]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
14 years ago- The cx_oracle "decimal detection" logic, which takes place
Mike Bayer [Thu, 18 Nov 2010 17:19:31 +0000 (12:19 -0500)] 
- The cx_oracle "decimal detection" logic, which takes place
for for result set columns with ambiguous numeric characteristics,
now uses the decimal point character determined by the locale/
NLS_LANG setting, using an on-first-connect detection of
this character.  cx_oracle 5.0.3 or greater is also required
when using a non-period-decimal-point NLS_LANG setting.
[ticket:1953].

14 years ago- An error is raised if __table_args__ is not in tuple
Mike Bayer [Wed, 17 Nov 2010 15:55:10 +0000 (10:55 -0500)] 
- An error is raised if __table_args__ is not in tuple
or dict format, and is not None.  [ticket:1972]

14 years ago- primary_key arg to mapper() is a list
Mike Bayer [Tue, 16 Nov 2010 22:42:18 +0000 (17:42 -0500)] 
- primary_key arg to mapper() is a list
- other modernizations to mapper() doc

14 years ago- Added a bind processor for booleans which coerces
Mike Bayer [Mon, 15 Nov 2010 17:40:55 +0000 (12:40 -0500)] 
- Added a bind processor for booleans which coerces
to int, for DBAPIs such as pymssql that naively call
str() on values.

14 years ago- Fixed indexing of Query objects by -1. It was erroneously
Mike Bayer [Mon, 15 Nov 2010 14:55:43 +0000 (09:55 -0500)] 
- Fixed indexing of Query objects by -1. It was erroneously
transformed to the empty slice -1:0 that resulted in
IndexError. [ticket:1968]

14 years ago- add a few choice TypeDecorator recipes
Mike Bayer [Mon, 15 Nov 2010 14:12:45 +0000 (09:12 -0500)] 
- add a few choice TypeDecorator recipes

14 years ago- add helpstring for site-mako
Mike Bayer [Sun, 14 Nov 2010 19:02:11 +0000 (14:02 -0500)] 
- add helpstring for site-mako

14 years ago- The 'info' attribute of Column is copied during
Mike Bayer [Sun, 14 Nov 2010 17:42:34 +0000 (12:42 -0500)] 
- The 'info' attribute of Column is copied during
Column.copy(), i.e. as occurs when using columns
in declarative mixins.  [ticket:1967]

14 years ago- move inline "import" statements to use new "util.importlater()" construct. cuts
Mike Bayer [Sat, 13 Nov 2010 18:19:36 +0000 (13:19 -0500)] 
- move inline "import" statements to use new "util.importlater()" construct.  cuts
down on clutter, timeit says there's a teeny performance gain, at least where
the access is compared against attr.subattr.  these aren't super-critical
calls anyway
- slight inlining in _class_to_mapper

14 years ago- additional fix for [ticket:1964]
Mike Bayer [Sat, 13 Nov 2010 17:05:13 +0000 (12:05 -0500)] 
- additional fix for [ticket:1964]
- [ticket:1965] is invalid

14 years ago- nose3 seems to support "addplugins" now
Mike Bayer [Sat, 13 Nov 2010 16:26:05 +0000 (11:26 -0500)] 
- nose3 seems to support "addplugins" now
- add a note about sqla_nose.py

14 years agocomment
Mike Bayer [Sat, 13 Nov 2010 16:18:18 +0000 (11:18 -0500)] 
comment

14 years ago- On the same theme, the REFERENCES clause in a CREATE TABLE
Mike Bayer [Fri, 12 Nov 2010 15:49:17 +0000 (10:49 -0500)] 
- On the same theme, the REFERENCES clause in a CREATE TABLE
that includes a remote schema to a *different* schema
than that of the parent table doesn't render at all,
as cross-schema references do not appear to be supported.

