- renamed SessionTransaction autoflush to reentrant_flush to more clearly state its purpose
- added _enable_transaction_accounting, flag for Mike Bernson which disables the whole 0.5 transaction state management; the system depends on expiry on rollback in order to function.
Session whereby asynchronous GC could remove unmodified,
no longer referenced items from the session as they were
present in a list of items to be processed, typically
during session.expunge_all() and dependent methods.
- added query_cls keyword argument to sessionmaker(); allows user-defined Query subclasses to be generated by query().
- added @attributes.on_reconstitute decorator, MapperExtension.on_reconstitute, both receieve 'on_load' attribute event allowing
non-__init__ dependent instance initialization routines.
- push memusage to the top to avoid pointless heisenbugs
- renamed '_foostate'/'_fooclass_manager' to '_sa_instance_state'/'_sa_class_manager'
- removed legacy instance ORM state accessors
- query._get() will use _remove_newly_deleted instead of expunge() on ObjectDeleted, so that transaction rollback
restores the previous state
- removed MapperExtension.get(); replaced by a user-defined Query subclass
- removed needless **kwargs from query.get()
- removed Session.get(cls, id); this is redundant against Session.query(cls).get(id)
- removed Query.load() and Session.load(); the use case for this method has never been clear, and the same functionality is available in more explicit ways
- Guards are now present on all public Session methods and passing in an
unmapped hoho anywhere yields helpful exception messages, going to some
effort to provide hints for debugging situations that would otherwise seem
hopeless, such as broken user instrumentation or half-pickles.
- @unsupported now only accepts a single target and demands a reason
for not running the test.
- @exclude also demands an exclusion reason
- Greatly expanded @testing.requires.<feature>, eliminating many
decorators in the suite and signficantly easing integration of
multi-driver support.
- New ORM test base class, and a featureful base for mapped tests
- Usage of 'global' for shared setup going away, * imports as well
case it's evaluated at statement execution time to
get the value.
- expressions used in filter(), filter_by() and others,
when they make usage of a clause generated from a
relation using the identity of a child object
(e.g. filter(Parent.child==<somechild>)), evaluate
the actual primary key value of <somechild> at
execution time so that the autoflush step of the
Query can complete, thereby populating the PK value
of <somechild> in the case that <somechild> was
pending.
- cleanup of attributes.get_committed_value() to never return
the NO_VALUE value; evaluates to None
- session.transaction now always refers to the innermost active transaction, even when commit/rollback are called directly on the session transaction object.
- when preparing a two-phase transaction fails on one connection all the connections are rolled back.
- two phase transactions can now be prepared.
- session.close() didn't close all transactions when nested transactions were used.
- rollback() previously erroneously set the current transaction directly to the parent of the transaction that could be rolled back to.
- autoflush for commit() wasn't flushing for simple subtransactions.
- Importing testenv has no side effects- explicit functions provide similar behavior to the old immediate behavior of testbase
- testing.db has the configured db
- Fixed up the perf/* scripts
more than once in the unitofwork.new collection; most typically
reproduced when using a combination of inheriting mappers and
ScopedSession.mapper, as the multiple __init__ calls per instance
could save() the object with distinct _state objects
__init__() method will now raise TypeError if non-empty *args
or **kwargs are present at instance construction time (and are
not consumed by any extensions such as the scoped_session mapper),
consistent with the behavior of normal Python classes [ticket:908]
- cleanup of mapper._instance, query.instances(). mapper identifies objects which are part of the
current load using a app-unique id on the query context.
- attributes refactor; attributes now mostly use copy-on-modify instead of copy-on-load behavior,
simplified get_history(), added a new set of tests
- fixes to OrderedSet such that difference(), intersection() and others can accept an iterator
- OrderedIdentitySet passes in OrderedSet to the IdentitySet superclass for usage in difference/intersection/etc. operations so that these methods actually work with ordering behavior.
