]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
11 years ago- reinstate E123, E125, E128
Mike Bayer [Wed, 16 Jul 2014 18:30:35 +0000 (14:30 -0400)] 
- reinstate E123, E125, E128
- edit strategies.py to conform

11 years ago- forwards-port changelog for ref #3093
Mike Bayer [Wed, 16 Jul 2014 17:43:57 +0000 (13:43 -0400)] 
- forwards-port changelog for ref #3093

11 years ago- use an odict here to ensure ordering of properties
Mike Bayer [Tue, 15 Jul 2014 22:59:44 +0000 (18:59 -0400)] 
- use an odict here to ensure ordering of properties

11 years ago- add a full env for pypy, name it funny to avoid hidden/undocumented
Mike Bayer [Tue, 15 Jul 2014 19:55:12 +0000 (15:55 -0400)] 
- add a full env for pypy, name it funny to avoid hidden/undocumented
tox behavior

11 years ago- fully flake8 this file
Mike Bayer [Tue, 15 Jul 2014 17:34:55 +0000 (13:34 -0400)] 
- fully flake8 this file

11 years ago- Fixed a regression caused by :ticket:`2976` released in 0.9.4 where
Mike Bayer [Tue, 15 Jul 2014 17:20:55 +0000 (13:20 -0400)] 
- Fixed a regression caused by :ticket:`2976` released in 0.9.4 where
the "outer join" propagation along a chain of joined eager loads
would incorrectly convert an "inner join" along a sibling join path
into an outer join as well, when only descendant paths should be
receiving the "outer join" propagation; additionally, fixed related
issue where "nested" join propagation would take place inappropriately
between two sibling join paths.

this is accomplished by re-introducing the removed flag "allow_innerjoin",
now inverted and named "chained_from_outerjoin".  Propagating this flag
allows us to know when we have encountered an outerjoin along a load
path, without confusing it for state obtained from a sibling path.

fixes #3131
ref #2976

11 years ago- Fixed a SQLite join rewriting issue where a subquery that is embedded
Mike Bayer [Tue, 15 Jul 2014 16:25:38 +0000 (12:25 -0400)] 
- Fixed a SQLite join rewriting issue where a subquery that is embedded
as a scalar subquery such as within an IN would receive inappropriate
substitutions from the enclosing query, if the same table were present
inside the subquery as were in the enclosing query such as in a
joined inheritance scenario.
fixes #3130

11 years ago- wrestle with conv() and tests some more
Mike Bayer [Tue, 15 Jul 2014 02:44:33 +0000 (22:44 -0400)] 
- wrestle with conv() and tests some more

11 years ago- allow the compilation rule that gets the formatted name
Mike Bayer [Tue, 15 Jul 2014 01:11:16 +0000 (21:11 -0400)] 
- allow the compilation rule that gets the formatted name
to again have the chance to veto rendering, as the naming convention
can make the decision that the name is "none" or not now.

11 years ago- Fix bug in naming convention feature where using a check
Mike Bayer [Tue, 15 Jul 2014 00:26:38 +0000 (20:26 -0400)] 
- Fix bug in naming convention feature where using a check
constraint convention that includes ``constraint_name`` would
then force all :class:`.Boolean` and :class:`.Enum` types to
require names as well, as these implicitly create a
constraint, even if the ultimate target backend were one that does
not require generation of the constraint such as Postgresql.
The mechanics of naming conventions for these particular
constraints has been reorganized such that the naming
determination is done at DDL compile time, rather than at
constraint/table construction time.
fixes #3067

11 years ago- Fixed a regression from 0.9.5 caused by :ticket:`3025` where the
Mike Bayer [Mon, 14 Jul 2014 23:16:11 +0000 (19:16 -0400)] 
- Fixed a regression from 0.9.5 caused by :ticket:`3025` where the
query used to determine "default schema" is invalid in SQL Server 2000.
For SQL Server 2000 we go back to defaulting to the "schema name"
parameter of the dialect, which is configurable but defaults
to 'dbo'. fixes #3025

11 years ago- Added statement encoding to the "SET IDENTITY_INSERT"
Mike Bayer [Mon, 14 Jul 2014 23:02:20 +0000 (19:02 -0400)] 
- Added statement encoding to the "SET IDENTITY_INSERT"
statements which operate when an explicit INSERT is being
interjected into an IDENTITY column, to support non-ascii table
identifiers on drivers such as pyodbc + unix + py2k that don't
support unicode statements.
ref #3091 as this fix is also in that issue's patch, but is
a different issue.

