]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
14 years ago- more liberal check here
Mike Bayer [Fri, 8 Apr 2011 19:16:12 +0000 (15:16 -0400)] 
- more liberal check here
- dont encode enums in py3k

14 years ago- assume in py3k that description encoding is None unless the dialect really
Mike Bayer [Fri, 8 Apr 2011 18:54:45 +0000 (14:54 -0400)] 
- assume in py3k that description encoding is None unless the dialect really
overrides it
- psycopg2 + 3k supports unicode statements...

14 years ago- Some fixes to the state handling regarding
Mike Bayer [Thu, 7 Apr 2011 21:56:01 +0000 (17:56 -0400)] 
- Some fixes to the state handling regarding
backrefs, typically when autoflush=False, where
the back-referenced collection wouldn't
properly handle add/removes with no net
change.  Thanks to Richard Murri for the
test case + patch.  [ticket:2123]
(also in 0.6.7).

14 years agofix tests so that document related is in a separate fixture base
Mike Bayer [Thu, 7 Apr 2011 17:38:12 +0000 (13:38 -0400)] 
fix tests so that document related is in a separate fixture base

14 years ago- The limit/offset keywords to select() as well
Mike Bayer [Thu, 7 Apr 2011 17:34:38 +0000 (13:34 -0400)] 
- The limit/offset keywords to select() as well
as the value passed to select.limit()/offset()
will be coerced to integer.  [ticket:2116]
(also in 0.6.7)
- Oracle dialect adds use_binds_for_limits=False
create_engine() flag, will render the LIMIT/OFFSET
values inline instead of as binds, reported to
modify the execution plan used by Oracle.
[ticket:2116] (Also in 0.6.7)

14 years ago- Some fixes to "evaulate" and "fetch" evaluation
Mike Bayer [Thu, 7 Apr 2011 16:49:29 +0000 (12:49 -0400)] 
- Some fixes to "evaulate" and "fetch" evaluation
when query.update(), query.delete() are called.
The retrieval of records is done after autoflush
in all cases, and before update/delete is
emitted, guarding against unflushed data present
as well as expired objects failing during
the evaluation.  [ticket:2122]

14 years ago - Still more wording adjustments when a query option
Mike Bayer [Wed, 6 Apr 2011 14:36:33 +0000 (10:36 -0400)] 
  - Still more wording adjustments when a query option
    can't find the target entity.  Explain that the
    path must be from one of the root entities.

14 years ago- add link to concrete base declarative recipe
Mike Bayer [Tue, 5 Apr 2011 19:10:22 +0000 (15:10 -0400)] 
- add link to concrete base declarative recipe

14 years agochangelog
Mike Bayer [Tue, 5 Apr 2011 18:26:44 +0000 (14:26 -0400)] 
changelog

14 years ago- The "pool.manage" feature doesn't use pickle
Mike Bayer [Tue, 5 Apr 2011 18:26:28 +0000 (14:26 -0400)] 
- The "pool.manage" feature doesn't use pickle
anymore to hash the arguments for each pool.

14 years ago- REAL has been added to the core types. Supported
Mike Bayer [Tue, 5 Apr 2011 16:40:55 +0000 (12:40 -0400)] 
- REAL has been added to the core types.  Supported
by Postgresql, SQL Server, MySQL, SQLite.  Note
that the SQL Server and MySQL versions, which
add extra arguments, are also still available
from those dialects.  [ticket:2081]

14 years ago- The 'useexisting' flag on Table has been superceded
Mike Bayer [Tue, 5 Apr 2011 16:04:42 +0000 (12:04 -0400)] 
- The 'useexisting' flag on Table has been superceded
by a new pair of flags 'keep_existing' and
'extend_existing'.   'extend_existing' is equivalent
to 'useexisting' - the existing Table is returned,
and additional constructor elements are added.
With 'keep_existing', the existing Table is returned,
but additional constructor elements are not added -
these elements are only applied when the Table
is newly created.   [ticket:2109]