- query.order_by() takes into account aliased joins, i.e. query.join('orders', aliased=True).order_by(Order.id)
- cleanup etc.
the tree can now run the full suite of tests out of the box.
- Migrated most @supported to @fails_on, @fails_on_everything_but, or (last
resort) @unsupported. @fails_on revealed a slew of bogus test skippage,
which was corrected.
- Added @fails_on_everything_but. Yes, the first usage *was*
"fails_on_everything_but('postgres')". How did you guess!
- Migrated @supported in dialect/* to the new test-class attribute __only_on__.
- Test classes can also have __unsupported_on__ and __excluded_on__.
- topological.py cleaned up, presents three public facing functions which
return list/tuple based structures, without exposing any internals. only
the third function returns the "hierarchical" structure. when results
include "cycles" or "child" items, 2- or 3- tuples are used to represent
results.
- unitofwork uses InstanceState almost exclusively now. new and deleted lists
are now dicts which ref the actual object to provide a strong ref for the
duration that they're in those lists. IdentitySet is only used for the public
facing versions of "new" and "deleted".
- unitofwork topological sort no longer uses the "hierarchical" version of the sort
for the base sort, only for the "per-object" secondary sort where it still
helps to group non-dependent operations together and provides expected insert
order. the default sort deals with UOWTasks in a straight list and is greatly
simplified. Tests all pass but need to see if svilen's stuff still works,
one block of code in _sort_cyclical_dependencies() seems to not be needed anywhere
but i definitely put it there for a reason at some point; if not hopefully we
can derive more test coverage from that.
- the UOWEventHandler is only applied to object-storing attributes, not
scalar (i.e. column-based) ones. cuts out a ton of overhead when setting
non-object based attributes.
- InstanceState also used throughout the flush process, i.e. dependency.py,
mapper.save_obj()/delete_obj(), sync.execute() all expect InstanceState objects
in most cases now.
- mapper/property cascade_iterator() takes InstanceState as its argument,
but still returns lists of object instances so that they are not dereferenced.
- a few tricks needed when dealing with InstanceState, i.e. when loading a list
of items that are possibly fresh from the DB, you *have* to get the actual objects
into a strong-referencing datastructure else they fall out of scope immediately.
dependency.py caches lists of dependent objects which it loads now (i.e. history
collections).
- AttributeHistory is gone, replaced by a function that returns a 3-tuple of
added, unchanged, deleted. these collections still reference the object
instances directly for the strong-referencing reasons mentiontioned, but
it uses less IdentitySet logic to generate.
e.g. if you pickle a series of objects and unpickle (i.e. as in a Pylons HTTP session
or similar), they can go into a new session without any conflict
- added stricter checks around session.delete() similar to update()
- shored up some old "validate" stuff in session/uow
- session.update() raises an error when updating an instance that is already in the session with a different identity
- adjusted zoomarks lower limits so I can get a nice clean run
single UPDATE; many-to-many relations on the parent object update properly.
[ticket:841]
- it's an error to session.save() an object which is already persistent
[ticket:840]
- changed a bunch of repr(obj) calls in session.py exceptions to use mapperutil.instance_str()
as occurs within a flush operation; setting include_collections=False gives the same
result as is used when the flush determines whether or not to issue an UPDATE for the
instance's row.
deals primarily with the InstanceState and almost never with the instrumented object
directly. This reduces lookups and complexity since we need the state for just about
everything, now its the one place for everything internally.
we also merged the new weak referencing identity map, which will go out in beta6 and
we'll see how that goes !
if it's the outermost transaction, otherwise just ends without affecting
the outer transaction.
- transactional and non-transactional Session integrates better with bound
connection; a close() will ensure that connection transactional state is
the same as that which existed on it before being bound to the Session.
- changed testing connection closer to work on _ConnectionFairy instances, resulting in
pool checkins, not actual closes
- disabled session two phase test for now, needs work
- added some two-phase support to TLEngine, not tested
- TLTransaction is now a wrapper
Mike Bayer