11 years ago- In the SQL Server pyodbc dialect, repaired the implementation
Mike Bayer [Mon, 14 Jul 2014 22:54:23 +0000 (18:54 -0400)] 
- In the SQL Server pyodbc dialect, repaired the implementation
for the ``description_encoding`` dialect parameter, which when
not explicitly set was preventing  cursor.description from
being parsed correctly in the case of result sets that
contained names in alternate encodings.  This parameter
shouldn't be needed going forward.
fixes #3091

11 years ago- Fixed a regression from 0.9.0 due to :ticket:`2736` where the
Mike Bayer [Mon, 14 Jul 2014 22:49:06 +0000 (18:49 -0400)] 
- Fixed a regression from 0.9.0 due to :ticket:`2736` where the
:meth:`.Query.select_from` method no longer set up the "from
entity" of the :class:`.Query` object correctly, so that
subsequent :meth:`.Query.filter_by` or :meth:`.Query.join`
calls would fail to check the appropriate "from" entity when
searching for attributes by string name.
fixes #3083

11 years ago -Fixed bug in common table expressions whereby positional bound
Mike Bayer [Mon, 14 Jul 2014 22:34:32 +0000 (18:34 -0400)] 
 -Fixed bug in common table expressions whereby positional bound
parameters could be expressed in the wrong final order
when CTEs were nested in certain ways.
fixes #3090

11 years ago- Fixed bug where multi-valued :class:`.Insert` construct would fail
Mike Bayer [Mon, 14 Jul 2014 22:28:07 +0000 (18:28 -0400)] 
- Fixed bug where multi-valued :class:`.Insert` construct would fail
to check subsequent values entries beyond the first one given
for literal SQL expressions.
fixes #3069

11 years agoneed a 1.0.0 here
Mike Bayer [Mon, 14 Jul 2014 22:10:34 +0000 (18:10 -0400)] 
need a 1.0.0 here

11 years ago- Added a "str()" step to the dialect_kwargs iteration for
Mike Bayer [Mon, 14 Jul 2014 22:09:54 +0000 (18:09 -0400)] 
- Added a "str()" step to the dialect_kwargs iteration for
Python version < 2.6.5, working around the
"no unicode keyword arg" bug as these args are passed along as
keyword args within some reflection processes.
fixes #3123

11 years ago- add some order bys
Mike Bayer [Mon, 14 Jul 2014 20:31:16 +0000 (16:31 -0400)] 
- add some order bys

11 years ago- ensure usedevelop is present so that C exts are in the tests
Mike Bayer [Mon, 14 Jul 2014 15:33:27 +0000 (11:33 -0400)] 
- ensure usedevelop is present so that C exts are in the tests

11 years ago- adjust this to work on PG also
Mike Bayer [Mon, 14 Jul 2014 03:59:12 +0000 (23:59 -0400)] 
- adjust this to work on PG also

11 years agoMerge branch 'pep8'
Mike Bayer [Mon, 14 Jul 2014 00:19:25 +0000 (20:19 -0400)] 
Merge branch 'pep8'

11 years agoPEP8 style fixes
Brian Jarrett [Thu, 10 Jul 2014 22:28:49 +0000 (16:28 -0600)] 
PEP8 style fixes

11 years ago- Fixed bug in :class:`.Enum` and other :class:`.SchemaType`
Mike Bayer [Sun, 13 Jul 2014 22:55:18 +0000 (18:55 -0400)] 
- Fixed bug in :class:`.Enum` and other :class:`.SchemaType`
subclasses where direct association of the type with a
:class:`.MetaData` would lead to a hang when events
(like create events) were emitted on the :class:`.MetaData`.
fixes #3124

11 years agoupdate test runners to be module-based
Mike Bayer [Fri, 11 Jul 2014 00:41:48 +0000 (20:41 -0400)] 
update test runners to be module-based

11 years agoOK use tox
Mike Bayer [Fri, 11 Jul 2014 00:06:47 +0000 (20:06 -0400)] 
OK use tox

11 years agoMerge branch 'master' of https://github.com/celttechie/sqlalchemy
Brian Jarrett [Thu, 10 Jul 2014 22:31:07 +0000 (16:31 -0600)] 
Merge branch 'master' of https://github.com/celttechie/sqlalchemy

