]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
11 years ago- default dialect, plus test against the default dialect...
Mike Bayer [Tue, 2 Sep 2014 19:04:12 +0000 (15:04 -0400)] 
- default dialect, plus test against the default dialect...

11 years ago- An adjustment to table/index reflection such that if an index
Mike Bayer [Tue, 2 Sep 2014 18:18:09 +0000 (14:18 -0400)] 
- An adjustment to table/index reflection such that if an index
reports a column that isn't found to be present in the table,
a warning is emitted and the column is skipped.  This can occur
for some special system column situations as has been observed
with Oracle. fixes #3180

11 years agodefault dialect
Mike Bayer [Tue, 2 Sep 2014 17:39:16 +0000 (13:39 -0400)] 
default dialect

11 years ago- add logic to compiler such that if stack is empty, we just
Mike Bayer [Tue, 2 Sep 2014 14:49:46 +0000 (10:49 -0400)] 
- add logic to compiler such that if stack is empty, we just
stringify a _label_reference() as is.
- add .key to _label_reference(), so that when _make_proxy()
is called, we don't call str() on it anyway.
- add a test to exercise Query's behavior of adding all the order_by
expressions to the columns list of the select, assert that things
work out when we have a _label_reference there, that it gets sucked
into the columns list and spit out on the other side, it's referred
to appropriately, etc.   _label_reference() could theoretically
be resolved at the point we iterate _raw_columns() but
it's better to just let things work as they already do (except
nicer, since we get "tablename.colname" instead of just "somename"
 in the columns list) so that we aren't adding a ton of overhead
to _columns_plus_names in the common case.

11 years ago- reorganize
Mike Bayer [Tue, 2 Sep 2014 00:31:00 +0000 (20:31 -0400)] 
- reorganize

11 years ago- The :func:`~.expression.column` and :func:`~.expression.table`
Mike Bayer [Tue, 2 Sep 2014 00:19:54 +0000 (20:19 -0400)] 
- The :func:`~.expression.column` and :func:`~.expression.table`
constructs are now importable from the "from sqlalchemy" namespace,
just like every other Core construct.
- The implicit conversion of strings to :func:`.text` constructs
when passed to most builder methods of :func:`.select` as
well as :class:`.Query` now emits a warning with just the
plain string sent.   The textual conversion still proceeds normally,
however.  The only method that accepts a string without a warning
are the "label reference" methods like order_by(), group_by();
these functions will now at compile time attempt to resolve a single
string argument to a column or label expression present in the
selectable; if none is located, the expression still renders, but
you get the warning again. The rationale here is that the implicit
conversion from string to text is more unexpected than not these days,
and it is better that the user send more direction to the Core / ORM
when passing a raw string as to what direction should be taken.
Core/ORM tutorials have been updated to go more in depth as to how text
is handled.
fixes #2992

11 years ago- more updates to text docs, literal_column, column etc. in prep
Mike Bayer [Mon, 1 Sep 2014 21:20:49 +0000 (17:20 -0400)] 
- more updates to text docs, literal_column, column etc.  in prep
for ticket 2992.

11 years ago- walk back these literal SQL lectures into something much more succinct.
Mike Bayer [Mon, 1 Sep 2014 00:26:40 +0000 (20:26 -0400)] 
- walk back these literal SQL lectures into something much more succinct.
the ORM one in particular was really long winded and I don't really care if people
use text() anyway, they'll figure it out ;)

11 years ago- improve from_statement() doc
Mike Bayer [Sun, 31 Aug 2014 22:05:17 +0000 (18:05 -0400)] 
- improve from_statement() doc

11 years ago- optional 'b' here for py3k
Mike Bayer [Sun, 31 Aug 2014 22:02:24 +0000 (18:02 -0400)] 
- optional 'b' here for py3k

11 years ago- refine this a bit to better check for exception type
Mike Bayer [Sun, 31 Aug 2014 22:00:49 +0000 (18:00 -0400)] 
- refine this a bit to better check for exception type

11 years ago- these tests were squashing deprecation warnings all along...
Mike Bayer [Sun, 31 Aug 2014 22:00:34 +0000 (18:00 -0400)] 
- these tests were squashing deprecation warnings all along...

