]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
18 years agoUpdated ignores.
Jason Kirtland [Fri, 7 Sep 2007 00:20:51 +0000 (00:20 +0000)] 
Updated ignores.

18 years agomysql SETs and ENUMs now unescape embedded quotes before storage in .enums and .value...
Jason Kirtland [Thu, 6 Sep 2007 18:46:53 +0000 (18:46 +0000)] 
mysql SETs and ENUMs now unescape embedded quotes before storage in .enums and .values.  An ancient bug.

18 years agoassociationproxy relies upon a "sweep" through the attributes at the class level,
Mike Bayer [Thu, 6 Sep 2007 16:58:02 +0000 (16:58 +0000)] 
associationproxy relies upon a "sweep" through the attributes at the class level,
restored the equivalent functionality from previous releases

18 years ago- column defaults and onupdates, executing inline, will add parenthesis
Mike Bayer [Thu, 6 Sep 2007 16:37:37 +0000 (16:37 +0000)] 
- column defaults and onupdates, executing inline,  will add parenthesis
  for subqueries and other parenthesis-requiring expressions

18 years agodo the cheaper check first....
Mike Bayer [Wed, 5 Sep 2007 23:05:39 +0000 (23:05 +0000)] 
do the cheaper check first....

18 years agoFixed reflection of the empty string for mysql enums.
Jason Kirtland [Wed, 5 Sep 2007 19:39:07 +0000 (19:39 +0000)] 
Fixed reflection of the empty string for mysql enums.

18 years agomerged current entity_management brach r3457-r3462. cleans up
Mike Bayer [Wed, 5 Sep 2007 17:25:32 +0000 (17:25 +0000)] 
merged current entity_management brach r3457-r3462.  cleans up
'_state' mamangement in attributes, moves __init__() instrumntation into attributes.py,
and reduces method call overhead by removing '_state' property.
future enhancements may include _state maintaining a weakref to the instance and a
strong ref to its __dict__ so that garbage-collected instances can get added to 'dirty',
when weak-referenced identity map is used.

18 years agoSmall change in ActiveMapper to make it work with Python 2.3
Paul Johnston [Tue, 4 Sep 2007 22:35:00 +0000 (22:35 +0000)] 
Small change in ActiveMapper to make it work with Python 2.3

18 years agoremove unused method
Mike Bayer [Tue, 4 Sep 2007 18:43:53 +0000 (18:43 +0000)] 
remove unused method

18 years agobuildbot reported a lower number for test 1a...
Mike Bayer [Tue, 4 Sep 2007 18:20:48 +0000 (18:20 +0000)] 
buildbot reported a lower number for test 1a...

18 years ago- took out method calls for oid_column
Mike Bayer [Tue, 4 Sep 2007 18:07:16 +0000 (18:07 +0000)] 
- took out method calls for oid_column
- reduced complexity of parameter handling during execution; __distill_params does all
parameter munging, executioncontext.parameters always holds a list of parameter structures
(lists, tuples, or dicts).

18 years agoSet supports_sane_multi_rowcount for MSSQL
Paul Johnston [Tue, 4 Sep 2007 14:45:36 +0000 (14:45 +0000)] 
Set supports_sane_multi_rowcount for MSSQL

18 years ago- removed "parameters" argument from clauseelement.compile(), replaced with
Mike Bayer [Tue, 4 Sep 2007 00:08:57 +0000 (00:08 +0000)] 
- removed "parameters" argument from clauseelement.compile(), replaced with
  "column_keys".  the parameters sent to execute() only interact with the
  insert/update statement compilation process in terms of the column names
  present but not the values for those columns.
  produces more consistent execute/executemany behavior, simplifies things a
  bit internally.

18 years ago- various cruft removal and optimizations to load process.
Mike Bayer [Mon, 3 Sep 2007 18:18:09 +0000 (18:18 +0000)] 
- various cruft removal and optimizations to load process.
removes about 15K method calls from masseagerload.py test.