Pulling from new fork of SQLAlchemy

11 years agofixed style to conform to PEP8
Brian Jarrett [Thu, 10 Jul 2014 22:28:49 +0000 (16:28 -0600)] 
fixed style to conform to PEP8

11 years ago- determine the root cause of the mysqlconnector issue, report
Mike Bayer [Thu, 10 Jul 2014 20:10:46 +0000 (16:10 -0400)] 
- determine the root cause of the mysqlconnector issue, report
it and move on

11 years ago- mark tests failing for mysqlconnector, oursql
Mike Bayer [Thu, 10 Jul 2014 15:47:31 +0000 (11:47 -0400)] 
- mark tests failing for mysqlconnector, oursql

11 years agoimports gone bad
Mike Bayer [Thu, 10 Jul 2014 15:36:57 +0000 (11:36 -0400)] 
imports gone bad

11 years ago- The :meth:`.TypeEngine.with_variant` method will now accept a
Mike Bayer [Thu, 10 Jul 2014 15:26:13 +0000 (11:26 -0400)] 
- The :meth:`.TypeEngine.with_variant` method will now accept a
type class as an argument which is internally converted to an
instance, using the same convention long established by other
constructs such as :class:`.Column`. fixes #3122

11 years ago- fully flake8 test/aaa_profiling
Mike Bayer [Wed, 9 Jul 2014 22:59:16 +0000 (18:59 -0400)] 
- fully flake8 test/aaa_profiling

11 years ago- break up the <authors> copyright comment as part of a pass
Mike Bayer [Wed, 9 Jul 2014 22:26:55 +0000 (18:26 -0400)] 
- break up the <authors> copyright comment as part of a pass
to get all flake8 passing

11 years ago- support __only_on__ and __backend__ at the same time
Mike Bayer [Wed, 9 Jul 2014 22:12:32 +0000 (18:12 -0400)] 
- support __only_on__ and __backend__ at the same time

11 years ago- Changed the default value of "raise_on_warnings" to False for
Mike Bayer [Wed, 9 Jul 2014 20:04:07 +0000 (16:04 -0400)] 
- Changed the default value of "raise_on_warnings" to False for
MySQLconnector.  This was set at True for some reason.  The "buffered"
flag unfortunately must stay at True as MySQLconnector does not allow
a cursor to be closed unless all results are fully fetched. fixes #2515
- lots of MySQL tests seemed to not be hitting all backends, so we should
be getting some mysqlconnector failures now

11 years ago- add 0.9 changelog
Mike Bayer [Wed, 9 Jul 2014 19:51:47 +0000 (15:51 -0400)] 
- add 0.9 changelog

11 years ago- Fixed bug where "python setup.py test" wasn't calling into
Mike Bayer [Wed, 9 Jul 2014 19:50:17 +0000 (15:50 -0400)] 
- Fixed bug where "python setup.py test" wasn't calling into
distutils appropriately, and errors would be emitted at the end
of the test suite.

11 years ago-do an autoflake8 here
Mike Bayer [Tue, 8 Jul 2014 23:22:44 +0000 (19:22 -0400)] 
-do an autoflake8 here

11 years ago- The "evaulator" for query.update()/delete() won't work with multi-table
Mike Bayer [Tue, 8 Jul 2014 23:08:42 +0000 (19:08 -0400)] 
- The "evaulator" for query.update()/delete() won't work with multi-table
updates, and needs to be set to `synchronize_session=False` or
`synchronize_session='fetch'`; this now raises an exception, with a
message to change the synchronize setting.  This will be only a warning
in 0.9.7.  fixes #3117

11 years agoMerge pull request #103 from tlocke/master
mike bayer [Tue, 8 Jul 2014 22:34:28 +0000 (18:34 -0400)] 
Merge pull request #103 from tlocke/master

pg8000 passing test/sql

11 years ago- add postgresql_regconfig argument to PG dialect for match() operator,
jonathan vanasco [Tue, 10 Jun 2014 22:56:27 +0000 (18:56 -0400)] 
- add postgresql_regconfig argument to PG dialect for match() operator,
implements PG's to_tsquery('regconfig', 'arg') pattern. fixes #3078