11 years ago- A new style of warning can be emitted which will "filter" up to
Mike Bayer [Sun, 31 Aug 2014 19:22:00 +0000 (15:22 -0400)] 
- A new style of warning can be emitted which will "filter" up to
N occurrences of a parameterized string.   This allows parameterized
warnings that can refer to their arguments to be delivered a fixed
number of times until allowing Python warning filters to squelch them,
and prevents memory from growing unbounded within Python's
warning registries.
fixes #3178

11 years ago- updated callcounts
Mike Bayer [Sun, 31 Aug 2014 19:21:42 +0000 (15:21 -0400)] 
- updated callcounts

11 years ago- start encouraging the use of text() for injection of string-based SQL
Mike Bayer [Sun, 31 Aug 2014 15:44:51 +0000 (11:44 -0400)] 
- start encouraging the use of text() for injection of string-based SQL
rather than straight strings.  reference #2992

11 years ago- alter the yield_per eager restriction such that joined many-to-one loads
Mike Bayer [Sat, 30 Aug 2014 19:45:50 +0000 (15:45 -0400)] 
- alter the yield_per eager restriction such that joined many-to-one loads
are still OK, since these should be fine.

11 years ago- continue moving things out that don't need to be there
Mike Bayer [Sat, 30 Aug 2014 05:52:36 +0000 (01:52 -0400)] 
- continue moving things out that don't need to be there
- an existing state shouldn't need its load_options/load_path updated;
it should maintain those from its original Query source.  there's no
tests that check this behavior

11 years ago- clean up zoomark a little and try to get new profiles written
Mike Bayer [Sat, 30 Aug 2014 04:11:32 +0000 (00:11 -0400)] 
- clean up zoomark a little and try to get new profiles written

11 years ago- do the polymorphic thing as a decorator so it's out of the way otherwise
Mike Bayer [Sat, 30 Aug 2014 00:10:54 +0000 (20:10 -0400)] 
- do the polymorphic thing as a decorator so it's out of the way otherwise

11 years ago- bump callcounts down
Mike Bayer [Fri, 29 Aug 2014 23:16:29 +0000 (19:16 -0400)] 
- bump callcounts down

11 years ago- pull out populators back into separate functions, though still very inlined
Mike Bayer [Fri, 29 Aug 2014 21:46:57 +0000 (17:46 -0400)] 
- pull out populators back into separate functions, though still very inlined

11 years ago- further move things vertically, at which point things are inlined enough
Mike Bayer [Fri, 29 Aug 2014 20:57:12 +0000 (16:57 -0400)] 
- further move things vertically, at which point things are inlined enough
that I'd like to start de-inlining again in the hopes of making this readable.

11 years ago- need to use safe_discard() at least in _restore_snapshot(), let's use it
Mike Bayer [Fri, 29 Aug 2014 20:32:08 +0000 (16:32 -0400)] 
- need to use safe_discard() at least in _restore_snapshot(), let's use it
everywhere in Session since the optimized one only applies to loading

11 years ago- use default dialect for compile test here
Mike Bayer [Fri, 29 Aug 2014 20:29:46 +0000 (16:29 -0400)] 
- use default dialect for compile test here

11 years ago- defaultdict benchmarks faster than a namedtuple; OK
Mike Bayer [Fri, 29 Aug 2014 20:28:19 +0000 (16:28 -0400)] 
- defaultdict benchmarks faster than a namedtuple; OK
- inline the column-based expiration operations as well

11 years ago- reorganize how create_row_processor() communicates up to
Mike Bayer [Fri, 29 Aug 2014 19:07:11 +0000 (15:07 -0400)] 
- reorganize how create_row_processor() communicates up to
instances(), using a named tuple it can assign to directly.  this way
we never have to worry about that structure changing anymore, though
we are still having it append (key, fn) which is kind of awkward.
- inline _populators() into instance(), it's a little verbose but
saves an fn call

