]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
14 years ago- move session/mapper events/extension tests into orm/test_events.py
Mike Bayer [Fri, 31 Dec 2010 16:02:56 +0000 (11:02 -0500)] 
- move session/mapper events/extension tests into orm/test_events.py

14 years ago- removes the "on_" prefix.
Mike Bayer [Thu, 30 Dec 2010 19:22:43 +0000 (14:22 -0500)] 
- removes the "on_" prefix.

14 years ago- factor out the dependency on the "on_" name
Mike Bayer [Thu, 30 Dec 2010 19:07:15 +0000 (14:07 -0500)] 
- factor out the dependency on the "on_" name

14 years ago- they don't want "on_". First step, change the naming convention on Events
Mike Bayer [Thu, 30 Dec 2010 18:43:39 +0000 (13:43 -0500)] 
- they don't want "on_".   First step, change the naming convention on Events
so that non-events are just _name.

14 years ago- A little more verbiage to the "primaryjoin" error,
Mike Bayer [Thu, 30 Dec 2010 17:14:32 +0000 (12:14 -0500)] 
- A little more verbiage to the "primaryjoin" error,
in an unusual condition that the join condition
"works" for viewonly but doesn't work for non-viewonly,
and foreign_keys wasn't used - adds "foreign_keys" to
the suggestion.  Also add "foreign_keys" to the
suggestion for the generic "direction" error.

14 years ago- mutable examples now move into sqlalchemy.ext.mutable
Mike Bayer [Wed, 29 Dec 2010 20:04:35 +0000 (15:04 -0500)] 
- mutable examples now move into sqlalchemy.ext.mutable
- streamline interfaces, get Mutable/MutableComposite to be as minimal
in usage as possible
- docs for mutable, warnings regrarding mapper events being global
- move MutableType/mutable=True outwards, move orm tests to its
own module, note in all documentation
- still need more events/tests for correct pickling support of
composites, mutables.  in the case of composites its needed
even without mutation.  see [ticket:2009]

14 years ago- restore mapper.get_property() to use the _props dict. at the moment
Mike Bayer [Wed, 29 Dec 2010 03:23:13 +0000 (22:23 -0500)] 
- restore mapper.get_property() to use the _props dict.   at the moment
synonyms for relationships might just be taken out altogether, since they aren't
documented and are of little use.   a plain proxying descriptor, combined with
attribute-based usage with Query (as opposted to naming it by string)
 can do the same thing more simply.
- add event support to composites, change the model around so that the composite
is generated at the point of load.
- add a recipe for tracking mutations on composites.  will probably make both
of these mutations examples into extensions since they're intricate, should
have a lot of test coverage, and what they need to do is fairly straightforward.
Will use metaclasses so that no extra userland step is needed beyond usage
of the type.

14 years ago- simplified the descriptor system to no longer use the hybrid extension, instead...
Mike Bayer [Tue, 28 Dec 2010 22:50:36 +0000 (17:50 -0500)] 
- simplified the descriptor system to no longer use the hybrid extension, instead presenting
the Proxy(QueryableAttribute) object as the public facing interface.   This simplifies
the descriptor system and will allow straightforward integration with attribute events.

14 years ago- add standalone tests for descriptor instrumentation. these begin to replace
Mike Bayer [Tue, 28 Dec 2010 22:11:01 +0000 (17:11 -0500)] 
- add standalone tests for descriptor instrumentation.   these begin to replace
what we're testing with synonyms and comparables in test_mapper

14 years ago- merge rfc0b9df5d9e0 from 0.6 branch
Mike Bayer [Tue, 28 Dec 2010 20:01:19 +0000 (15:01 -0500)] 
- merge rfc0b9df5d9e0 from 0.6 branch

14 years ago- remove some more 2.4 targets
Mike Bayer [Mon, 27 Dec 2010 19:34:22 +0000 (14:34 -0500)] 
- remove some more 2.4 targets

14 years agopy3k fixes
Mike Bayer [Mon, 27 Dec 2010 19:17:10 +0000 (14:17 -0500)] 
py3k fixes

14 years ago- link to the events we use
Mike Bayer [Sun, 26 Dec 2010 22:32:41 +0000 (17:32 -0500)] 
- link to the events we use