18 years agoMinor fixes to MSSQL reflection
Paul Johnston [Mon, 3 Sep 2007 13:57:08 +0000 (13:57 +0000)] 
Minor fixes to MSSQL reflection

18 years ago- mapper compilation has been reorganized such that most compilation
Mike Bayer [Sun, 2 Sep 2007 19:55:33 +0000 (19:55 +0000)] 
- mapper compilation has been reorganized such that most compilation
  occurs upon mapper construction.  this allows us to have fewer
  calls to mapper.compile() and also to allow class-based properties
  to force a compilation (i.e. User.addresses == 7 will compile all
  mappers; this is [ticket:758]).  The only caveat here is that
  an inheriting mapper now looks for its inherited mapper upon construction;
  so mappers within inheritance relationships need to be constructed in
  inheritance order (which should be the normal case anyway).

18 years agolowered value, fewer calls on pybot ?
Mike Bayer [Sat, 1 Sep 2007 23:10:13 +0000 (23:10 +0000)] 
lowered value, fewer calls on pybot ?

18 years agoentity refs
Mike Bayer [Sat, 1 Sep 2007 23:08:30 +0000 (23:08 +0000)] 
entity refs

18 years agoedits
Mike Bayer [Sat, 1 Sep 2007 23:07:46 +0000 (23:07 +0000)] 
edits

18 years agoadded zoomark profile
Mike Bayer [Sat, 1 Sep 2007 22:48:07 +0000 (22:48 +0000)] 
added zoomark profile

18 years agoadjusting firebird, obviously needs someone to test
Mike Bayer [Sat, 1 Sep 2007 22:46:16 +0000 (22:46 +0000)] 
adjusting firebird, obviously needs someone to test

18 years agofix to oracle sequence exec
Mike Bayer [Sat, 1 Sep 2007 22:44:14 +0000 (22:44 +0000)] 
fix to oracle sequence exec

18 years agosequence pre-executes dont create an ExecutionContext, use straight cursor
Mike Bayer [Sat, 1 Sep 2007 22:42:51 +0000 (22:42 +0000)] 
sequence pre-executes dont create an ExecutionContext, use straight cursor

18 years agofactored out uses_sequences_for_inserts() into
Mike Bayer [Sat, 1 Sep 2007 21:25:46 +0000 (21:25 +0000)] 
factored out uses_sequences_for_inserts() into
preexecute_sequence dialect attribute

18 years ago- got all examples working
Mike Bayer [Sat, 1 Sep 2007 21:21:29 +0000 (21:21 +0000)] 
- got all examples working
- inline default execution occurs for *all* non-PK columns
unconditionally - preexecute only for non-executemany PK cols on
PG, Oracle, etc.
- new default docs

18 years agowhats a big commit without some errant print statements ? :)
Mike Bayer [Sat, 1 Sep 2007 19:51:25 +0000 (19:51 +0000)] 
whats a big commit without some errant print statements ? :)

18 years ago- merged inline inserts branch
Mike Bayer [Sat, 1 Sep 2007 19:49:26 +0000 (19:49 +0000)] 
- merged inline inserts branch
- all executemany() style calls put all sequences and SQL defaults inline into a single SQL statement
and don't do any pre-execution
- regular Insert and Update objects can have inline=True, forcing all executions to be inlined.
- no last_inserted_ids(), lastrow_has_defaults() available with inline execution
- calculation of pre/post execute pushed into compiler; DefaultExecutionContext greatly simplified
- fixed postgres reflection of primary key columns with no sequence/default generator, sets autoincrement=False
- fixed postgres executemany() behavior regarding sequences present, not present, passivedefaults, etc.
- all tests pass for sqlite, mysql, postgres; oracle tests pass as well as they did previously including all
insert/update/default functionality

18 years agoupdate on pool status
Mike Bayer [Fri, 31 Aug 2007 20:57:12 +0000 (20:57 +0000)] 
update on pool status