11 years ago- re-establish and test some behavior from previous versions, that
Mike Bayer [Fri, 29 Aug 2014 18:25:09 +0000 (14:25 -0400)] 
- re-establish and test some behavior from previous versions, that
if a load() or refresh() event changes history (which...why...but anyway)
the state of the object is the same; currently it seems that history
gets reset but on a refresh, the object still goes into session.dirty
- simplify what we store in partials

11 years ago- The :class:`.Query` will raise an exception when :meth:`.Query.yield_per`
Mike Bayer [Fri, 29 Aug 2014 18:01:38 +0000 (14:01 -0400)] 
- The :class:`.Query` will raise an exception when :meth:`.Query.yield_per`
is used with mappings or options where eager loading, either
joined or subquery, would take place.  These loading strategies are
not currently compatible with yield_per, so by raising this error,
the method is safer to use - combine with sending False to
:meth:`.Query.enable_eagerloads` to disable the eager loaders.

11 years ago- we've got 25% improvement so let's talk about it
Mike Bayer [Fri, 29 Aug 2014 17:35:25 +0000 (13:35 -0400)] 
- we've got 25% improvement so let's talk about it

11 years ago- add some more docs to yield_per
Mike Bayer [Fri, 29 Aug 2014 17:35:03 +0000 (13:35 -0400)] 
- add some more docs to yield_per

11 years ago- use a faster discard when loading
Mike Bayer [Fri, 29 Aug 2014 17:11:52 +0000 (13:11 -0400)] 
- use a faster discard when loading
- don't do a bool on identity map since it calls __len__

11 years ago- inline the commit of partials too
Mike Bayer [Fri, 29 Aug 2014 16:49:06 +0000 (12:49 -0400)] 
- inline the commit of partials too

11 years ago- Changed the approach by which the "single inheritance criterion"
Mike Bayer [Fri, 29 Aug 2014 16:05:00 +0000 (12:05 -0400)] 
- Changed the approach by which the "single inheritance criterion"
is applied, when using :meth:`.Query.from_self`, or its common
user :meth:`.Query.count`.  The criteria to limit rows to those
with a certain type is now indicated on the inside subquery,
not the outside one, so that even if the "type" column is not
available in the columns clause, we can filter on it on the "inner"
query.
fixes #3177

11 years ago- dropping callcounts
Mike Bayer [Fri, 29 Aug 2014 15:22:54 +0000 (11:22 -0400)] 
- dropping callcounts

11 years agoinlines galore
Mike Bayer [Fri, 29 Aug 2014 14:59:13 +0000 (10:59 -0400)] 
inlines galore

11 years ago- major refactoring/inlining to loader.instances(), though not really
Mike Bayer [Fri, 29 Aug 2014 00:06:12 +0000 (20:06 -0400)] 
- major refactoring/inlining to loader.instances(), though not really
any speed improvements :(.   code is in a much better place to be run into
C, however
- The ``proc()`` callable passed to the ``create_row_processor()``
method of custom :class:`.Bundle` classes now accepts only a single
"row" argument.
- Deprecated event hooks removed:  ``populate_instance``,
``create_instance``, ``translate_row``, ``append_result``
- the getter() idea is somewhat restored; see ref #3175

11 years ago- Made a small adjustment to the mechanics of lazy loading,
Mike Bayer [Thu, 28 Aug 2014 21:57:48 +0000 (17:57 -0400)] 
- Made a small adjustment to the mechanics of lazy loading,
such that it has less chance of interfering with a joinload() in the
very rare circumstance that an object points to itself; in this
scenario, the object refers to itself while loading its attributes
which can cause a mixup between loaders.   The use case of
"object points to itself" is not fully supported, but the fix also
removes some overhead so for now is part of testing.
fixes #3145

11 years ago- flake8
Mike Bayer [Thu, 28 Aug 2014 21:42:41 +0000 (17:42 -0400)] 
- flake8