14 years ago- The REFERENCES clause in a CREATE TABLE that includes
Mike Bayer [Fri, 12 Nov 2010 15:36:03 +0000 (10:36 -0500)] 
- The REFERENCES clause in a CREATE TABLE that includes
a remote schema name now renders the remote name without
the schema clause, as required by SQLite.  [ticket:1851]

14 years agowhoopsie
Mike Bayer [Fri, 12 Nov 2010 00:27:35 +0000 (19:27 -0500)] 
whoopsie

14 years ago- Fixed bug regarding "subqueryload" strategy whereby
Mike Bayer [Fri, 12 Nov 2010 00:15:26 +0000 (19:15 -0500)] 
- Fixed bug regarding "subqueryload" strategy whereby
strategy would fail if the entity was an aliased()
construct.  [ticket:1964]

14 years agostupid tab got committed
Mike Bayer [Tue, 9 Nov 2010 17:19:11 +0000 (12:19 -0500)] 
stupid tab got committed

14 years agoadd a note to use column_property with declarative map to a join
Mike Bayer [Tue, 9 Nov 2010 17:09:41 +0000 (12:09 -0500)] 
add a note to use column_property with declarative map to a join

14 years ago- and fix everything from the last checkin that breaks on py 2.4, 2.5
Mike Bayer [Tue, 9 Nov 2010 16:46:52 +0000 (11:46 -0500)] 
- and fix everything from the last checkin that breaks on py 2.4, 2.5

14 years ago- Implemented sequence check capability for the C
Mike Bayer [Tue, 9 Nov 2010 16:40:23 +0000 (11:40 -0500)] 
- Implemented sequence check capability for the C
version of RowProxy, as well as 2.7 style
"collections.Sequence" registration for RowProxy.
[ticket:1871]

14 years ago- "innerjoin" flag doesn't take effect along the chain
Mike Bayer [Fri, 5 Nov 2010 05:02:48 +0000 (01:02 -0400)] 
- "innerjoin" flag doesn't take effect along the chain
of joinedload() joins if a previous join in that chain
is an outer join, thus allowing primary rows without
a referenced child row to be correctly returned
in results.  [ticket:1954]

14 years ago- mysql
Mike Bayer [Wed, 3 Nov 2010 19:45:21 +0000 (15:45 -0400)] 
- mysql
  - Fixed error handling for Jython + zxjdbc, such that
    has_table() property works again.  Regression from
    0.6.3 (we don't have a Jython buildbot, sorry)
    [ticket:1960]

14 years agofurther edits
Mike Bayer [Mon, 1 Nov 2010 20:07:47 +0000 (16:07 -0400)] 
further edits

14 years ago- write a new section describing the "subtransactions=True" flag in full detail
Mike Bayer [Mon, 1 Nov 2010 19:59:53 +0000 (15:59 -0400)] 
- write a new section describing the "subtransactions=True" flag in full detail

14 years ago- Fixed bug whereby a non-"mutable" attribute modified event
Mike Bayer [Tue, 26 Oct 2010 20:12:04 +0000 (16:12 -0400)] 
- Fixed bug whereby a non-"mutable" attribute modified event
which occurred on an object that was clean except for
preceding mutable attribute changes would fail to strongly
reference itself in the identity map. This would cause the
object to be garbage collected, losing track of any changes
that weren't previously saved in the "mutable changes"
dictionary.

14 years ago- engine status updates
Mike Bayer [Sun, 24 Oct 2010 20:08:14 +0000 (16:08 -0400)] 
- engine status updates

14 years agoAdded tag rel_0_6_5 for changeset b96e6232a42918692e95b32b7dd3723cd4a242bb
Mike Bayer [Sun, 24 Oct 2010 17:39:11 +0000 (13:39 -0400)] 
Added tag rel_0_6_5 for changeset b96e6232a42918692e95b32b7dd3723cd4a242bb

14 years agocleanup most egregiously long lines rel_0_6_5
Mike Bayer [Sun, 24 Oct 2010 17:26:34 +0000 (13:26 -0400)] 
cleanup most egregiously long lines

14 years ago- remove remaining create_session() calls from examples, replace with Session
Mike Bayer [Sun, 24 Oct 2010 16:42:47 +0000 (12:42 -0400)] 
- remove remaining create_session() calls from examples, replace with Session
- replace all flush()/expunge_all() with commit()

14 years ago- mssql+pymssql dialect now honors the "port" portion
Mike Bayer [Sun, 24 Oct 2010 16:00:58 +0000 (12:00 -0400)] 
- mssql+pymssql dialect now honors the "port" portion
of the URL instead of discarding it.  [ticket:1952]
- testing.only_on() accepts db specs optionally as a list