14 years ago- docs for mutable events example
Mike Bayer [Sun, 26 Dec 2010 22:28:56 +0000 (17:28 -0500)] 
- docs for mutable events example

14 years ago- on_expire event, since we are starting to build off of events around
Mike Bayer [Sat, 25 Dec 2010 19:08:03 +0000 (14:08 -0500)] 
- on_expire event, since we are starting to build off of events around
full lifecycle
- composite will use events to do everything we want it to, i.e.
storing the composite in __dict__, invalidating it on change
of any of the columns.
- will reinstate mutability of composites via callable attached
to the composite - but userland code will still need to establish
change event listening on the composite itself, perhaps via
a "mutable" mixin like the scalars.py example, perhaps via
addition of descriptors to the mutable object.

14 years ago- add a new "on mapper configured" event - handy !
Mike Bayer [Sat, 25 Dec 2010 18:22:12 +0000 (13:22 -0500)] 
- add a new "on mapper configured" event - handy !

14 years agobugfix
Mike Bayer [Fri, 24 Dec 2010 16:39:56 +0000 (11:39 -0500)] 
bugfix

14 years ago- more skippage of 2.4
Mike Bayer [Thu, 23 Dec 2010 20:51:41 +0000 (15:51 -0500)] 
- more skippage of 2.4

14 years ago- restore declarative support for "composite"
Mike Bayer [Thu, 23 Dec 2010 20:27:47 +0000 (15:27 -0500)] 
- restore declarative support for "composite"
- add an example of mutable scalars with events

14 years ago- added flag to mark any attribute as "modified"
Mike Bayer [Thu, 23 Dec 2010 19:10:14 +0000 (14:10 -0500)] 
- added flag to mark any attribute as "modified"

14 years ago- slight simplify to state.modified_event()
Mike Bayer [Thu, 23 Dec 2010 18:34:26 +0000 (13:34 -0500)] 
- slight simplify to state.modified_event()

14 years ago- allow cextension version to fall back to non-cextension
Mike Bayer [Thu, 23 Dec 2010 01:36:20 +0000 (20:36 -0500)] 
- allow cextension version to fall back to non-cextension
- start taking out "default" version, 2.4 version, only need these tests
in a relative sense

14 years ago- remove OrderedSet usage from a critical area
Mike Bayer [Thu, 23 Dec 2010 00:07:04 +0000 (19:07 -0500)] 
- remove OrderedSet usage from a critical area

14 years ago- fix beaker caching example to use new ".callable" param on bind
Mike Bayer [Wed, 22 Dec 2010 16:37:07 +0000 (11:37 -0500)] 
- fix beaker caching example to use new ".callable" param on bind

14 years ago- arg callcounts change from 2.6.4 to 2.6.6
Mike Bayer [Wed, 22 Dec 2010 02:52:26 +0000 (21:52 -0500)] 
- arg callcounts change from 2.6.4 to 2.6.6

14 years ago- another heap of inlinings and now I really have to be done with this
Mike Bayer [Wed, 22 Dec 2010 02:37:52 +0000 (21:37 -0500)] 
- another heap of inlinings and now I really have to be done with this

14 years ago- apply pep8 to compiler.py
Mike Bayer [Tue, 21 Dec 2010 21:34:00 +0000 (16:34 -0500)] 
- apply pep8 to compiler.py
- deprecate Compiled.compile() - have __init__ do compilation
if statement is present.

14 years ago- merged the test for [ticket:1997], r1b1f195483aa , into
Mike Bayer [Tue, 21 Dec 2010 21:01:38 +0000 (16:01 -0500)] 
- merged the test for [ticket:1997], r1b1f195483aa , into
0.7.  the fix itself no longer applies.