14 years ago- add some function examples, [ticket:2107]
Mike Bayer [Sat, 2 Apr 2011 20:32:05 +0000 (16:32 -0400)] 
- add some function examples, [ticket:2107]
- have "packagenames" be present on FunctionElement by default
so that compiler.visit_function() can be called
- add a test for that

14 years ago- Fixed mutable extension docs to show the
Mike Bayer [Sat, 2 Apr 2011 20:00:35 +0000 (16:00 -0400)] 
- Fixed mutable extension docs to show the
correct type-association methods.
[ticket:2118]

14 years ago- Added @event.listens_for() decorator, given
Mike Bayer [Sat, 2 Apr 2011 17:29:11 +0000 (13:29 -0400)] 
- Added @event.listens_for() decorator, given
target + event name, applies the decorated
function as a listener.  [ticket:2106]
- remove usage of globals from test.base.test_events

14 years agosome additional clarification on typeengine docs
Mike Bayer [Wed, 30 Mar 2011 18:22:41 +0000 (14:22 -0400)] 
some additional clarification on typeengine docs

14 years ago- beef up the "augmenting existing types" section
Mike Bayer [Wed, 30 Mar 2011 18:07:27 +0000 (14:07 -0400)] 
- beef up the "augmenting existing types" section
- add docstirngs for everything TypeDecorator [ticket:2086]

14 years ago- AssertionPool now stores the traceback indicating
Mike Bayer [Wed, 30 Mar 2011 16:30:54 +0000 (12:30 -0400)] 
- AssertionPool now stores the traceback indicating
where the currently checked out connection was
acquired; this traceback is reported within
the assertion raised upon a second concurrent
checkout; courtesy Gunnlaugur Briem
[ticket:2103]

14 years ago- and fix the message that was the actual subject of [ticket:2063]
Mike Bayer [Mon, 28 Mar 2011 01:14:06 +0000 (21:14 -0400)] 
- and fix the message that was the actual subject of [ticket:2063]

14 years ago- Reworded the exception raised when a flush
Mike Bayer [Mon, 28 Mar 2011 01:06:32 +0000 (21:06 -0400)] 
- Reworded the exception raised when a flush
is attempted of a subclass that is not polymorphic
against the supertype.  [ticket:2063]

14 years ago- add a note about snapshot isolation [ticket:2078]
Mike Bayer [Mon, 28 Mar 2011 00:50:47 +0000 (20:50 -0400)] 
- add a note about snapshot isolation [ticket:2078]

14 years ago- Documented SQLite DATE/TIME/DATETIME types.
Mike Bayer [Mon, 28 Mar 2011 00:34:39 +0000 (20:34 -0400)] 
- Documented SQLite DATE/TIME/DATETIME types.
[ticket:2029] (also in 0.6.7)
- add "currentmodule" directive to all the dialect type docs to
ensure users import from the dialect package, not the "base" module

14 years ago- Using column names that would require quotes
Mike Bayer [Sun, 27 Mar 2011 23:39:39 +0000 (19:39 -0400)] 
- Using column names that would require quotes
for the column itself or for a name-generated
bind parameter, such as names with special
characters, underscores, non-ascii characters,
now properly translate bind parameter keys when
talking to cx_oracle.  [ticket:2100]  (Also
in 0.6.7)

14 years ago- do a close() here so that objects are guaranteed loaded new
Mike Bayer [Sun, 27 Mar 2011 23:26:47 +0000 (19:26 -0400)] 
- do a close() here so that objects are guaranteed loaded new

14 years ago- fix a callcount
Mike Bayer [Sun, 27 Mar 2011 22:16:40 +0000 (18:16 -0400)] 
- fix a callcount
- remove extra _load_fixtures() method.   using the classbound version as the fixtures
seem to have everything classlevel at the moment (wondering if we want to change that
for non-classlevel setup/teardown)