14 years ago- Rewrote the reflection of indexes to use sys.
Mike Bayer [Sat, 23 Oct 2010 22:09:36 +0000 (18:09 -0400)] 
- Rewrote the reflection of indexes to use sys.
catalogs, so that column names of any configuration
(spaces, embedded commas, etc.) can be reflected.
Note that reflection of indexes requires SQL
Server 2005 or greater.  [ticket:1770]

14 years ago- make a futile attempt to combat the needless usage of clear_mappers() in the wild,
Mike Bayer [Sat, 23 Oct 2010 21:09:53 +0000 (17:09 -0400)] 
- make a futile attempt to combat the needless usage of clear_mappers() in the wild,
[ticket:1861]

14 years ago- Added type_coerce(expr, type_) expression element.
Mike Bayer [Sat, 23 Oct 2010 20:40:39 +0000 (16:40 -0400)] 
- Added type_coerce(expr, type_) expression element.
Treats the given expression as the given type when evaluating
expressions and processing result rows, but does not
affect the generation of SQL, other than an anonymous label.

14 years ago- Oracle - the implicit_retunring argument to create_engine()
Mike Bayer [Sat, 23 Oct 2010 19:27:21 +0000 (15:27 -0400)] 
- Oracle - the implicit_retunring argument to create_engine()
is now honored regardless of detected version of
Oracle.  Previously, the flag would be forced
to False if server version info was < 10.
[ticket:1878]

14 years agofix typo
Mike Bayer [Sat, 23 Oct 2010 17:43:39 +0000 (13:43 -0400)] 
fix typo

14 years ago- New Query methods: query.label(name), query.as_scalar(),
Mike Bayer [Sat, 23 Oct 2010 17:07:56 +0000 (13:07 -0400)] 
- New Query methods: query.label(name), query.as_scalar(),
return the query's statement as a scalar subquery
with /without label [ticket:1920];
query.with_entities(*ent), replaces the SELECT list of
the query with new entities.
Roughly equivalent to a generative form of query.values()
which accepts mapped entities as well as column
expressions.

14 years ago- A mixin can now specify a column that overrides
Mike Bayer [Thu, 21 Oct 2010 20:54:39 +0000 (16:54 -0400)] 
- A mixin can now specify a column that overrides
a column of the same name associated with a superclass.
Thanks to Oystein Haaland.

14 years ago- Fixed bug in TypeDecorator whereby the dialect-specific
Mike Bayer [Thu, 21 Oct 2010 20:42:32 +0000 (16:42 -0400)] 
- Fixed bug in TypeDecorator whereby the dialect-specific
type was getting pulled in to generate the DDL for a
given type, which didn't always return the correct result.

- TypeDecorator can now have a fully constructed type
specified as its "impl", in addition to a type class.

- TypeDecorator will now place itself as the resulting
type for a binary expression where the type coercion
rules would normally return its impl type - previously,
a copy of the impl type would be returned which would
have the TypeDecorator embedded into it as the "dialect"
impl, this was probably an unintentional way of achieving
the desired effect.

- TypeDecorator.load_dialect_impl() returns "self.impl" by
default, i.e. not the dialect implementation type of
"self.impl".   This to support compilation correctly.
Behavior can be user-overridden in exactly the same way
as before to the same effect.

14 years ago- document is_modified() caveats including that the "old" value isn't always
Mike Bayer [Wed, 20 Oct 2010 21:58:35 +0000 (17:58 -0400)] 
- document is_modified() caveats including that the "old" value isn't always
present, and that this results in a positive response for "changed"
[ticket:1928]

14 years ago- Fixed mysql bug involving reflection of CURRENT_TIMESTAMP
Mike Bayer [Wed, 20 Oct 2010 21:44:35 +0000 (17:44 -0400)] 
- Fixed mysql bug involving reflection of CURRENT_TIMESTAMP
default used with ON UPDATE clause, thanks to
Taavi Burns [ticket:1940]

14 years ago- Fixed bug which prevented "domain" built from a
Mike Bayer [Wed, 20 Oct 2010 21:28:08 +0000 (17:28 -0400)] 
- Fixed bug which prevented "domain" built from a
custom type such as "enum" from being reflected.
[ticket:1933]

14 years agomerge tip
Mike Bayer [Wed, 20 Oct 2010 20:22:30 +0000 (16:22 -0400)] 
merge tip

14 years ago- Added "as_tuple" flag to pg ARRAY type, returns results
Mike Bayer [Wed, 20 Oct 2010 20:17:17 +0000 (16:17 -0400)] 
- Added "as_tuple" flag to pg ARRAY type, returns results
as tuples instead of lists to allow hashing.