18 years ago- fixed bugs in determining proper sync clauses from custom inherit
Mike Bayer [Thu, 30 Aug 2007 15:55:41 +0000 (15:55 +0000)] 
- fixed bugs in determining proper sync clauses from custom inherit
  conditions [ticket:769]

18 years agoremove() should issue a close() on existing session
Mike Bayer [Thu, 30 Aug 2007 00:22:17 +0000 (00:22 +0000)] 
remove() should issue a close() on existing session

18 years agoExtended 'engine_from_config' coercion for QueuePool size / overflow. [ticket:763]
Jason Kirtland [Wed, 29 Aug 2007 23:17:36 +0000 (23:17 +0000)] 
Extended 'engine_from_config' coercion for QueuePool size / overflow. [ticket:763]
Added a set of coercion tests.

18 years agoengine.url cleanups [ticket:742]
Jason Kirtland [Wed, 29 Aug 2007 22:27:45 +0000 (22:27 +0000)] 
engine.url cleanups [ticket:742]
- translate_connect_args can now take kw args or the classic list
- in-tree dialects updated to supply their overrides as keywords
- tweaked url parsing in the spirit of the #742 patch, more or less

18 years agoFix for scoped_session's `mapper(extension=<scalar>)` [ticket:760]
Jason Kirtland [Wed, 29 Aug 2007 20:23:02 +0000 (20:23 +0000)] 
Fix for scoped_session's `mapper(extension=<scalar>)` [ticket:760]

18 years agoCatch-up entries for b5.
Jason Kirtland [Tue, 28 Aug 2007 23:52:48 +0000 (23:52 +0000)] 
Catch-up entries for b5.

18 years ago- Restored reflection for mysql VIEWs [ticket:748]
Jason Kirtland [Tue, 28 Aug 2007 23:44:00 +0000 (23:44 +0000)] 
- Restored reflection for mysql VIEWs [ticket:748]
- Fixed anonymous pk reflection for mysql 5.1
- Tested table and view reflection against the 'sakila' database from
  MySQL AB on 3.23 - 6.0. (with some schema adjustments, obviously)
  Maybe this will go into the SA test suite someday.
- Tweaked mysql server version tuplification, now also splitting on hyphens
- Light janitorial

18 years agoFixed OrderedProperties pickling [ticket:762]
Jason Kirtland [Tue, 28 Aug 2007 15:17:26 +0000 (15:17 +0000)] 
Fixed OrderedProperties pickling [ticket:762]

18 years agoFixed signature for orm's BETWEEN operator.
Jason Kirtland [Mon, 27 Aug 2007 19:14:26 +0000 (19:14 +0000)] 
Fixed signature for orm's BETWEEN operator.

18 years agofix typos in assoc_proxy doc
Gaëtan de Menten [Mon, 27 Aug 2007 10:15:36 +0000 (10:15 +0000)] 
fix typos in assoc_proxy doc

18 years agoExpand custom assocproxy getter/setter support to scalar proxies
Jason Kirtland [Thu, 23 Aug 2007 15:48:51 +0000 (15:48 +0000)] 
Expand custom assocproxy getter/setter support to scalar proxies

18 years agoAllow custom getter/setters to be specified for a standard AssociationProxy
Jason Kirtland [Thu, 23 Aug 2007 15:35:03 +0000 (15:35 +0000)] 
Allow custom getter/setters to be specified for a standard AssociationProxy

18 years agoFix for reflecting mysql keys that have USING
Jason Kirtland [Wed, 22 Aug 2007 22:20:56 +0000 (22:20 +0000)] 
Fix for reflecting mysql keys that have USING

18 years agobump
Mike Bayer [Wed, 22 Aug 2007 20:39:32 +0000 (20:39 +0000)] 
bump