14 years ago- reorganizing the test modules and fixtures, but keeping the essential method of
Mike Bayer [Sun, 27 Mar 2011 21:52:30 +0000 (17:52 -0400)] 
- reorganizing the test modules and fixtures, but keeping the essential method of
creating fixture classes the same, i.e. run a method, then inspect __subclasses__() of
a marked base class, seems to have broken that approach permanently.   While the tests run
locally just fine, on the buildbot most of the environments continue to fail in entirely
random places.   One of the key changes of the reorganization is that the big bunch of
classes from _fixtures.py is now created ad-hoc instead of reusing a set of module-bound
classes.  This means there's a lot more class creation/teardown traffic occuring in the
cPython interpreter.   Taking a look at the docs for __subclasses__() confirms that the
list of subclasses is of course a weak-referenced collection.   So in the absense of the
tests working just fine for a couple of years, it would seem obvious that the bug
here is setup_classes()/mappers() is called, we come back, the new classes are now
not referenced at all, we then call __subclasses__() and maybe we get them, maybe we don't.
But it's totally not clear how this behavior just hasn't occurred at all for a couple
of years (except, possibly, for very rare occasions that I cannot reproduce).   So in this changeset we switch
to a method of class tracking that I use in a lot of other places, that of using a metaclass
to positively stick classes in the registry the moment they are created.   Test classes
are already built against a base that is created per-fixture run so we already have the advantage
of a local system.   I send this changeset to the buildbot and beg of it to please allow our
tests to pass once again !

14 years ago- dont clear the classes collection here either...
Mike Bayer [Sun, 27 Mar 2011 21:28:41 +0000 (17:28 -0400)] 
- dont clear the classes collection here either...

14 years ago- move all the comments that got shoved below the fixture grabs back up
Mike Bayer [Sun, 27 Mar 2011 21:19:00 +0000 (17:19 -0400)] 
- move all the comments that got shoved below the fixture grabs back up

14 years ago- remove test.sql._base, test.engine._base, test.orm._base, move those classes to...
Mike Bayer [Sun, 27 Mar 2011 20:27:27 +0000 (16:27 -0400)] 
- remove test.sql._base, test.engine._base, test.orm._base, move those classes to a new test.lib.fixtures module
- move testing.TestBase to test.lib.fixtures
- massive search and replace

14 years ago- move all the setup_classes(cls) and setup_mappers(cls) to use a
Mike Bayer [Sun, 27 Mar 2011 19:04:06 +0000 (15:04 -0400)] 
- move all the setup_classes(cls) and setup_mappers(cls) to use a
local cls.Basic, cls.Comparable base class so that there is no ambiguity
or hash identity behaviors getting in the way of class registration.

14 years agofixes for savepoint test
Mike Bayer [Sun, 27 Mar 2011 01:52:31 +0000 (21:52 -0400)] 
fixes for savepoint test

14 years ago- convert @provide_metadata to use self access
Mike Bayer [Sun, 27 Mar 2011 01:40:05 +0000 (21:40 -0400)] 
- convert @provide_metadata to use self access
- having occasional issues with BasicEntity grabbing, if it persists
may have to pass an explicit base class into setup_classes()/setup_mappers()

14 years ago- move _fixtures to work via the normal methods of _base.MappedTest, convert
Mike Bayer [Sun, 27 Mar 2011 00:44:47 +0000 (20:44 -0400)] 
- move _fixtures to work via the normal methods of _base.MappedTest, convert
all referncing tests to not use globals
- tests that deal with pickle specifically load the fixture classes
from test.lib.pickleable, which gets some more classes added
- removed weird sa05 pickling tests that don't matter

14 years ago- move _fixtures into the traditional FixtureTest form. will adjust
Mike Bayer [Sat, 26 Mar 2011 23:30:40 +0000 (19:30 -0400)] 
- move _fixtures into the traditional FixtureTest form.  will adjust
old style tests