14 years ago- Fixed bug where aliasing of tables with "schema" would
Mike Bayer [Wed, 20 Oct 2010 20:15:54 +0000 (16:15 -0400)] 
- Fixed bug where aliasing of tables with "schema" would
fail to compile properly.  [ticket:1943]

14 years agoUpdate for new nose plugin location.
Jason Kirtland [Tue, 19 Oct 2010 23:44:15 +0000 (16:44 -0700)] 
Update for new nose plugin location.

14 years ago- the NoseSQLAlchemyPlugin has been moved to a
Mike Bayer [Sun, 17 Oct 2010 19:23:14 +0000 (15:23 -0400)] 
- the NoseSQLAlchemyPlugin has been moved to a
new package "sqlalchemy_nose" which installs
along with "sqlalchemy".  This so that the "nosetests"
script works as always but also allows the
--with-coverage option to turn on coverage before
SQLAlchemy modules are imported, allowing coverage
to work correctly.
- added some new ignores

14 years ago- forgot to add the immediate loader tests
Mike Bayer [Sun, 17 Oct 2010 19:10:08 +0000 (15:10 -0400)] 
- forgot to add the immediate loader tests

14 years agoedit
Mike Bayer [Sun, 17 Oct 2010 04:07:13 +0000 (00:07 -0400)] 
edit

14 years agonote expire_on_commit in the docstring for commit() as well
Mike Bayer [Sun, 17 Oct 2010 03:50:33 +0000 (23:50 -0400)] 
note expire_on_commit in the docstring for commit() as well

14 years ago- Fixed MSSQL reflection bug which did not properly handle
Mike Bayer [Sat, 16 Oct 2010 16:15:40 +0000 (12:15 -0400)] 
- Fixed MSSQL reflection bug which did not properly handle
reflection of unknown types.  [ticket:1946]

14 years ago- Added a new "lazyload" option "immediateload".
Mike Bayer [Fri, 15 Oct 2010 15:59:02 +0000 (11:59 -0400)] 
- Added a new "lazyload" option "immediateload".
Issues the usual "lazy" load operation automatically
as the object is populated.   The use case
here is when loading objects to be placed in
an offline cache, or otherwise used after
the session isn't available, and straight 'select'
loading, not 'joined' or 'subquery', is desired.
[ticket:1914]

14 years ago - engine_from_config() now accepts 'debug' for
Mike Bayer [Fri, 15 Oct 2010 15:43:59 +0000 (11:43 -0400)] 
 - engine_from_config() now accepts 'debug' for
'echo', 'echo_pool', 'force' for 'convert_unicode',
boolean values for 'use_native_unicode'.
[ticket:1899]

14 years ago- Fixed labeling bug in Query whereby the NamedTuple
Mike Bayer [Thu, 14 Oct 2010 16:04:51 +0000 (12:04 -0400)] 
- Fixed labeling bug in Query whereby the NamedTuple
would mis-apply labels if any of the column
expressions were un-labeled.

14 years agowarn when two classes with the same name end up in the declarative registry
Chris Withers [Wed, 6 Oct 2010 11:04:25 +0000 (12:04 +0100)] 
warn when two classes with the same name end up in the declarative registry

14 years agodocument creating indexes for columns in mixins
Chris Withers [Mon, 4 Oct 2010 09:14:54 +0000 (10:14 +0100)] 
document creating indexes for columns in mixins

14 years ago- default compilation of a type will check if the type class's module is
Mike Bayer [Sun, 3 Oct 2010 17:11:41 +0000 (13:11 -0400)] 
- default compilation of a type will check if the type class's module is
inside of sqlalchemy.dialects, in which case that dialect's default
will be used instead of DefaultDialect.