18 years ago- ugh ! beta4 is double logging....fixed that....
Mike Bayer [Wed, 22 Aug 2007 20:35:40 +0000 (20:35 +0000)] 
- ugh  ! beta4 is double logging....fixed that....
- added test/fixed eager aliasizing for self-referential m2m relations

18 years agochangeset about connection pool rel_0_4beta4
Mike Bayer [Wed, 22 Aug 2007 19:36:54 +0000 (19:36 +0000)] 
changeset about connection pool

18 years agorestored WeakValueDict for threadlocal connections + profiler test, addressing
Mike Bayer [Wed, 22 Aug 2007 19:14:30 +0000 (19:14 +0000)] 
restored WeakValueDict for threadlocal connections + profiler test, addressing
[ticket:754]

18 years ago- a "collection-holding" InstrumentedAttribute is now identified
Mike Bayer [Wed, 22 Aug 2007 18:21:25 +0000 (18:21 +0000)] 
- a "collection-holding" InstrumentedAttribute is now identified
by the presence of a "get_collection" method.
- added "get_collection" to DynamicCollectionAttribute so its
treated as a collection.

18 years ago- tightened down the screws on logging a little bit
Mike Bayer [Wed, 22 Aug 2007 18:08:10 +0000 (18:08 +0000)] 
- tightened down the screws on logging a little bit

18 years ago- added **modifiers to _get_from_objects
Mike Bayer [Wed, 22 Aug 2007 17:53:01 +0000 (17:53 +0000)] 
- added **modifiers to _get_from_objects
- fixed up PG distinct flag

18 years ago- restored engine.echo flag
Mike Bayer [Wed, 22 Aug 2007 16:58:26 +0000 (16:58 +0000)] 
- restored engine.echo flag
- changelog

18 years agoAdjusted ColumnDefault default function fitness check to only insure that a given...
Jason Kirtland [Wed, 22 Aug 2007 15:15:52 +0000 (15:15 +0000)] 
Adjusted ColumnDefault default function fitness check to only insure that a given function had no more than one non-defaulted positional arg.

18 years agoHousekeeping.
Jason Kirtland [Wed, 22 Aug 2007 08:33:09 +0000 (08:33 +0000)] 
Housekeeping.

18 years agoAdded a test for the SELECT DISTINCT ON postgresqlism.
Jason Kirtland [Wed, 22 Aug 2007 07:33:03 +0000 (07:33 +0000)] 
Added a test for the SELECT DISTINCT ON postgresqlism.
Test currently fails due to two problems in postgres.py, but I'm leaving
it uncorrected for now as its not clear what the original intent was
for lists.

18 years agoDeleting an entity having a dynamic loader with cascade="all" has some issues at...
Jason Kirtland [Wed, 22 Aug 2007 05:40:50 +0000 (05:40 +0000)] 
Deleting an entity having a dynamic loader with cascade="all" has some issues at the moment.

18 years agoAssociation example updates, round two.
Jason Kirtland [Wed, 22 Aug 2007 04:20:14 +0000 (04:20 +0000)] 
Association example updates, round two.

18 years agobasic 0.4 update
Jason Kirtland [Wed, 22 Aug 2007 03:25:09 +0000 (03:25 +0000)] 
basic 0.4 update

18 years agoFor sqlite NUMERIC, send Decimal bind values as strings instead of converting to...
Jason Kirtland [Wed, 22 Aug 2007 00:21:06 +0000 (00:21 +0000)] 
For sqlite NUMERIC, send Decimal bind values as strings instead of converting to floats.

18 years agofixed imports
Mike Bayer [Tue, 21 Aug 2007 23:42:34 +0000 (23:42 +0000)] 
fixed imports

18 years ago-removed echo_property() function, moved logging checks to
Mike Bayer [Tue, 21 Aug 2007 23:06:12 +0000 (23:06 +0000)] 
-removed echo_property() function, moved logging checks to
static variables

18 years ago`from foo import (name, name)` isn't valid syntax for 2.3. ah well.
Jason Kirtland [Tue, 21 Aug 2007 22:20:52 +0000 (22:20 +0000)] 
`from foo import (name, name)` isn't valid syntax for 2.3.  ah well.
omitting modules from sqlalchemy.__all__...