11 years agodocument JSONB
Mike Bayer [Mon, 7 Jul 2014 21:59:47 +0000 (17:59 -0400)] 
document JSONB

11 years agochangelog
Mike Bayer [Mon, 7 Jul 2014 14:42:19 +0000 (10:42 -0400)] 
changelog

11 years agoMerge pull request #101 from ddimmich/master
mike bayer [Mon, 7 Jul 2014 14:40:03 +0000 (10:40 -0400)] 
Merge pull request #101 from ddimmich/master

Postgres 9.4 Jsonb support

11 years agoPEP8 tidy of subset of test/sql/*.py
Tony Locke [Mon, 23 Jun 2014 21:34:01 +0000 (22:34 +0100)] 
PEP8 tidy of subset of test/sql/*.py

11 years agopg8000 passing test/sql
Tony Locke [Mon, 23 Jun 2014 19:45:16 +0000 (20:45 +0100)] 
pg8000 passing test/sql

11 years agoMerge pull request #100 from plaes/typos
mike bayer [Sun, 6 Jul 2014 21:17:59 +0000 (17:17 -0400)] 
Merge pull request #100 from plaes/typos

Typo fixes

11 years agochangelog, fixes #3093
Mike Bayer [Sun, 6 Jul 2014 21:16:26 +0000 (17:16 -0400)] 
changelog, fixes #3093

11 years agoMerged in therve/bug-3093/bug/3093 (pull request #24)
Mike Bayer [Sun, 6 Jul 2014 21:12:23 +0000 (17:12 -0400)] 
Merged in therve/bug-3093/bug/3093 (pull request #24)

Return the assigned value in MultableDict.setdefault

11 years agoadd pg8000
Mike Bayer [Sun, 6 Jul 2014 21:00:38 +0000 (17:00 -0400)] 
add pg8000

11 years agoMerge pull request #99 from tlocke/master
mike bayer [Sun, 6 Jul 2014 20:59:46 +0000 (16:59 -0400)] 
Merge pull request #99 from tlocke/master

pg8000 passing test/sql/test_types.py

11 years ago- changelog
Mike Bayer [Sun, 6 Jul 2014 20:55:28 +0000 (16:55 -0400)] 
- changelog

11 years agoMerge pull request #98 from abbec/mssql2008-multivalues-insert
mike bayer [Sun, 6 Jul 2014 20:55:05 +0000 (16:55 -0400)] 
Merge pull request #98 from abbec/mssql2008-multivalues-insert

Fixed support for multivalue inserts for MS SQL 2008

11 years ago- add test support for disconnect modification
Mike Bayer [Fri, 4 Jul 2014 20:49:11 +0000 (16:49 -0400)] 
- add test support for disconnect modification

11 years ago- clarify that the pg8000 backend has no impact of any kind on
Mike Bayer [Fri, 4 Jul 2014 19:49:01 +0000 (15:49 -0400)] 
- clarify that the pg8000 backend has no impact of any kind on
the pg8000 DBAPI's encoding behavior, fixes #3112.

11 years ago- rework the entire approach to #3076. As we need to catch all exceptions
Mike Bayer [Fri, 4 Jul 2014 19:40:47 +0000 (15:40 -0400)] 
- rework the entire approach to #3076. As we need to catch all exceptions
in all cases unconditionally, the number of use cases that go beyond what
dbapi_error() is expecting has gone too far for an 0.9 release.
Additionally, the number of things we'd like to track is really a lot
more than the five arguments here, and ExecutionContext is really not
suitable as totally public API for this.   So restore dbapi_error
to its old version, deprecate, and build out handle_error instead.
This is a lot more extensible and doesn't get in the way of anything
compatibility-wise.

11 years ago- Added new attributes :attr:`.ExecutionContext.exception` and
Mike Bayer [Fri, 4 Jul 2014 01:49:10 +0000 (21:49 -0400)] 
- Added new attributes :attr:`.ExecutionContext.exception` and
:attr:`.ExecutionContext.is_disconnect` which are meaningful within
the :meth:`.ConnectionEvents.dbapi_error` handler to see both the
original DBAPI error as well as whether or not it represents
a disconnect.