14 years ago- remove @testing.resolve_artifact_names, replace with direct attribute
Mike Bayer [Sat, 26 Mar 2011 23:03:11 +0000 (19:03 -0400)] 
- remove @testing.resolve_artifact_names, replace with direct attribute
access to the cls/self.tables/classes registries
- express orm/_base.py ORMTest in terms of engine/_base.py TablesTest,
factor out common steps into TablesTest, remove AltEngineTest as a
separate class.   will further consolidate these base classes

14 years agoindent the string here
Mike Bayer [Sat, 26 Mar 2011 19:31:15 +0000 (15:31 -0400)] 
indent the string here

14 years ago- changes to CHANGES
Mike Bayer [Sat, 26 Mar 2011 15:53:07 +0000 (11:53 -0400)] 
- changes to CHANGES

- The "-declarative" changes will now be listed
directly under the "-orm" section, as these
are closely related.

- The 0.5 series changes have been moved to
the file CHANGES_PRE_06 which replaces
CHANGES_PRE_05.

- The changelog for 0.6.7 and subsequent within
the 0.6 series is now listed only in the
CHANGES file within the 0.6 branch.
In the 0.7 CHANGES file all the
0.6 changes are listed inline within the 0.7
section in which they were also applied
(since all 0.6 changes are in 0.7 as well).
Changes that apply to an 0.6 version here
are noted as are if any differences in
implementation/behavior are present.

14 years ago- adjust the previous checkin a bit. need to transfer sequences over from pickle
Mike Bayer [Wed, 23 Mar 2011 21:38:54 +0000 (17:38 -0400)] 
- adjust the previous checkin a bit.  need to transfer sequences over from pickle
even if they are *not* restated.  its necessary here to do a "replacement"
scheme when an incoming sequence replaces the previous.  Theoretically
we could do what Table does here, i.e. use a singleton constructor, but
this is heavyhanded.   The most recent sequence placed in is the winner
is likely the most expected behavior.

14 years ago- Fixed regression whereby MetaData() coming
Mike Bayer [Wed, 23 Mar 2011 21:25:57 +0000 (17:25 -0400)] 
- Fixed regression whereby MetaData() coming
back from unpickling did not keep track of
new things it keeps track of now, i.e.
collection of Sequence objects, list
of schema names.  [ticket:2104]

14 years ago- Fixed bug in QueuePool, SingletonThreadPool whereby
Mike Bayer [Wed, 23 Mar 2011 03:39:05 +0000 (23:39 -0400)] 
- Fixed bug in QueuePool, SingletonThreadPool whereby
connections that were discarded via overflow or periodic
cleanup() were not explicitly closed, leaving garbage
collection to the task instead.   This generally only
affects non-reference-counting backends like Jython
and Pypy.  Thanks to Jaimy Azle for spotting
this.  [ticket:2102]

14 years ago- Restored the "catchall" constructor on the base
Mike Bayer [Mon, 21 Mar 2011 15:49:43 +0000 (11:49 -0400)] 
- Restored the "catchall" constructor on the base
TypeEngine class, with a deprecation warning.
This so that code which does something like
Integer(11) still succeeds.

14 years agoAdded tag rel_0_7b3 for changeset b6423d26280cac2ff74ab1450ce5549789ffa61e
Mike Bayer [Sun, 20 Mar 2011 19:14:15 +0000 (15:14 -0400)] 
Added tag rel_0_7b3 for changeset b6423d26280cac2ff74ab1450ce5549789ffa61e

14 years ago- remove all the fanfare from dynamic_loader() and de-emphasize it in rel_0_7b3
Mike Bayer [Sun, 20 Mar 2011 18:54:35 +0000 (14:54 -0400)] 
- remove all the fanfare from dynamic_loader() and de-emphasize it in
the docs for [ticket:2080].   Not deprecating for now.