18 years agoUpdated adjencytree examples
Jason Kirtland [Tue, 21 Aug 2007 21:05:23 +0000 (21:05 +0000)] 
Updated adjencytree examples

18 years agofixed generative behavior to copy collections, [ticket:752]
Mike Bayer [Tue, 21 Aug 2007 20:43:54 +0000 (20:43 +0000)] 
fixed generative behavior to copy collections, [ticket:752]

18 years agoadded first profile tests per [ticket:753]
Mike Bayer [Tue, 21 Aug 2007 19:15:55 +0000 (19:15 +0000)] 
added first profile tests per [ticket:753]

18 years agochanging Pool to use weakref callback for auto-cleanup, instead of __del__.
Mike Bayer [Tue, 21 Aug 2007 18:18:21 +0000 (18:18 +0000)] 
changing Pool to use weakref callback for auto-cleanup, instead of __del__.
Still leaving the RLock in Queue however since I see no guarantee that the weakref callback
isn't called at an arbitrary time.

18 years agoInlined ClauseParameters.set_parameter (simple assignment) in construct_params
Jason Kirtland [Tue, 21 Aug 2007 16:58:49 +0000 (16:58 +0000)] 
Inlined ClauseParameters.set_parameter (simple assignment) in construct_params
Big drop in function count for inserts (22%) with about a 3% wall clock improvement.

18 years agofixed "SmallInteger"
Mike Bayer [Tue, 21 Aug 2007 16:56:22 +0000 (16:56 +0000)] 
fixed "SmallInteger"

18 years ago- implemented __len__() accessor on RowProxy
Mike Bayer [Tue, 21 Aug 2007 15:50:17 +0000 (15:50 +0000)] 
- implemented __len__() accessor on RowProxy
- implemented jek's ClauseParameters optimization for named params

18 years agotweak that construct_params optimization, one of the adjustments wasn't needed
Jason Kirtland [Tue, 21 Aug 2007 08:19:21 +0000 (08:19 +0000)] 
tweak that construct_params optimization, one of the adjustments wasn't needed

18 years agoA couple critical path optimizations
Jason Kirtland [Tue, 21 Aug 2007 07:55:43 +0000 (07:55 +0000)] 
A couple critical path optimizations
(some sql operations faster by nearly 10% wallclock, general orm around 3%)

18 years ago- omitted 'table' and 'column' from 'from sqlalchemy import *'
Jason Kirtland [Tue, 21 Aug 2007 01:31:23 +0000 (01:31 +0000)] 
- omitted 'table' and 'column' from 'from sqlalchemy import *'
- also omitted all modules and classes that aren't expicitly public
- omitted 'Smallinteger' (small i), but it's still in schema
- omitted NullType-related items from types.__all__
- patched up a few tests to use sql.table and sql.column, other related.

18 years agoadded stub/import tests for all dialects
Jason Kirtland [Tue, 21 Aug 2007 00:47:48 +0000 (00:47 +0000)] 
added stub/import tests for all dialects
post-refactor fix for access dialect

18 years agofix line ending svn prop
Jason Kirtland [Tue, 21 Aug 2007 00:43:47 +0000 (00:43 +0000)] 
fix line ending svn prop

18 years agopool_threadlocal on by default
Mike Bayer [Mon, 20 Aug 2007 23:34:58 +0000 (23:34 +0000)] 
pool_threadlocal on by default

18 years agoremoved unnecessary _branch calls
Mike Bayer [Mon, 20 Aug 2007 22:06:31 +0000 (22:06 +0000)] 
removed unnecessary _branch calls