14 years agotiny inlinings
Mike Bayer [Tue, 21 Dec 2010 16:44:46 +0000 (11:44 -0500)] 
tiny inlinings

14 years ago- refactor expire_attributes into two simpler methods
Mike Bayer [Tue, 21 Dec 2010 16:08:12 +0000 (11:08 -0500)] 
- refactor expire_attributes into two simpler methods

14 years ago- modest inlinings into the MapperProperty.setup/row_processor chain
Mike Bayer [Tue, 21 Dec 2010 01:30:29 +0000 (20:30 -0500)] 
- modest inlinings into the MapperProperty.setup/row_processor chain

14 years agocallcount reduction
Mike Bayer [Tue, 21 Dec 2010 00:00:36 +0000 (19:00 -0500)] 
callcount reduction

14 years agodont need the wrapper here
Mike Bayer [Mon, 20 Dec 2010 22:59:08 +0000 (17:59 -0500)] 
dont need the wrapper here

14 years ago- convert built in AttributExtensions to event listener fns
Mike Bayer [Mon, 20 Dec 2010 22:50:57 +0000 (17:50 -0500)] 
- convert built in AttributExtensions to event listener fns

14 years ago- with composites gone all the get_col_value crap is gone too
Mike Bayer [Mon, 20 Dec 2010 21:41:26 +0000 (16:41 -0500)] 
- with composites gone all the get_col_value crap is gone too

14 years ago- clean it up a bit
Mike Bayer [Mon, 20 Dec 2010 21:21:31 +0000 (16:21 -0500)] 
- clean it up a bit
- don't need __set_composite_values__()
- break the bad news about mutations

14 years ago- move the "descriptor" properties into a separate module.
Mike Bayer [Mon, 20 Dec 2010 21:04:05 +0000 (16:04 -0500)] 
- move the "descriptor" properties into a separate module.

14 years ago- migrate composites to its own test suite, break up tests into individual feature...
Mike Bayer [Mon, 20 Dec 2010 20:18:58 +0000 (15:18 -0500)] 
- migrate composites to its own test suite, break up tests into individual feature tests

14 years ago- crudely, this replaces CompositeProperty's base to be
Mike Bayer [Mon, 20 Dec 2010 18:47:48 +0000 (13:47 -0500)] 
- crudely, this replaces CompositeProperty's base to be
DescriptorProperty.   We have to lose mutability (yikes
composites were using mutable too !).   Also the getter
is not particularly efficient since it recreates the composite
every time, probably want to stick it in __dict__.  also
rewrite the unit tests

14 years agomerge tip
Mike Bayer [Mon, 20 Dec 2010 15:51:49 +0000 (10:51 -0500)] 
merge tip

14 years ago- merge versioning example fix from 0.6
Mike Bayer [Mon, 20 Dec 2010 15:51:24 +0000 (10:51 -0500)] 
- merge versioning example fix from 0.6

14 years agomerge tip
Michael Trier [Mon, 20 Dec 2010 04:17:55 +0000 (23:17 -0500)] 
merge tip

14 years agoRidding the world of a few wasteful imports.
Michael Trier [Mon, 20 Dec 2010 04:16:50 +0000 (23:16 -0500)] 
Ridding the world of a few wasteful imports.

14 years agoskip some len calls
Mike Bayer [Mon, 20 Dec 2010 04:00:19 +0000 (23:00 -0500)] 
skip some len calls

14 years agomore inlines
Mike Bayer [Mon, 20 Dec 2010 03:48:39 +0000 (22:48 -0500)] 
more inlines

14 years agoestablish default for returns_unicode_strings in py3k
Mike Bayer [Mon, 20 Dec 2010 01:47:27 +0000 (20:47 -0500)] 
establish default for returns_unicode_strings in py3k

14 years agocallcounts
Mike Bayer [Mon, 20 Dec 2010 01:14:39 +0000 (20:14 -0500)] 
callcounts

14 years ago- try to make get_all_pending() a little easier to read without adding calls
Mike Bayer [Mon, 20 Dec 2010 01:13:03 +0000 (20:13 -0500)] 
- try to make get_all_pending() a little easier to read without adding calls

14 years ago- merge tip
Mike Bayer [Mon, 20 Dec 2010 00:55:55 +0000 (19:55 -0500)] 
- merge tip

14 years ago- further optimize what get_all_pending() returns to reduce the work of receivers
Mike Bayer [Mon, 20 Dec 2010 00:55:35 +0000 (19:55 -0500)] 
- further optimize what get_all_pending() returns to reduce the work of receivers

14 years agomerge tip
Michael Trier [Mon, 20 Dec 2010 00:38:03 +0000 (19:38 -0500)] 
merge tip

14 years agoAdded NULLS FIRST and NULLS LAST support.
Michael Trier [Mon, 20 Dec 2010 00:25:33 +0000 (19:25 -0500)] 
Added NULLS FIRST and NULLS LAST support.