11 years ago- this is small optimization, currently it's the best we can do
Mike Bayer [Thu, 28 Aug 2014 16:41:50 +0000 (12:41 -0400)] 
- this is small optimization, currently it's the best we can do
for #3175.  fixes #3175 (for now)

11 years ago- A new implementation for :class:`.KeyedTuple` used by the
Mike Bayer [Thu, 28 Aug 2014 16:25:21 +0000 (12:25 -0400)] 
- A new implementation for :class:`.KeyedTuple` used by the
:class:`.Query` object offers dramatic speed improvements when
fetching large numbers of column-oriented rows.
fixes #3176

11 years ago- The behavior of :paramref:`.joinedload.innerjoin` as well as
Mike Bayer [Tue, 26 Aug 2014 21:23:23 +0000 (17:23 -0400)] 
- The behavior of :paramref:`.joinedload.innerjoin` as well as
:paramref:`.relationship.innerjoin` is now to use "nested"
inner joins, that is, right-nested, as the default behavior when an
inner join joined eager load is chained to an outer join eager load.
fixes #3008

11 years ago- hard-won pep8 formatting of one of the giants...
Mike Bayer [Tue, 26 Aug 2014 20:34:41 +0000 (16:34 -0400)] 
- hard-won pep8 formatting of one of the giants...

11 years ago- updates to migration / changelog for 1.0
Mike Bayer [Tue, 26 Aug 2014 18:44:50 +0000 (14:44 -0400)] 
- updates to migration / changelog for 1.0

11 years ago- changelog for pr bitbucket:28
Mike Bayer [Mon, 25 Aug 2014 23:14:47 +0000 (19:14 -0400)] 
- changelog for pr bitbucket:28

11 years agoMerge branch 'mutable-dict-update' of https://bitbucket.org/goodscloud/sqlalchemy...
Mike Bayer [Mon, 25 Aug 2014 23:13:47 +0000 (19:13 -0400)] 
Merge branch 'mutable-dict-update' of https://bitbucket.org/goodscloud/sqlalchemy into pr27

11 years ago- changelog for pr bitbucket:27
Mike Bayer [Mon, 25 Aug 2014 23:13:25 +0000 (19:13 -0400)] 
- changelog for pr bitbucket:27

11 years agoMerge branch 'mutable-dict-coerce-fix' of https://bitbucket.org/goodscloud/sqlalchemy...
Mike Bayer [Mon, 25 Aug 2014 23:10:43 +0000 (19:10 -0400)] 
Merge branch 'mutable-dict-coerce-fix' of https://bitbucket.org/goodscloud/sqlalchemy into pr27

11 years ago- mention that FOUND_ROWS is hardcoded; fixes #3146
Mike Bayer [Mon, 25 Aug 2014 21:06:28 +0000 (17:06 -0400)] 
- mention that FOUND_ROWS is hardcoded; fixes #3146

11 years ago- The "resurrect" ORM event has been removed. This event hook had
Mike Bayer [Mon, 25 Aug 2014 21:00:21 +0000 (17:00 -0400)] 
- The "resurrect" ORM event has been removed.  This event hook had
no purpose since the old "mutable attribute" system was removed
in 0.8.
fixes #3171

11 years agoMerge branch 'pr129'
Mike Bayer [Sat, 23 Aug 2014 19:21:54 +0000 (15:21 -0400)] 
Merge branch 'pr129'

Conflicts:
doc/build/changelog/changelog_10.rst

11 years ago- pep8 formatting for pg table opts feature, tests
Mike Bayer [Sat, 23 Aug 2014 19:21:16 +0000 (15:21 -0400)] 
- pep8 formatting for pg table opts feature, tests
- add support for PG INHERITS
- fix mis-named tests
- changelog
fixes #2051

11 years agoMerge remote-tracking branch 'origin/pr/131'
Mike Bayer [Thu, 21 Aug 2014 21:38:23 +0000 (17:38 -0400)] 
Merge remote-tracking branch 'origin/pr/131'

11 years agoMerge pull request #130 from gthb/docfix-1
mike bayer [Thu, 21 Aug 2014 21:33:26 +0000 (17:33 -0400)] 
Merge pull request #130 from gthb/docfix-1

Fix doc typos and copy-paste error

11 years agoAdd note on begin_nested requiring rollback/commit
Gunnlaugur Þór Briem [Thu, 21 Aug 2014 18:42:08 +0000 (18:42 +0000)] 
Add note on begin_nested requiring rollback/commit

Avoid confusion about rollback/commit "must be issued" after
``session.begin_nested()`` --- this might be taken to mean call must be
*added*, but that's only true if not using the return value as a context
manager.

