- 19 Feb, 2004 12 commits
-
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
Include ZODB.serialize tests.
-
Jeremy Hylton authored
DemoStorage is too frail to bother using.
-
Jeremy Hylton authored
Blast! Why didn't doctest complain about this?
-
Jeremy Hylton authored
- fold in not-actually-shared pickle support code into cPersistence.c, removing pickle.c - remove __getnewargs__(), since we want to allow ghost creation to avoid the extra database lookup when possible - persistent objects no longer get a default __getnewargs__()
-
Jeremy Hylton authored
- inline getClassMetadata() into serialize() - get __getnewargs__() from the object instance, not the class, since the PersistentClassMetaClass does things strangely
-
Jeremy Hylton authored
-
Jeremy Hylton authored
Added support for persistent weak references and PersistentWeakKeyDictionary. Add _p_invalidate() method to persistence API. _p_deactivate() is advisory from the cache. _p_invalidate() is called when database invalidates object. Port support for getattr/setattr hacks from ZODB4. The doctest tests describe the new technique pretty well. Remove unused arguments from some cPickleCache calls.
-
Jeremy Hylton authored
Add support for persistent weak references. The implementation is in pure Python right now; coptimizations.c was disabled. We need to restore the C code before going to beta. The persistent reference format has evolved a little, but the code on the branch doesn't document it. The ref is usually a an oid-class pair, where the class is an actual class object. It can also be a list, for weak references, or a bare oid. Add support for data managers written against the ZODB4 transaction API. The transaction implementation provides an adapter between the two interfaces. There's a substantial doctest test suite for this feature. Add add() method to Connection to explicitly add an object and get it an assigned _p_oid. It's possible to add an unconnected object; this isn't recommended, because it will be lost at pack time.
-
- 18 Feb, 2004 6 commits
-
-
Chris McDonough authored
port we plan to listen on.
-
Jeremy Hylton authored
Missed two tests in earlier checkins.
-
Tres Seaver authored
-
Jeremy Hylton authored
The old undo was not used by Zope and shouldn't have been used by any other client. The newly named undo() is the preferred version of undo. XXX DemoStorage didn't implement transactionalUndo, so now it doesn't implementation undo() at all. We need to replace it with the demo storage from ZODB4. There are a few changes related to ZODB4 removal in this checkin.
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
- 17 Feb, 2004 3 commits
-
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
- 14 Feb, 2004 1 commit
-
-
Jeremy Hylton authored
-
- 12 Feb, 2004 1 commit
-
-
Jeremy Hylton authored
I've removed all the details about Python versions prior to 2.3 and Berkeley before 4.1.25. I also removed all the configuration instructions. You "just" use the standard zconfig mechanism, which may or may not be documented elsewhere.
-
- 11 Feb, 2004 11 commits
-
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Fred Drake authored
-
Fred Drake authored
and set it in more places
-
Fred Drake authored
an older ZConfig API which has not been supported for some time
-
Fred Drake authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
-
Jeremy Hylton authored
I think it's safe to lose the revision history for these files. setup.py is a copy of ZODB3/setup.py with all the paths edited. test.py is a copy of ZODB4/test.py with no changes.
-
- 06 Feb, 2004 1 commit
-
-
Fred Drake authored
-
- 05 Feb, 2004 2 commits
-
-
Fred Drake authored
-
Fred Drake authored
-
- 04 Feb, 2004 1 commit
-
-
Jeremy Hylton authored
-
- 19 Jan, 2004 1 commit
-
-
Jeremy Hylton authored
-
- 16 Jan, 2004 1 commit
-
-
Tim Peters authored
trouble here on Windows, due to the poor granulatiry of Windows time.time() feeding into pack-time selection in this test. Moved the snooze() from before pack time selection to after it, which appears to be a sufficient fix; this can't hurt Linux, because the snooze() calls aren't needed at all there (but don't hurt either).
-