11 years ago- The mechanics of the :meth:`.ConnectionEvents.dbapi_error` handler
Mike Bayer [Thu, 3 Jul 2014 21:30:49 +0000 (17:30 -0400)] 
- The mechanics of the :meth:`.ConnectionEvents.dbapi_error` handler
have been enhanced such that the function handler is now capable
of raising or returning a new exception object, which will replace
the exception normally being thrown by SQLAlchemy.
fixes #3076

11 years ago- add link to dialect docs for SQLite autoincrement from column autoincrement,
Mike Bayer [Thu, 3 Jul 2014 15:34:51 +0000 (11:34 -0400)] 
- add link to dialect docs for SQLite autoincrement from column autoincrement,
fixes #3110

11 years ago- ensure ACID set up on this table for MySQL
Mike Bayer [Tue, 1 Jul 2014 19:28:16 +0000 (15:28 -0400)] 
- ensure ACID set up on this table for MySQL

11 years ago- Fixed bug where items that were persisted, deleted, or had a
Mike Bayer [Tue, 1 Jul 2014 16:12:27 +0000 (12:12 -0400)] 
- Fixed bug where items that were persisted, deleted, or had a
primary key change within a savepoint block would not
participate in being restored to their former state (not in
session, in session, previous PK) after the outer transaction
were rolled back. fixes #3108

11 years agojsonb support for <@, ?| and ?& added.
Damian Dimmich [Tue, 1 Jul 2014 09:24:30 +0000 (13:24 +0400)] 
jsonb support for <@, ?| and ?& added.

need to see if equality already works.

11 years ago- repair the _enable_single_crit method, it was named the same
Mike Bayer [Tue, 1 Jul 2014 00:25:04 +0000 (20:25 -0400)] 
- repair the _enable_single_crit method, it was named the same
as the attribute and probably just replaced itself, so that is
now _set_enable_single_crit
- as a side effect of the main issue fixed here, correct the case in
adjust_for_single_inheritance where the same mapper appears more
than once in mapper_adapter_map; run through a set() for uniqueness.
- Fixed bug in subquery eager loading in conjunction with
:func:`.with_polymorphic`, the targeting of entities and columns
in the subquery load has been made more accurate with respect
to this type of entity and others. Fixes #3106

11 years ago- add new section Working with Raw DBAPI Connections, fixes #2218.
Mike Bayer [Mon, 30 Jun 2014 15:08:05 +0000 (11:08 -0400)] 
- add new section Working with Raw DBAPI Connections, fixes #2218.

11 years ago- attach the ResultMetaData to the Compiled object, when we detect that
Mike Bayer [Sun, 29 Jun 2014 04:10:59 +0000 (00:10 -0400)] 
- attach the ResultMetaData to the Compiled object, when we detect that
the compiled cache is used.  That allows us to cache the whole metadata
and save on creating it at result time, when compiled cache is used.

11 years ago- cut out some calls by inlining the calls to get_history()
Mike Bayer [Sun, 29 Jun 2014 03:34:05 +0000 (23:34 -0400)] 
- cut out some calls by inlining the calls to get_history()

11 years agoit's OK to pass a dict in - it does the right thing, no need to quote it
Damian Dimmich [Sat, 28 Jun 2014 19:24:36 +0000 (23:24 +0400)] 
it's OK to pass a dict in - it does the right thing, no need to quote it
in the tests.

11 years agominor cleanup of the jsonb - had extraneous operators that where copied
Damian Dimmich [Sat, 28 Jun 2014 19:11:03 +0000 (23:11 +0400)] 
minor cleanup of the jsonb - had extraneous operators that where copied
from hstore that don't apply.

Add tests for ? and @> operators.

11 years agoadd has_key & contains operators for jsonb (ported over from hstore)
Damian Dimmich [Sat, 28 Jun 2014 18:47:20 +0000 (22:47 +0400)] 
add has_key & contains operators for jsonb (ported over from hstore)

11 years agoand tests for JSONB - as this is a superset of JSON i've subclassed
Damian Dimmich [Sat, 28 Jun 2014 16:02:24 +0000 (20:02 +0400)] 
and tests for JSONB - as this is a superset of JSON i've subclassed
the JSON tests as all of these should be applicable as well.