It's implemented as an extension to the asc() and desc() operators, called
nullsfirst() and nullslast().  [ticket:723]

14 years agotriple check that the dict is empty
Mike Bayer [Mon, 20 Dec 2010 00:13:40 +0000 (19:13 -0500)] 
triple check that the dict is empty

14 years ago- merge tip
Mike Bayer [Sun, 19 Dec 2010 23:53:23 +0000 (18:53 -0500)] 
- merge tip

14 years ago- callcounts
Mike Bayer [Sun, 19 Dec 2010 23:52:46 +0000 (18:52 -0500)] 
- callcounts

14 years ago- use a subset of inspect.getargspec() so that get_func_kwargs()/constructor_copy...
Mike Bayer [Sun, 19 Dec 2010 23:49:26 +0000 (18:49 -0500)] 
- use a subset of inspect.getargspec() so that get_func_kwargs()/constructor_copy() don't take up 20 function calls

14 years agonarrow down cascades in session some more
Mike Bayer [Sun, 19 Dec 2010 22:47:40 +0000 (17:47 -0500)] 
narrow down cascades in session some more

14 years ago- duh, compiled is per dialect
Mike Bayer [Sun, 19 Dec 2010 21:01:40 +0000 (16:01 -0500)] 
- duh, compiled is per dialect

14 years agoAdded vim undo files to the ignore list.
Michael Trier [Sun, 19 Dec 2010 02:31:06 +0000 (21:31 -0500)] 
Added vim undo files to the ignore list.

14 years agotrivial inlinings
Mike Bayer [Sat, 18 Dec 2010 18:05:20 +0000 (13:05 -0500)] 
trivial inlinings

14 years ago- some good inlinings to the whole cascade_iterator() thing.
Mike Bayer [Sat, 18 Dec 2010 17:14:23 +0000 (12:14 -0500)] 
- some good inlinings to the whole cascade_iterator() thing.
cascade_iterator() should probably not yield the "instance" at all
and only deal in states.   30-40K methods taken off the orm2010 test.

14 years agothis innocuous change allows sqlite to work with pypy. sadly, pypy is twice as slow
Mike Bayer [Sat, 18 Dec 2010 00:27:01 +0000 (19:27 -0500)] 
this innocuous change allows sqlite to work with pypy.  sadly, pypy is twice as slow
even if i run the same test script four or five times to prime the jit.

14 years agomerge tip
Mike Bayer [Fri, 17 Dec 2010 19:19:22 +0000 (14:19 -0500)] 
merge tip

14 years ago- use get_all_pending in per_state_flush_actions(), but we'd like to
Mike Bayer [Fri, 17 Dec 2010 18:59:24 +0000 (13:59 -0500)] 
- use get_all_pending in per_state_flush_actions(), but we'd like to
streamline get_history() in any case

14 years ago- callcounts
Mike Bayer [Fri, 17 Dec 2010 02:44:48 +0000 (21:44 -0500)] 
- callcounts

14 years ago- remove the need to use LoadDeferredColumns, LoadLazyAttribute in most cases,
Mike Bayer [Fri, 17 Dec 2010 02:08:26 +0000 (21:08 -0500)] 
- remove the need to use LoadDeferredColumns, LoadLazyAttribute in most cases,
these are back to being part of LoaderStrategy
- simplify attribute.get()
- inline the dict get inside of attribute.__get__()
- revert some memoized attrs from InstanceState which are called in almost
all cases regardless
- inlining

14 years ago- inlinings
Mike Bayer [Thu, 16 Dec 2010 23:46:17 +0000 (18:46 -0500)] 
- inlinings
- in particular, mapper.primary_key is a tuple now

14 years ago- MSSQL - the String/Unicode types, and their counterparts VARCHAR/
Mike Bayer [Thu, 16 Dec 2010 22:52:43 +0000 (17:52 -0500)] 
- MSSQL - the String/Unicode types, and their counterparts VARCHAR/
NVARCHAR, emit "max" as the length when no length is
specified.   This makes it more compatible with Postgresql's
VARCHAR type which is similarly unbounded when no length
specified.