14 years ago- added "views=True" option to metadata.reflect(),
Mike Bayer [Sat, 2 Oct 2010 22:05:19 +0000 (18:05 -0400)] 
- added "views=True" option to metadata.reflect(),
will add the list of available views to those
being reflected.  [ticket:1936]

14 years ago- this test began working for oracle as of r14af1f73fbbd. Additionally,
Mike Bayer [Sat, 2 Oct 2010 05:30:17 +0000 (01:30 -0400)] 
- this test began working for oracle as of r14af1f73fbbd.   Additionally,
the operation wasn't actually "working" on sqlite, mysql prior to that
rev, but no assertions checked. added assertions for those.

14 years ago- *Major* cleanup / modernization of the Informix
Mike Bayer [Sat, 2 Oct 2010 01:26:05 +0000 (21:26 -0400)] 
- *Major* cleanup / modernization of the Informix
dialect for 0.6, courtesy Florian Apolloner.
[ticket:1906]

14 years ago- Fixed bug in query.update() where 'evaluate' or 'fetch'
Mike Bayer [Sat, 2 Oct 2010 00:55:42 +0000 (20:55 -0400)] 
- Fixed bug in query.update() where 'evaluate' or 'fetch'
expiration would fail if the column expression key was
a class attribute with a different keyname as the
actual column name.  [ticket:1935]

14 years agocorrect that only MySQL doesn't support CHECK constraints, [ticket:1927]
Mike Bayer [Sat, 2 Oct 2010 00:40:56 +0000 (20:40 -0400)] 
correct that only MySQL doesn't support CHECK constraints, [ticket:1927]

14 years ago- Make it clear that the second mapping of User is against User.__table__, [ticket...
Mike Bayer [Sat, 2 Oct 2010 00:39:35 +0000 (20:39 -0400)] 
- Make it clear that the second mapping of User is against User.__table__, [ticket:1934]

14 years ago- add additional logic that duplicates mapper's prop.copy(); prop.columns.append...
Mike Bayer [Sat, 2 Oct 2010 00:29:04 +0000 (20:29 -0400)] 
- add additional logic that duplicates mapper's prop.copy(); prop.columns.append(col)
logic when columns are present in a joined subclass with an attribute name different
than the column name itself [ticket:1931]
- add coverage to verify that we need to check (obj.name or name) when
deciding if a Column from a mixin should be added to the mapped table

14 years ago- reworked the internals of mapper.cascade_iterator() to
Mike Bayer [Fri, 1 Oct 2010 18:23:01 +0000 (14:23 -0400)] 
- reworked the internals of mapper.cascade_iterator() to
cut down method calls by about 9% in some circumstances.
[ticket:1932]

14 years agopy2.4 didn't seem to like super() on an exception
Mike Bayer [Wed, 29 Sep 2010 14:34:45 +0000 (10:34 -0400)] 
py2.4 didn't seem to like super() on an exception

14 years agodon't need itertools here
Mike Bayer [Wed, 29 Sep 2010 06:15:07 +0000 (02:15 -0400)] 
don't need itertools here

14 years ago- CircularDependencyError now has .cycles and .edges
Mike Bayer [Wed, 29 Sep 2010 06:11:38 +0000 (02:11 -0400)] 
- CircularDependencyError now has .cycles and .edges
members, which are the set of elements involved in
one or more cycles, and the set of edges as 2-tuples.
[ticket:1890]

14 years ago- add Sphinx 1.0 makefile and config options, [ticket:1929]
Mike Bayer [Wed, 29 Sep 2010 05:52:25 +0000 (01:52 -0400)] 
- add Sphinx 1.0 makefile and config options, [ticket:1929]

14 years ago- Fixed bug whereby columns on a mixin wouldn't propagate
Mike Bayer [Tue, 28 Sep 2010 16:43:36 +0000 (12:43 -0400)] 
- Fixed bug whereby columns on a mixin wouldn't propagate
correctly to a single-table inheritance scheme where
the attribute name is different than that of the column.
[ticket:1930].   Note [ticket:1931] which is the same
issue for joined inh, not yet resolved.