11 years agoFix doc typo 'conjunection'
Gunnlaugur Þór Briem [Thu, 31 Jul 2014 23:26:18 +0000 (23:26 +0000)] 
Fix doc typo 'conjunection'

11 years agoFix copy-paste error in Delete doc
Gunnlaugur Þór Briem [Thu, 31 Jul 2014 23:23:56 +0000 (23:23 +0000)] 
Fix copy-paste error in Delete doc

11 years ago- fix link
Mike Bayer [Thu, 21 Aug 2014 14:29:21 +0000 (10:29 -0400)] 
- fix link

11 years ago- The INSERT...FROM SELECT construct now implies ``inline=True``
Mike Bayer [Thu, 21 Aug 2014 00:14:20 +0000 (20:14 -0400)] 
- The INSERT...FROM SELECT construct now implies ``inline=True``
on :class:`.Insert`.  This helps to fix a bug where an
INSERT...FROM SELECT construct would inadvertently be compiled
as "implicit returning" on supporting backends, which would
cause breakage in the case of an INSERT that inserts zero rows
(as implicit returning expects a row), as well as arbitrary
return data in the case of an INSERT that inserts multiple
rows (e.g. only the first row of many).
A similar change is also applied to an INSERT..VALUES
with multiple parameter sets; implicit RETURNING will no longer emit
for this statement either.  As both of these constructs deal
with varible numbers of rows, the
:attr:`.ResultProxy.inserted_primary_key` accessor does not
apply.   Previously, there was a documentation note that one
may prefer ``inline=True`` with INSERT..FROM SELECT as some databases
don't support returning and therefore can't do "implicit" returning,
but there's no reason an INSERT...FROM SELECT needs implicit returning
in any case.   Regular explicit :meth:`.Insert.returning` should
be used to return variable numbers of result rows if inserted
data is needed.
fixes #3169

11 years ago- pep8
Mike Bayer [Wed, 20 Aug 2014 23:12:32 +0000 (19:12 -0400)] 
- pep8

11 years ago- factor out determination of current version id out of
Mike Bayer [Wed, 20 Aug 2014 18:24:45 +0000 (14:24 -0400)] 
- factor out determination of current version id out of
_collect_update_commands and _collect_delete_commands

11 years ago- Fixed bug in connection pool logging where the "connection checked out"
Mike Bayer [Wed, 20 Aug 2014 16:01:20 +0000 (12:01 -0400)] 
- Fixed bug in connection pool logging where the "connection checked out"
debug logging message would not emit if the logging were set up using
``logging.setLevel()``, rather than using the ``echo_pool`` flag.
Tests to assert this logging have been added.  This is a
regression that was introduced in 0.9.0.
fixes #3168

11 years ago- pep8 cleanup
Mike Bayer [Wed, 20 Aug 2014 13:08:59 +0000 (09:08 -0400)] 
- pep8 cleanup

11 years ago- simplify PK logic in update for row switch
Mike Bayer [Tue, 19 Aug 2014 20:14:57 +0000 (16:14 -0400)] 
- simplify PK logic in update for row switch

11 years ago- optimize collection of cols we insert as none
Mike Bayer [Mon, 18 Aug 2014 21:12:06 +0000 (17:12 -0400)] 
- optimize collection of cols we insert as none

11 years ago- move out checks for table in mapper._pks_by_table
Mike Bayer [Mon, 18 Aug 2014 21:02:52 +0000 (17:02 -0400)] 
- move out checks for table in mapper._pks_by_table