14 years agoarg, fix again
Mike Bayer [Wed, 15 Dec 2010 22:19:48 +0000 (17:19 -0500)] 
arg, fix again

14 years ago- merge rbbd81cb9a341 from 0.6 branch
Mike Bayer [Wed, 15 Dec 2010 21:30:41 +0000 (16:30 -0500)] 
- merge rbbd81cb9a341 from 0.6 branch

14 years ago- fix broken py3k block
Mike Bayer [Wed, 15 Dec 2010 18:25:16 +0000 (13:25 -0500)] 
- fix broken py3k block

14 years agomerge tip
Mike Bayer [Wed, 15 Dec 2010 17:46:21 +0000 (12:46 -0500)] 
merge tip

14 years ago- an approach I like better, remove most adapt() methods and use a generic
Mike Bayer [Wed, 15 Dec 2010 17:44:37 +0000 (12:44 -0500)] 
- an approach I like better, remove most adapt() methods and use a generic
copier
- mssql reflection fix, but this will come in again from the tip merge

14 years ago- merge changelog updates from 0.6
Mike Bayer [Wed, 15 Dec 2010 17:32:06 +0000 (12:32 -0500)] 
- merge changelog updates from 0.6

14 years ago- we shouldn't be raising warnings on non-unicode for String type,
Mike Bayer [Wed, 15 Dec 2010 17:21:55 +0000 (12:21 -0500)] 
- we shouldn't be raising warnings on non-unicode for String type,
even if convert_unicode=True

14 years ago- fix mssql index bug, regression post 0.6.4
Mike Bayer [Wed, 15 Dec 2010 17:16:28 +0000 (12:16 -0500)] 
- fix mssql index bug, regression post 0.6.4
- dont emit unicode warning if _warn_on_bytestring is false, including
for in-python string conversion

14 years ago- fixes for mysql and I'm ready to replace this whole adapt() business with a constru...
Mike Bayer [Wed, 15 Dec 2010 00:24:17 +0000 (19:24 -0500)] 
- fixes for mysql and I'm ready to replace this whole adapt() business with a constructor copy

14 years agosome fixes to get PG on board
Mike Bayer [Wed, 15 Dec 2010 00:04:05 +0000 (19:04 -0500)] 
some fixes to get PG on board

14 years agosome tests, should be OK
Mike Bayer [Tue, 14 Dec 2010 01:23:24 +0000 (20:23 -0500)] 
some tests, should be OK

14 years ago- system to cache the bind/result processors in a dialect-wide registry.
Mike Bayer [Mon, 13 Dec 2010 17:53:56 +0000 (12:53 -0500)] 
- system to cache the bind/result processors in a dialect-wide registry.
its an idea with pointy edges.

14 years ago- callcounts. need to find some way to make this less tedious.
Mike Bayer [Mon, 13 Dec 2010 15:23:59 +0000 (10:23 -0500)] 
- callcounts.  need to find some way to make this less tedious.

14 years agospoke too soon on that.
Mike Bayer [Mon, 13 Dec 2010 15:01:42 +0000 (10:01 -0500)] 
spoke too soon on that.

14 years agomerge tip
Mike Bayer [Mon, 13 Dec 2010 05:16:04 +0000 (00:16 -0500)] 
merge tip

14 years ago- why type.dialect_impl(dialect).bind_processor(dialect), caching just the impl?
Mike Bayer [Mon, 13 Dec 2010 05:15:32 +0000 (00:15 -0500)] 
- why type.dialect_impl(dialect).bind_processor(dialect), caching just the impl?
just call type._cached_bind_processor(dialect), cache the impl *and* the processor function.
same for result sets.
- use plain dict + update for defaultexecutioncontext.execution_options

14 years agoc is undefined, we mean connection here
Philip Jenvey [Mon, 13 Dec 2010 01:36:57 +0000 (17:36 -0800)] 
c is undefined, we mean connection here

14 years ago- another easy win, cache the calc of bind processors in the compiled object
Mike Bayer [Sun, 12 Dec 2010 22:50:25 +0000 (17:50 -0500)] 
- another easy win, cache the calc of bind processors in the compiled object

14 years agocallcounts
Mike Bayer [Sun, 12 Dec 2010 20:35:21 +0000 (15:35 -0500)] 
callcounts