18 years ago- method call removal
Mike Bayer [Mon, 20 Aug 2007 21:50:59 +0000 (21:50 +0000)] 
- method call removal

18 years ago- Engine and TLEngine assume "threadlocal" behavior on Pool; both use connect()
Mike Bayer [Mon, 20 Aug 2007 19:07:07 +0000 (19:07 +0000)] 
- Engine and TLEngine assume "threadlocal" behavior on Pool; both use connect()
for contextual connection, unique_connection() for non-contextual.
- Pool use_threadlocal defaults to True, can be set to false at create_engine()
level with pool_threadlocal=False
- made all logger statements in pool conditional based on a flag calcualted once.
- chagned WeakValueDictionary() used for "threadlocal" pool to be a regular dict
referencing weakref objects.  WVD had a lot of overhead, apparently.  *CAUTION* -
im pretty confident about this change, as the threadlocal dict gets explicitly managed
anyway, tests pass with PG etc., but keep a close eye on this one regardless.

18 years agoan early out processing insert/update column parameters was a bit too early.
Ants Aasma [Sun, 19 Aug 2007 23:01:44 +0000 (23:01 +0000)] 
an early out processing insert/update column parameters was a bit too early.

18 years agolight docstring tweaks to the pool
Jason Kirtland [Sun, 19 Aug 2007 18:34:24 +0000 (18:34 +0000)] 
light docstring tweaks to the pool
more pedantic DBAPI -> DB-API changes in docstrings

18 years agoCHANGES for r3372
Jason Kirtland [Sun, 19 Aug 2007 18:32:55 +0000 (18:32 +0000)] 
CHANGES for r3372

18 years ago- Connection.begin() no longer accepts nested=True, a possible source of confusion...
Jason Kirtland [Sun, 19 Aug 2007 17:57:51 +0000 (17:57 +0000)] 
- Connection.begin() no longer accepts nested=True, a possible source of confusion as two forms of nesting are supported.  SAVEPOINT-style nesting logic is now contained soley in begin_nested().
- Docstring love for the engine package.  More is needed.

18 years agoReST docstring fix
Jason Kirtland [Sun, 19 Aug 2007 17:48:46 +0000 (17:48 +0000)] 
ReST docstring fix

18 years agomssql unit test fixes
Paul Johnston [Sun, 19 Aug 2007 15:30:02 +0000 (15:30 +0000)] 
mssql unit test fixes

18 years agoOne more change of preparer() to identifier_preparer
Paul Johnston [Sun, 19 Aug 2007 14:51:36 +0000 (14:51 +0000)] 
One more change of preparer() to identifier_preparer

18 years agopyflakes mop-up
Jason Kirtland [Sun, 19 Aug 2007 14:48:04 +0000 (14:48 +0000)] 
pyflakes mop-up

18 years agoUse identifier_preparer instead of preparer()
Paul Johnston [Sun, 19 Aug 2007 14:26:53 +0000 (14:26 +0000)] 
Use identifier_preparer instead of preparer()

18 years agoFix broken imports
Paul Johnston [Sun, 19 Aug 2007 14:13:29 +0000 (14:13 +0000)] 
Fix broken imports

18 years agodocstring compile fixup
Mike Bayer [Sat, 18 Aug 2007 22:07:33 +0000 (22:07 +0000)] 
docstring compile fixup

18 years agoadding an "already exists" catch for CREATE DOMAIN
Mike Bayer [Sat, 18 Aug 2007 21:43:53 +0000 (21:43 +0000)] 
adding an "already exists" catch for CREATE DOMAIN

18 years ago1. Module layout. sql.py and related move into a package called "sql".
Mike Bayer [Sat, 18 Aug 2007 21:37:48 +0000 (21:37 +0000)] 
1. Module layout.  sql.py and related move into a package called "sql".
2. compiler names changed to be less verbose, unused classes removed.
3. Methods on Dialect which return compilers, schema generators, identifier preparers
have changed to direct class references, typically on the Dialect class itself
or optionally as attributes on an individual Dialect instance if conditional behavior is needed.
This takes away the need for Dialect subclasses to know how to instantiate these
objects, and also reduces method overhead by one call for each one.
4. as a result of 3., some internal signatures have changed for things like compiler() (now statement_compiler()), preparer(), etc., mostly in that the dialect needs to be passed explicitly as the first argument (since they are just class references now).  The compiler() method on Engine and Connection is now also named statement_compiler(), but as before does not take the dialect as an argument.

