. See issue 139.persistent
is no longer required at setup time. See issue 119.Connection.close
andConnection.open
no longer race onself.transaction_manager
, which could lead toAttributeError
. This was a bug introduced in 5.0.1. See issue 142.
- Fixed:
ZODB.Connection.TransactionMetaData
didn't support custom data storage that some storages rely on.
ZODB now translates transaction meta data,
user
anddescription
from text to bytes before passing them to storages, and converts them back to text when retrieving them from storages in thehistory
,undoLog
andundoInfo
methods.The
IDatabase
interface was updated to reflect thathistory
,undoLog
andundoInfo
are available on database objects. (They were always available, but not documented in the interface.)
- Fix an AttributeError that DemoStorage could raise if it was asked to store a blob into a temporary changes before reading a blob. See issue 103.
- Call _p_resolveConflict() even if a conflicting change doesn't change the state. This reverts to the behaviour of 3.10.3 and older.
- Closing a Connection now reverts its
transaction_manager
to None. This helps prevent errors and release resources when thetransaction_manager
was the (default) thread-local manager. See issue 114. - Many docstrings have been improved.
Major internal improvements and cleanups plus:
Added a connection
prefetch
method that can be used to request that a storage prefetch data an application will need:conn.prefetch(obj, ...)
Where arguments can be objects, object ids, or iterables of objects or object ids.
Added optional
prefetch
methods to the storage APIs. If a storage doesn't support prefetch, then the connection prefetch method is a noop.fstail: print the txn offset and header size, instead of only the data offset. fstail can now be used to truncate a DB at the right offset.
Drop support for old commit protocol. All of the build-in storages implement the new protocol. This new protocol allows storages to provide better write performance by allowing multiple commits to execute in parallel.
- fstail: print the txn offset and header size, instead of only the data offset. fstail can now be used to truncate a DB at the right offset.
Numerous internal cleanups, including:
- Changed the way the root object was created. Now the root object is created using a database connection, rather than by making low-level storage calls.
- Drop support for the old commit protocol.
- Internal FileStorage-undo fixes that should allow undo in some cases where it didn't work before.
- Drop the
version
argument to some methods where it was the last argument and optional.
Added a connection
prefetch
method that can be used to request that a storage prefect data an application will need:conn.prefetch(obj, ...)
Where arguments can be objects, object ids, or iterables of objects or object ids.
Added optional
prefetch
methods to the storage APIs. If a storage doesn't support prefetch, then the connection prefetch method is a noop.
Drop support for old commit protocol. All of the build-in storages implement the new protocol. This new protocol allows storages to provide better write performance by allowing multiple commits to execute in parallel.
See 4.4.2.
See 4.4.1.
See 4.4.0.
Major internal implementation changes to the Multi Version Concurrency Control (MVCC) implementation:
For storages that implement IMVCCStorage (RelStorage), no longer implement MVCC in ZODB.
For other storages, MVCC is implemented using an additional storage layer. This underlying layer works by calling
loadBefore
. The low-level storageload
method isn't used any more.This change allows server-based storages like ZEO and NEO to be implemented more simply and cleanly.
- Internal FileStorage-undo fixes that should allow undo in some cases where it didn't work before.
- fstail: print the txn offset and header size, instead of only the data offset. fstail can now be used to truncate a DB at the right offset.
Better support of the new commit protocol. This fixes issues with blobs and undo. See pull requests #77, #80, #83
Added IMultiCommitStorage to directly represent the changes in the 4.4.0 release and to make complient storages introspectable.
This release begins evolution to a more effcient commit protocol that allows storage implementations, like NEO, to support multiple transactions committing at the same time, for greater write parallelism.
This release updates IStorage:
- The committed transaction's ID is returned by
tpc_finish
, rather than being returned in response store and tpc_vote results. tpc_vote
is now expected to returnNone
or a list of object ids for objects for which conflicts were resolved.
This release works with storages that implemented the older version of the storage interface, but also supports storages that implement the updated interface.
- Fixed: FileStorage loadBefore didn't handle deleted/undone data correctly.
- Drop support for Python 2.6 and 3.2.
- Make the
zodbpickle
dependency required and not conditional. This fixes various packaging issues involving pip and its wheel cache. zodbpickle was only optional under Python 2.6 so this change only impacts users of that version. See #42. - Add support for Python 3.5.
- Avoid failure during cleanup of nested databases that provide MVCC on storage level (Relstorage). #45
- Remove useless dependency to zdaemon in setup.py. Remove ZEO documentation. Both were leftovers from the time where ZEO was part of this repository.
- Fix possible data corruption after FileStorage is truncated to roll back a transaction. #52
- DemoStorage: add support for conflict resolution and fix history() #58
- Fixed a test that depended on implementation-specific behavior in tpc_finish
- Declare conditional dependencies using PEP-426 environment markers (fixing interation between pip 7's wheel cache and tox). See #36.
- Log failed conflict resolution attempts at
DEBUG
level. See: #29. - Fix command-line parsing of
--verbose
and--verify
arguments. (The short versions,-v
and-V
, were parsed correctly.) - Add support for PyPy.
- Fix the methods in
ZODB.serialize
that find object references under Python 2.7 (used in scripts likereferrers
,netspace
, andfsrecover
among others). This requires the addition of thezodbpickle
dependency. - FileStorage: fix an edge case when disk space runs out while packing,
do not leave the
.pack
file around. That would block any write to the to-be-packedData.fs
, because the disk would stay at 0 bytes free. See #21.
Fix registration of custom logging level names ("BLATHER", "TRACE").
We have been registering them in the wrong order since 2004. Before Python 3.4, the stdlib
logging
module masked the error by registering them in both directions.Add support for Python 3.4.
- Fix
POSKeyError
duringtransaction.commit
when aftersavepoint.rollback
. See #16 - Ensure that the pickler used in PyPy always has a
persistent_id
attribute (inst_persistent_id
is not present on the pure-Python pickler). (PR #17) - Provide better error reporting when trying to load an object on a closed connection.
Finally released.
- Switch to using non-backward-compatible pickles (protocol 3, without storing bytes as strings) under Python 3. Updated the magic number for file-storage files under Python3 to indicate the incompatibility.
- Fixed: A
UnicodeDecodeError
could happen for non-ASCII OIDs when using bushy blob layout.
- Extended the filename renormalizer used for blob doctests to support the filenames used by ZEO in non-shared mode.
- Added
url
parameter tosetup()
(PyPI says it is required).
- Skipped non-unit tests in
setup.py test
. Use the buildout to run tests requiring "layer" support. - Included the filename in the exception message to support debugging in case
loadBlob
does not find the file. - Added support for Python 3.2 / 3.3.
Note
ZODB 4.0.x is supported on Python 3.x for new applications only. Due to changes in the standard library's pickle support, the Python3 support does not provide forward- or backward-compatibility at the data level with Python2. A future version of ZODB may add such support.
Applications which need migrate data from Python2 to Python3 should plan to script this migration using separte databases, e.g. via a "dump-and-reload" approach, or by providing explicit fix-ups of the pickled values as transactions are copied between storages.
- 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
persistent
andBTrees
packages 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
repozo
share the same timestamp.Pinned the
transaction
andmanuel
dependencies to Python 2.5- compatible versions when installing under Python 2.5.
Note
Please see https://github.com/zopefoundation/ZODB/blob/master/HISTORY.rst for older versions of ZODB.