14 years ago- add a perf script to illustrate 0.7 speed improvements
Mike Bayer [Sun, 12 Dec 2010 20:02:22 +0000 (15:02 -0500)] 
- add a perf script to illustrate 0.7 speed improvements

14 years agodont need these imports
Mike Bayer [Sun, 12 Dec 2010 18:43:02 +0000 (13:43 -0500)] 
dont need these imports

14 years ago- inlinings and callcount reductions
Mike Bayer [Sun, 12 Dec 2010 18:01:34 +0000 (13:01 -0500)] 
- inlinings and callcount reductions
- add test coverage for the rare case of noload->lazyload + pickle

14 years ago- refactor query._get() into two methods - a static one that does
Mike Bayer [Sun, 12 Dec 2010 06:35:37 +0000 (01:35 -0500)] 
- refactor query._get() into two methods - a static one that does
just the identity map lookup + expired check, the other
which does the load unconditionally.  All the refresh/deferred load
calls use the unconditional load method, query.get() and
LoadLazyAttribute call the identity check by itself first.   m2o
lazyloads for object already in the identity map callcounts are
now cut in half, since no Query object is created.

14 years agoslight shrinking of the iterable
Mike Bayer [Sun, 12 Dec 2010 04:55:10 +0000 (23:55 -0500)] 
slight shrinking of the iterable

14 years ago- add full series of unit tests for PropertyOption._get_paths, thus enabling us to
Mike Bayer [Sun, 12 Dec 2010 00:50:41 +0000 (19:50 -0500)] 
- add full series of unit tests for PropertyOption._get_paths, thus enabling us to
rewrite it

14 years agobreak up test_query into three modules
Mike Bayer [Sat, 11 Dec 2010 23:21:52 +0000 (18:21 -0500)] 
break up test_query into three modules

14 years agohey we have .. note:: for notes, duh
Mike Bayer [Sat, 11 Dec 2010 23:01:31 +0000 (18:01 -0500)] 
hey we have .. note:: for notes, duh

14 years agocdecimal docs, link to pypi
Mike Bayer [Sat, 11 Dec 2010 22:53:02 +0000 (17:53 -0500)] 
cdecimal docs, link to pypi

14 years ago- support for cdecimal
Mike Bayer [Sat, 11 Dec 2010 22:44:46 +0000 (17:44 -0500)] 
- support for cdecimal
- add --with-cdecimal flag to tests, monkeypatches cdecimal in
- fix mssql/pyodbc.py to not use private '_int' accessor in decimal conversion
routines
- pyodbc version 2.1.8 is needed for cdecimal in any case as
previous versions also called '_int', 2.1.8 adds the same string
logic as our own dialect, so that logic is skipped for modern
pyodbc version
- make the imports for "Decimal" consistent across the whole lib.  not sure
yet how we should be importing "Decimal" or what the best way forward
is that would allow a clean user-invoked swap of cdecimal; for now,
added docs suggesting a global monkeypatch - the two decimal libs
are not compatible with each other so any chance of mixing produces
serious issues.  adding adapters to DBAPIs tedious and adds in-python
overhead.  suggestions welcome on how we should be doing
Decimal/cdecimal.

14 years agodo the mercurial dance (re-merge what I just merged...)
Mike Bayer [Sat, 11 Dec 2010 08:10:17 +0000 (03:10 -0500)] 
do the mercurial dance (re-merge what I just merged...)

14 years ago- clean up the batch insert thing
Mike Bayer [Sat, 11 Dec 2010 08:05:03 +0000 (03:05 -0500)] 
- clean up the batch insert thing
- add a test for batch inserts
- don't need elaborate _inserted_primary_key thing
- take some cruft out of ExecutionContext, ResultProxy,
EC members can be non-underscored, have mapper just call the
EC members for now.
- simplify "connection_callable", no need for a "flush_opts"
dictionary since this point of expansion is not needed

14 years ago[49145a6940062486a6eec66bfe5c9d95c5f76c7a] forgot the final arg to _handle_dbapi_exce...
Philip Jenvey [Sat, 11 Dec 2010 07:46:54 +0000 (23:46 -0800)] 
[49145a6940062486a6eec66bfe5c9d95c5f76c7a] forgot the final arg to _handle_dbapi_exception