11 years agotypo: s/thbe/the
Priit Laes [Sat, 28 Jun 2014 09:47:19 +0000 (12:47 +0300)] 
typo: s/thbe/the

11 years agotypo: s/tranasction/transaction
Priit Laes [Sat, 28 Jun 2014 09:46:46 +0000 (12:46 +0300)] 
typo: s/tranasction/transaction

11 years agotypo: s/founds/found
Priit Laes [Sat, 28 Jun 2014 09:45:35 +0000 (12:45 +0300)] 
typo: s/founds/found

11 years ago- Fixed a bug within the custom operator plus :meth:`.TypeEngine.with_variant`
Mike Bayer [Fri, 27 Jun 2014 20:08:42 +0000 (16:08 -0400)] 
- Fixed a bug within the custom operator plus :meth:`.TypeEngine.with_variant`
system, whereby using a :class:`.TypeDecorator` in conjunction with
variant would fail with an MRO error when a comparison operator was used.
fixes #3102

11 years agowrong version number
Mike Bayer [Fri, 27 Jun 2014 20:07:05 +0000 (16:07 -0400)] 
wrong version number

11 years ago- MySQL error 2014 "commands out of sync" appears to be raised as a
Mike Bayer [Fri, 27 Jun 2014 19:52:40 +0000 (15:52 -0400)] 
- MySQL error 2014 "commands out of sync" appears to be raised as a
ProgrammingError, not OperationalError, in modern MySQL-Python versions;
all MySQL error codes that are tested for "is disconnect" are now
checked within OperationalError and ProgrammingError regardless.
fixes #3101

11 years ago- Fixed bug involving dynamic attributes, that was again a regression
Mike Bayer [Thu, 26 Jun 2014 18:58:42 +0000 (14:58 -0400)] 
- Fixed bug involving dynamic attributes, that was again a regression
of :ticket:`3060` from verision 0.9.5.  A self-referential relationship
with lazy='dynamic' would raise a TypeError within a flush operation.
fixes #3099

11 years agoadd missing release date
Mike Bayer [Wed, 25 Jun 2014 18:31:18 +0000 (14:31 -0400)] 
add missing release date

11 years ago- Fixed bug when the declarative ``__abstract__`` flag was not being
Mike Bayer [Wed, 25 Jun 2014 18:30:25 +0000 (14:30 -0400)] 
- Fixed bug when the declarative ``__abstract__`` flag was not being
distinguished for when it was actually the value ``False``.
The ``__abstract__`` flag needs to acutally evaluate to a True
value at the level being tested.
fixes #3097

11 years ago- add a test to confirm #3096, we definitely get a timedelta
Mike Bayer [Wed, 25 Jun 2014 18:23:27 +0000 (14:23 -0400)] 
- add a test to confirm #3096, we definitely get a timedelta

11 years agoMerged in uralbash/sqlalchemy/uralbash/fix-pullreq-link-in-changelog_09rst-edit-14035...
Mike Bayer [Tue, 24 Jun 2014 16:27:01 +0000 (12:27 -0400)] 
Merged in uralbash/sqlalchemy/uralbash/fix-pullreq-link-in-changelog_09rst-edit-1403586718274 (pull request #23)

fix pullreq link in changelog_09.rst edited online with Bitbucket

11 years ago- write some really complete docs on MSSQL autoincrement, fixes #3094
Mike Bayer [Tue, 24 Jun 2014 16:25:07 +0000 (12:25 -0400)] 
- write some really complete docs on MSSQL autoincrement, fixes #3094

11 years agoReturn the assigned value in MultableDict.setdefault
Thomas Herve [Tue, 24 Jun 2014 07:55:53 +0000 (09:55 +0200)] 
Return the assigned value in MultableDict.setdefault

11 years agoinitial support for JSONB - this only allows you to define the JSONB
Damian Dimmich [Tue, 24 Jun 2014 07:23:21 +0000 (11:23 +0400)] 
initial support for JSONB - this only allows you to define the JSONB
datatype - this does not add any of the additional support for
querying/indexing yet.

11 years agofix pullreq link in changelog_09.rst edited online with Bitbucket
uralbash [Tue, 24 Jun 2014 05:12:14 +0000 (05:12 +0000)] 
fix pullreq link in changelog_09.rst edited online with Bitbucket

11 years ago- reverse course in #3061 so that we instead no longer set None in the attribute
Mike Bayer [Mon, 23 Jun 2014 23:50:23 +0000 (19:50 -0400)] 
- reverse course in #3061 so that we instead no longer set None in the attribute
when we do a get; we return the None as always but we leave the dict blank
and the loader callable still in place.  The case for this implicit get on a pending object is not
super common and there really should be no change in state at all when this
operation proceeds.   This change is more dramatic as it reverses
a behavior SQLA has had since the first release.
fixes #3061

11 years ago- Reverted the change for :ticket:`3060` - this is a unit of work
Mike Bayer [Mon, 23 Jun 2014 22:38:23 +0000 (18:38 -0400)] 
- Reverted the change for :ticket:`3060` - this is a unit of work
fix that is updated more comprehensively in 1.0 via :ticket:`3061`.
The fix in :ticket:`3060` unfortunately produces a new issue whereby
an eager load of a many-to-one attribute can produce an event
that is interpreted into an attribute change.