14 years ago- add attrstrings for all the important mapper() attrbutes, [ticket:2094].
Mike Bayer [Sun, 20 Mar 2011 18:00:16 +0000 (14:00 -0400)] 
- add attrstrings for all the important mapper() attrbutes, [ticket:2094].
Sticking with quotes here since that's what's used everywhere else.
- rename "API Documentation - XYZ" to "XYZ API".
- move connection API docs to the bottom of that section.  still
floating between separate API sections vs. inline API sections, not
sure how to be consistent.
- fix outdated paragraph about derived_attrbutes
- other target function fixes

14 years ago- Added new generic function "next_value()", accepts
Mike Bayer [Sun, 20 Mar 2011 16:49:28 +0000 (12:49 -0400)] 
- Added new generic function "next_value()", accepts
a Sequence object as its argument and renders the
appropriate "next value" generation string on the
target platform, if supported.  Also provides
".next_value()" method on Sequence itself.
[ticket:2085]
- added tests for all the conditions described
in [ticket:2085]
- postgresql dialect will exec/compile a Sequence
that has "optional=True".  the optional flag is now only
checked specifically in the context of a Table primary key
evaulation.
- func.next_value() or other SQL expression can
be embedded directly into an insert() construct,
and if implicit or explicit "returning" is used
in conjunction with a primary key column,
the newly generated value will be present in
result.inserted_primary_key. [ticket:2084]

14 years agoAdded docs requirements
Michael Trier [Sun, 20 Mar 2011 04:05:22 +0000 (00:05 -0400)] 
Added docs requirements

14 years agowhoops, only on SQL server
Mike Bayer [Fri, 18 Mar 2011 16:30:06 +0000 (12:30 -0400)] 
whoops, only on SQL server

14 years ago- some doc reorg
Mike Bayer [Fri, 18 Mar 2011 01:13:14 +0000 (21:13 -0400)] 
- some doc reorg
- change engine.Connection to _connection_cls so sphinx doesn't get upset
- globally add "." to all :class:`Foo`
- start naming sections that are mostly docstrings "API Documentation - blah blah"
- move some ad-hoc docstrings into "API" sections, there is some inconsistency here
and it may be that we just have to leave it that way
- add "internals" rsts to core, orm, I'm not super thrilled how these look but they are
targeted by some of the public api docs, users typically become aware of these anyway

14 years agochangelog update with 0.6 message
Mike Bayer [Thu, 17 Mar 2011 20:59:51 +0000 (16:59 -0400)] 
changelog update with 0.6 message

14 years ago- Fixed bug in query.options() whereby a path
Mike Bayer [Thu, 17 Mar 2011 20:22:25 +0000 (16:22 -0400)] 
- Fixed bug in query.options() whereby a path
applied to a lazyload using string keys could
overlap a same named attribute on the wrong
entity.  Note 0.6 has a more conservative fix
to this.  [ticket:2098]

14 years ago- merge
Mike Bayer [Wed, 16 Mar 2011 22:16:26 +0000 (18:16 -0400)] 
- merge

14 years ago- Fixed AssertionPool regression bug. A replacement of finalize
Mike Bayer [Wed, 16 Mar 2011 22:15:27 +0000 (18:15 -0400)] 
- Fixed AssertionPool regression bug.  A replacement of finalize
logic did the wrong thing, but then also the logic it replaced also
doesn't seem like its needed.  If it is, would rather have a test case
first so its out for now, added tests for assertionpool.  [ticket:2097]

14 years agomerged upstream
Alex Gaynor [Wed, 16 Mar 2011 18:21:56 +0000 (14:21 -0400)] 
merged upstream

14 years agothis test fails on PyPy because it checks for a lastrowid after the connection is...
Alex Gaynor [Wed, 16 Mar 2011 18:18:27 +0000 (14:18 -0400)] 
this test fails on PyPy because it checks for a lastrowid after the connection is lcosed