14 years ago@mapperproperty -> @declared_attr
Mike Bayer [Mon, 27 Sep 2010 14:01:48 +0000 (10:01 -0400)] 
@mapperproperty -> @declared_attr

14 years agofor internal link to top render #
Mike Bayer [Sun, 26 Sep 2010 16:55:51 +0000 (12:55 -0400)] 
for internal link to top render #

14 years ago- @classproperty 's official name/location for usage
Mike Bayer [Sat, 25 Sep 2010 23:25:31 +0000 (19:25 -0400)] 
- @classproperty 's official name/location for usage
with declarative is sqlalchemy.ext.declarative.mapperproperty.
Same thing, but moving there since it is more of a
"marker" that's specific to declararative,
not just an attribute technique.  [ticket:1915]

14 years ago- Fixed recursion overflow which could occur when operating
Mike Bayer [Sat, 25 Sep 2010 22:38:01 +0000 (18:38 -0400)] 
- Fixed recursion overflow which could occur when operating
with two expressions both of type "NullType", but
not the singleton NULLTYPE instance. [ticket:1907]

14 years ago- fix a typo that was apparently not really impacting anything
Mike Bayer [Sat, 25 Sep 2010 20:19:49 +0000 (16:19 -0400)] 
- fix a typo that was apparently not really impacting anything

14 years agofix the default here
Mike Bayer [Sat, 25 Sep 2010 20:15:32 +0000 (16:15 -0400)] 
fix the default here

14 years agomerge tip
Mike Bayer [Fri, 24 Sep 2010 15:11:43 +0000 (11:11 -0400)] 
merge tip

14 years ago- optimized_al doesn't need mapper extension
Mike Bayer [Fri, 24 Sep 2010 15:11:12 +0000 (11:11 -0400)] 
- optimized_al doesn't need mapper extension
- convert to non-bound metadata, Session, session.commit()

14 years agomake less ficticious ;-)
Chris Withers [Fri, 24 Sep 2010 11:10:43 +0000 (12:10 +0100)] 
make less ficticious ;-)

14 years agodoc edits
Mike Bayer [Thu, 23 Sep 2010 23:24:28 +0000 (19:24 -0400)] 
doc edits

14 years agodoc edits
Mike Bayer [Thu, 23 Sep 2010 03:48:17 +0000 (23:48 -0400)] 
doc edits

14 years ago- in depth docs about some merge() tips
Mike Bayer [Wed, 22 Sep 2010 18:22:16 +0000 (14:22 -0400)] 
- in depth docs about some merge() tips
- docs about backref cascade
- Another new flag on relationship(), cascade_backrefs,
disables the "save-update" cascade when the event was
initiated on the "reverse" side of a bidirectional
relationship.   This is a cleaner behavior so that
many-to-ones can be set on a transient object without
it getting sucked into the child object's session,
while still allowing the forward collection to
cascade.   We *might* default this to False in 0.7.

14 years ago- Query.select_from() has been beefed up to help
Mike Bayer [Tue, 21 Sep 2010 22:09:29 +0000 (18:09 -0400)] 
- Query.select_from() has been beefed up to help
ensure that a subsequent call to query.join()
will use the select_from() entity, assuming it's
a mapped entity and not a plain selectable,
as the default "left" side, not the first entity
in the Query object's list of entities.

14 years agouse default dialect for compile tests
Mike Bayer [Tue, 21 Sep 2010 15:12:33 +0000 (11:12 -0400)] 
use default dialect for compile tests

14 years ago- the logging message emitted by the engine when
Mike Bayer [Tue, 21 Sep 2010 15:10:07 +0000 (11:10 -0400)] 
- the logging message emitted by the engine when
a connection is first used is now "BEGIN (implicit)"
to emphasize that DBAPI has no explicit begin().

14 years ago- Patched a case where query.join() would adapt the
Mike Bayer [Tue, 21 Sep 2010 14:11:10 +0000 (10:11 -0400)] 
- Patched a case where query.join() would adapt the
right side to the right side of the left's join
inappropriately [ticket:1925]

14 years ago- scoped_session emits a warning when configure() is
Mike Bayer [Tue, 21 Sep 2010 13:33:30 +0000 (09:33 -0400)] 
- scoped_session emits a warning when configure() is
called if a Session is already present (checks only the
current thread) [ticket:1924]