11 years agoPEP8 tidy of test/sql/test_types.py
Tony Locke [Mon, 23 Jun 2014 21:34:01 +0000 (22:34 +0100)] 
PEP8 tidy of test/sql/test_types.py

11 years agopg8000 passing test/sql/test_types.py
Tony Locke [Mon, 23 Jun 2014 19:45:16 +0000 (20:45 +0100)] 
pg8000 passing test/sql/test_types.py

Opened up two tests that now pass with pg8000. Also, rewrote two tests
to use actual tables rather than having a round trip in a single select
statement. This is necessary for pg8000 because it sends strings to the
server with type 'unknown' and lets the server work out the type.

11 years ago0.9.5 released date
Mike Bayer [Mon, 23 Jun 2014 15:17:39 +0000 (11:17 -0400)] 
0.9.5 released date

11 years agofix changelog
Mike Bayer [Mon, 23 Jun 2014 15:12:47 +0000 (11:12 -0400)] 
fix changelog

11 years ago- use new Or() directive to test for both SQL variants here, which can vary
Mike Bayer [Sun, 22 Jun 2014 00:27:19 +0000 (20:27 -0400)] 
- use new Or() directive to test for both SQL variants here, which can vary
based on hash ordering of the ClassManager

11 years ago- fix test failures
Mike Bayer [Sat, 21 Jun 2014 22:11:04 +0000 (18:11 -0400)] 
- fix test failures

11 years ago- Additional checks have been added for the case where an inheriting
Mike Bayer [Fri, 20 Jun 2014 22:47:28 +0000 (18:47 -0400)] 
- Additional checks have been added for the case where an inheriting
mapper is implicitly combining one of its column-based attributes
with that of the parent, where those columns normally don't necessarily
share the same value.  This is an extension of an existing check that
was added via :ticket:`1892`; however this new check emits only a
warning, instead of an exception, to allow for applications that may
be relying upon the existing behavior.
fixes #3042

11 years agoadditional version notes
Mike Bayer [Fri, 20 Jun 2014 22:06:15 +0000 (18:06 -0400)] 
additional version notes

11 years agoadd missing version markers
Mike Bayer [Fri, 20 Jun 2014 22:04:26 +0000 (18:04 -0400)] 
add missing version markers

11 years ago- The :paramref:`.Column.nullable` flag is implicitly set to ``False``
Mike Bayer [Fri, 20 Jun 2014 22:03:28 +0000 (18:03 -0400)] 
- The :paramref:`.Column.nullable` flag is implicitly set to ``False``
when that :class:`.Column` is referred to in an explicit
:class:`.PrimaryKeyConstraint` for that table.  This behavior now
matches that of when the :class:`.Column` itself has the
:paramref:`.Column.primary_key` flag set to ``True``, which is
intended to be an exactly equivalent case.
fixes #3023

11 years ago- Added a new type :class:`.postgresql.OID` to the Postgresql dialect.
Mike Bayer [Fri, 20 Jun 2014 21:58:06 +0000 (17:58 -0400)] 
- Added a new type :class:`.postgresql.OID` to the Postgresql dialect.
While "oid" is generally a private type within PG that is not exposed
in modern versions, there are some PG use cases such as large object
support where these types might be exposed, as well as within some
user-reported schema reflection use cases.
fixes #3002

11 years ago- Fixed bug where column names added to ``mysql_length`` parameter
Mike Bayer [Wed, 18 Jun 2014 14:56:23 +0000 (10:56 -0400)] 
- Fixed bug where column names added to ``mysql_length`` parameter
on an index needed to have the same quoting for quoted names in
order to be recognized.  The fix makes the quotes optional but
also provides the old behavior for backwards compatibility with those
using the workaround.
fixes #3085