14 years ago- Rewrote the query used to get the definition of a view,
Mike Bayer [Wed, 16 Mar 2011 18:06:33 +0000 (14:06 -0400)] 
- Rewrote the query used to get the definition of a view,
typically when using the Inspector interface, to
use sys.sql_modules instead of the information schema,
thereby allowing views definitions longer than 4000
characters to be fully returned.  [ticket:2071]

14 years agomerged upstream
Alex Gaynor [Wed, 16 Mar 2011 17:59:23 +0000 (13:59 -0400)] 
merged upstream

14 years ago- alex gaynor's latest batch of pypy test fixes
Mike Bayer [Wed, 16 Mar 2011 17:04:07 +0000 (13:04 -0400)] 
- alex gaynor's latest batch of pypy test fixes

14 years agopypy fixes.
Alex Gaynor [Wed, 16 Mar 2011 16:56:48 +0000 (12:56 -0400)] 
pypy fixes.

14 years ago- The horizontal_shard ShardedSession class accepts the common
Mike Bayer [Wed, 16 Mar 2011 16:43:22 +0000 (12:43 -0400)] 
- The horizontal_shard ShardedSession class accepts the common
Session argument "query_cls" as a constructor argument,
to enable further subclassing of ShardedQuery.
[ticket:2090]
- The Beaker caching example allows a "query_cls" argument
to the query_callable() function.  [ticket:2090]

14 years ago- Improvements to the error messages emitted when
Mike Bayer [Wed, 16 Mar 2011 16:30:13 +0000 (12:30 -0400)] 
- Improvements to the error messages emitted when
querying against column-only entities in conjunction
with (typically incorrectly) using loader options,
where the parent entity is not fully present.
[ticket:2069]

14 years ago- Firebird - the "implicit_returning" flag on create_engine() is
Mike Bayer [Wed, 16 Mar 2011 15:22:28 +0000 (11:22 -0400)] 
- Firebird - the "implicit_returning" flag on create_engine() is
honored if set to False.  [ticket:2083]

14 years ago- Fixed the BIT type to allow a "length" parameter, "varying"
Mike Bayer [Wed, 16 Mar 2011 15:01:20 +0000 (11:01 -0400)] 
- Fixed the BIT type to allow a "length" parameter, "varying"
parameter.  Reflection also fixed.  [ticket:2073]

14 years agodouble quote these
Mike Bayer [Wed, 16 Mar 2011 14:43:42 +0000 (10:43 -0400)] 
double quote these

14 years ago- Added accessors to ResultProxy "returns_rows", "is_insert"
Mike Bayer [Wed, 16 Mar 2011 14:36:48 +0000 (10:36 -0400)] 
- Added accessors to ResultProxy "returns_rows", "is_insert"
[ticket:2089]

14 years agoblow away some globals in the search for what makes resolve_artifact_names fail ...
Mike Bayer [Wed, 16 Mar 2011 05:26:27 +0000 (01:26 -0400)] 
blow away some globals in the search for what makes resolve_artifact_names fail (possibly when there's already globals in the way...)

14 years agoFix 3 errors in py32, from patch from lsblakk. There are still 8 failures. See #2088.
Taavi Burns [Wed, 16 Mar 2011 04:32:33 +0000 (00:32 -0400)] 
Fix 3 errors in py32, from patch from lsblakk. There are still 8 failures. See #2088.