14 years ago- @classproperty (soon/now @mapperproperty) takes effect for
Mike Bayer [Mon, 20 Sep 2010 16:00:14 +0000 (12:00 -0400)] 
- @classproperty (soon/now @mapperproperty) takes effect for
__mapper_args__, __table_args__, __tablename__ on
a base class that is not a mixin, as well as mixins.
[ticket:1922]

14 years ago- as_scalar(), label() can be called on a selectable
Mike Bayer [Sat, 18 Sep 2010 17:34:04 +0000 (13:34 -0400)] 
- as_scalar(), label() can be called on a selectable
which contains a Column that is not yet named.
[ticket:1862]

14 years ago- An informative error message is raised if a Column
Mike Bayer [Sat, 18 Sep 2010 17:18:44 +0000 (13:18 -0400)] 
- An informative error message is raised if a Column
which has not yet been assigned a name, i.e. as in
declarative, is used in a context where it is
exported to the columns collection of an enclosing
select() construct, or if any construct involving
that column is compiled before its name is
assigned.  [ticket:1862]

14 years ago- add pk instructions for oracle
Mike Bayer [Sat, 18 Sep 2010 16:38:27 +0000 (12:38 -0400)] 
- add pk instructions for oracle

14 years agoadd oursql to has lastrowid
Mike Bayer [Sat, 18 Sep 2010 16:35:25 +0000 (12:35 -0400)] 
add oursql to has lastrowid

14 years ago- use "key in dict" rather than KeyError if the usual case
Mike Bayer [Sat, 18 Sep 2010 15:40:25 +0000 (11:40 -0400)] 
- use "key in dict" rather than KeyError if the usual case
is that the key is not present.
- don't need to uniquify Index schemes, just don't copy Indexes
that were known to be generated from the index=True flag
- user facing changes go in CHANGES
- Table.c allows string lookup

14 years agod.uh
Chris Withers [Fri, 17 Sep 2010 11:26:49 +0000 (12:26 +0100)] 
d.uh

14 years agomerge
Chris Withers [Thu, 16 Sep 2010 20:11:32 +0000 (21:11 +0100)] 
merge

14 years agoadd warning about dodgy use of tometadata
Chris Withers [Thu, 16 Sep 2010 20:10:17 +0000 (21:10 +0100)] 
add warning about dodgy use of tometadata

14 years agotometadata now copies indexes as well
Chris Withers [Thu, 16 Sep 2010 19:58:15 +0000 (20:58 +0100)] 
tometadata now copies indexes as well

14 years agomake sure kwargs are copied by tometadata
Chris Withers [Thu, 16 Sep 2010 19:03:18 +0000 (20:03 +0100)] 
make sure kwargs are copied by tometadata

14 years ago- Fixed a regression in 0.6.4 whereby the change that
Mike Bayer [Thu, 16 Sep 2010 15:29:01 +0000 (11:29 -0400)] 
- Fixed a regression in 0.6.4 whereby the change that
allowed cursor errors to be raised consistently broke
the result.lastrowid accessor.   Test coverage has
been added for result.lastrowid.   Note that lastrowid
is only supported by Pysqlite and some MySQL drivers,
so isn't super-useful in the general case.

14 years agohey lets pep-8 the setup.py file.
Mike Bayer [Thu, 16 Sep 2010 02:30:31 +0000 (22:30 -0400)] 
hey lets pep-8 the setup.py file.

14 years ago- fix test_single test to use default dialect
Mike Bayer [Thu, 16 Sep 2010 02:20:01 +0000 (22:20 -0400)] 
- fix test_single test to use default dialect
- The exception raised by Session when it is used
subsequent to a subtransaction rollback (which is what
happens when a flush fails in autocommit=False mode) has
now been reworded (this is the "inactive due to a
rollback in a subtransaction" message). In particular,
if the rollback was due to an exception during flush(),
the message states this is the case, and reiterates the
string form of the original exception that occurred
during flush. If the session is closed due to explicit
usage of subtransactions (not very common), the message
just states this is the case.

- The exception raised by Mapper when repeated requests to
its initialization are made after initialization already
failed no longer assumes the "hasattr" case, since
there's other scenarios in which this message gets
emitted, and the message also does not compound onto
itself multiple times - you get the same message for
each attempt at usage. The misnomer "compiles" is being
traded out for "initialize".