5. changed _process_row function on RowProxy to be a class reference, cuts out 50K method calls from insertspeed.py

18 years ago- fixed prefixes= argument to select()
Jason Kirtland [Sat, 18 Aug 2007 18:15:11 +0000 (18:15 +0000)] 
- fixed prefixes= argument to select()
- mysql can now generate DISTINCT or ALL for queries, select(..., distinct='ALL')
- documented 'prefixes' arg to select()
- rearranged doc order for select args to mirror that of a generated statement
- went nutty and fixed wrapping and line length on most docstrings in sql.py

18 years ago- modified SQL operator functions to be module-level operators, allowing
Mike Bayer [Sat, 18 Aug 2007 01:00:44 +0000 (01:00 +0000)] 
- modified SQL operator functions to be module-level operators, allowing
  SQL expressions to be pickleable [ticket:735]

- small adjustment to mapper class.__init__ to allow for Py2.6 object.__init__()
  behavior

18 years ago- Added a "legacy" adapter to types, such that user-defined TypeEngine
Mike Bayer [Fri, 17 Aug 2007 23:45:29 +0000 (23:45 +0000)] 
- Added a "legacy" adapter to types, such that user-defined TypeEngine
  and TypeDecorator classes which define convert_bind_param()/convert_result_value()
  will continue to function.  Also supports calling the super() version of
  those methods.

18 years agoadd a test for dupe tables in MetaData
Mike Bayer [Fri, 17 Aug 2007 22:20:55 +0000 (22:20 +0000)] 
add a test for dupe tables in MetaData

18 years ago- added close() method to Transaction. closes out a transaction using rollback
Mike Bayer [Fri, 17 Aug 2007 21:20:49 +0000 (21:20 +0000)] 
- added close() method to Transaction.  closes out a transaction using rollback
  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.

18 years agotransactional session rolls back bound connection
Mike Bayer [Fri, 17 Aug 2007 19:34:40 +0000 (19:34 +0000)] 
transactional session rolls back bound connection

18 years ago- threadlocal TLConnection, when closes for real, forces parent TLSession
Mike Bayer [Fri, 17 Aug 2007 19:13:51 +0000 (19:13 +0000)] 
- threadlocal TLConnection, when closes for real, forces parent TLSession
to rollback/dispose of transaction

18 years agoupdate activemapper backrefs for r3340
Jason Kirtland [Fri, 17 Aug 2007 19:13:12 +0000 (19:13 +0000)] 
update activemapper backrefs for r3340

18 years agodont commandeer warnings into logger
Mike Bayer [Fri, 17 Aug 2007 18:19:56 +0000 (18:19 +0000)] 
dont commandeer warnings into logger

18 years ago- turned twophase=True on in test
Mike Bayer [Fri, 17 Aug 2007 18:14:11 +0000 (18:14 +0000)] 
- turned twophase=True on in test
- TLEngine raises notimplemented for two-phase

18 years agoreinstated two_phase test. currently it passes on PG with and without threadlocal.
Mike Bayer [Fri, 17 Aug 2007 18:01:52 +0000 (18:01 +0000)] 
reinstated two_phase test.  currently it passes on PG with and without threadlocal.

18 years agomerge changset [3347] into trunk
Paul Johnston [Fri, 17 Aug 2007 18:00:30 +0000 (18:00 +0000)] 
merge changset [3347] into trunk