14 years ago- adjust gc_collect for pypy
Mike Bayer [Wed, 16 Mar 2011 02:11:34 +0000 (22:11 -0400)] 
- adjust gc_collect for pypy

14 years ago- qualify the use of locals() based on python version in use, don't run for pypy...
Mike Bayer [Wed, 16 Mar 2011 01:36:34 +0000 (21:36 -0400)] 
- qualify the use of locals() based on python version in use, don't run for pypy, jython, should fix [ticket:1073]

14 years agochange TypeError test to not care about string message
Mike Bayer [Wed, 16 Mar 2011 01:26:59 +0000 (21:26 -0400)] 
change TypeError test to not care about string message

14 years ago- merge
Mike Bayer [Tue, 15 Mar 2011 22:32:53 +0000 (18:32 -0400)] 
- merge

14 years agomerge
Taavi Burns [Tue, 15 Mar 2011 22:22:18 +0000 (18:22 -0400)] 
merge

14 years agoMerge
Michael Trier [Tue, 15 Mar 2011 22:01:00 +0000 (18:01 -0400)] 
Merge

14 years agoMerge
Michael Trier [Tue, 15 Mar 2011 22:00:17 +0000 (18:00 -0400)] 
Merge

14 years agoMerge
Denise Govindarajan [Tue, 15 Mar 2011 21:55:01 +0000 (17:55 -0400)] 
Merge

14 years agoPulling 32d2b5d1962961b4134463becdd988888f797e90 onto default from rel_0_6
Taavi Burns [Tue, 15 Mar 2011 21:54:11 +0000 (17:54 -0400)] 
Pulling 32d2b5d1962961b4134463becdd988888f797e90 onto default from rel_0_6

14 years agopsycopg2 2.4 no longer accepts % in bind place holders
Denise Govindarajan [Tue, 15 Mar 2011 21:47:36 +0000 (17:47 -0400)] 
psycopg2 2.4 no longer accepts % in bind place holders

14 years agoAdded required import (somehow we never seem to hit the unimported ref)
Taavi Burns [Tue, 15 Mar 2011 21:45:55 +0000 (17:45 -0400)] 
Added required import (somehow we never seem to hit the unimported ref)

14 years ago- add tests, CHANGES for [ticket:2095]
Mike Bayer [Tue, 15 Mar 2011 21:31:19 +0000 (17:31 -0400)] 
- add tests, CHANGES for [ticket:2095]

14 years agoInformix and postgresql reserved words changes -- almost done.
Jim Phares [Tue, 15 Mar 2011 20:56:45 +0000 (16:56 -0400)] 
Informix and postgresql reserved words changes -- almost done.

14 years agoRemoving unused symbol import
Taavi Burns [Tue, 15 Mar 2011 20:01:58 +0000 (16:01 -0400)] 
Removing unused symbol import

14 years agoinitial work on column reflect
Mike Bayer [Mon, 14 Mar 2011 20:36:08 +0000 (16:36 -0400)] 
initial work on column reflect

14 years agomerge
Mike Bayer [Mon, 14 Mar 2011 19:08:47 +0000 (15:08 -0400)] 
merge

14 years ago- Arguments in __mapper_args__ that aren't "hashable"
Mike Bayer [Mon, 14 Mar 2011 19:01:17 +0000 (15:01 -0400)] 
- Arguments in __mapper_args__ that aren't "hashable"
aren't mistaken for always-hashable, possibly-column
arguments.  [ticket:2091]

14 years agoTests for unicode columns with non-unicode defaults. See #2079.
Taavi Burns [Mon, 14 Mar 2011 18:42:11 +0000 (14:42 -0400)] 
Tests for unicode columns with non-unicode defaults.  See #2079.

14 years agoWarn about unicode column having non-unicode default. See #2079.
Diana Clarke [Mon, 14 Mar 2011 18:34:44 +0000 (14:34 -0400)] 
Warn about unicode column having non-unicode default.  See #2079.

14 years agoMerge
Michael Trier [Mon, 14 Mar 2011 16:04:36 +0000 (12:04 -0400)] 
Merge

14 years agoFixed order_by in test.orm.test_froms.MixedEntitiesTest.test_values.
Jim Phares [Mon, 14 Mar 2011 15:51:38 +0000 (11:51 -0400)] 
Fixed order_by in test.orm.test_froms.MixedEntitiesTest.test_values.

14 years agoCorrected order by in test joins.
Denise Govindarajan [Mon, 14 Mar 2011 15:44:38 +0000 (11:44 -0400)] 
Corrected order by in test joins.

14 years agofix "associated" at the source too
Mike Bayer [Mon, 14 Mar 2011 15:20:01 +0000 (11:20 -0400)] 
fix "associated" at the source too

14 years agoFixed doc problem. Thanks Toby Ho.
Michael Trier [Mon, 14 Mar 2011 04:43:08 +0000 (00:43 -0400)] 
Fixed doc problem. Thanks Toby Ho.

14 years agoAssociated was spelled wrong.
Michael Trier [Mon, 14 Mar 2011 04:20:34 +0000 (00:20 -0400)] 
Associated was spelled wrong.

14 years ago- Updated the association, association proxy examples
Mike Bayer [Sun, 13 Mar 2011 17:59:00 +0000 (13:59 -0400)] 
- Updated the association, association proxy examples
to use declarative, added a new example
dict_of_sets_with_default.py, a "pushing the envelope"
example of association proxy.

14 years ago- Changed the underlying approach to query.count().
Mike Bayer [Thu, 10 Mar 2011 00:17:45 +0000 (19:17 -0500)] 
- Changed the underlying approach to query.count().
query.count() is now in all cases exactly:

query.
    from_self(func.count(literal_column('1'))).
    scalar()

That is, "select count(1) from (<full query>)".
This produces a subquery in all cases, but
vastly simplifies all the guessing count()
tried to do previously, which would still
fail in many scenarios particularly when
joined table inheritance and other joins
were involved.  If the subquery produced
for an otherwise very simple count is really
an issue, use query(func.count()) as an
optimization.  [ticket:2093]

14 years agoModifications for PyPy support from Alex Gaynor.
Michael Trier [Wed, 9 Mar 2011 04:40:59 +0000 (23:40 -0500)] 
Modifications for PyPy support from Alex Gaynor.

14 years ago- A tweak to the unit of work causes it to order
Mike Bayer [Sun, 6 Mar 2011 20:37:18 +0000 (15:37 -0500)] 
- A tweak to the unit of work causes it to order
the flush along relationship() dependencies even if
the given objects don't have any inter-attribute
references in memory, which was the behavior in
0.5 and earlier, so a flush of Parent/Child with
only foreign key/primary key set will succeed.
This while still maintaining 0.6 and above's not
generating a ton of useless internal dependency
structures within the flush that don't correspond
to state actually within the current flush.
[ticket:2082]

14 years ago0.6.7 changelog message
Mike Bayer [Sat, 5 Mar 2011 17:28:16 +0000 (12:28 -0500)] 
0.6.7 changelog message

14 years ago - some changes to the identity map regarding
Mike Bayer [Sat, 5 Mar 2011 01:52:22 +0000 (20:52 -0500)] 
  - some changes to the identity map regarding
    rare weakref callbacks during iterations.
    The mutex has been removed as it apparently
    can cause a reentrant (i.e. in one thread) deadlock,
    perhaps when gc collects objects at the point of
    iteration in order to gain more memory.  It is hoped
    that "dictionary changed during iteration" will
    be exceedingly rare as iteration methods internally
    acquire the full list of objects in a single values()
    call. [ticket:2087]

14 years agoCorrected a bunch of spelling typos
Diana Clarke [Mon, 28 Feb 2011 18:21:08 +0000 (13:21 -0500)] 
Corrected a bunch of spelling typos

14 years agocorrected a bunch of spelling typos
Diana Clarke [Mon, 28 Feb 2011 18:17:46 +0000 (13:17 -0500)] 
corrected a bunch of spelling typos

14 years agoAdded changes information for #2060.
Michael Trier [Mon, 14 Mar 2011 16:46:21 +0000 (12:46 -0400)] 
Added changes information for #2060.