11 years ago- further reorganize collect_insert_commands to distinguish between
Mike Bayer [Mon, 18 Aug 2014 20:44:07 +0000 (16:44 -0400)] 
- further reorganize collect_insert_commands to distinguish between
setting up given values vs. defaults.  again trying to shoot for
making this of more general use

11 years ago- organize persistence methods in terms of generators,
Mike Bayer [Mon, 18 Aug 2014 20:32:48 +0000 (16:32 -0400)] 
- organize persistence methods in terms of generators,
narrow down argument lists and generator items for each function
down to just what each function needs.   This will help for them
to be of more multipurpose use for bulk operations

11 years ago- major simplification of _collect_update_commands. in particular,
Mike Bayer [Mon, 18 Aug 2014 16:50:29 +0000 (12:50 -0400)] 
- major simplification of _collect_update_commands.  in particular,
we only call upon the history API fully for primary key columns.
We also now skip the whole step of looking at PK columns and using
any history at all if no net changes are detected on the object.

11 years ago- Fixed bug where attribute "set" events or columns with
Mike Bayer [Mon, 18 Aug 2014 00:06:16 +0000 (20:06 -0400)] 
- Fixed bug where attribute "set" events or columns with
``@validates`` would have events triggered within the flush process,
when those columns were the targets of a "fetch and populate"
operation, such as an autoincremented primary key, a Python side
default, or a server-side default "eagerly" fetched via RETURNING.
fixes #3167

11 years agoAdding postgres create table options documentation
Malik Diarra [Sun, 17 Aug 2014 14:56:53 +0000 (16:56 +0200)] 
Adding postgres create table options documentation

11 years agoCorrecting options name from withoids to with_oids
Malik Diarra [Sun, 17 Aug 2014 00:48:21 +0000 (02:48 +0200)] 
Correcting options name from withoids to with_oids

11 years ago- oldest screwup in the book, forgot the file
Mike Bayer [Sun, 17 Aug 2014 00:17:08 +0000 (20:17 -0400)] 
- oldest screwup in the book, forgot the file

11 years agoquoting tablespace name in create table command in postgresql dialect
Malik Diarra [Sat, 16 Aug 2014 23:39:14 +0000 (01:39 +0200)] 
quoting tablespace name in create table command in postgresql dialect

11 years agoAdding oids and on_commit table options
Malik Diarra [Sat, 16 Aug 2014 22:39:36 +0000 (00:39 +0200)] 
Adding oids and on_commit table options

11 years agoAdding a tablespace options for postgresql create table
Malik Diarra [Tue, 12 Aug 2014 23:39:09 +0000 (01:39 +0200)] 
Adding a tablespace options for postgresql create table

11 years ago- rework profiling, zoomark tests into single tests so that
Mike Bayer [Sat, 16 Aug 2014 23:49:07 +0000 (19:49 -0400)] 
- rework profiling, zoomark tests into single tests so that
they can be used under xdist

11 years ago- changelog for pullreq github:125
Mike Bayer [Sat, 16 Aug 2014 17:57:46 +0000 (13:57 -0400)] 
- changelog for pullreq github:125
- add pg8000 version detection for the "sane multi rowcount" feature

11 years agoMerge remote-tracking branch 'origin/pr/125' into pr125
Mike Bayer [Sat, 16 Aug 2014 17:48:57 +0000 (13:48 -0400)] 
Merge remote-tracking branch 'origin/pr/125' into pr125

11 years ago- support dialects w/o sane multi row count again
Mike Bayer [Sat, 16 Aug 2014 17:46:15 +0000 (13:46 -0400)] 
- support dialects w/o sane multi row count again

11 years ago- need list() here for py3k
Mike Bayer [Sat, 16 Aug 2014 17:37:49 +0000 (13:37 -0400)] 
- need list() here for py3k

11 years ago- max failures 25
Mike Bayer [Sat, 16 Aug 2014 17:33:02 +0000 (13:33 -0400)] 
- max failures 25
- guard against some potential pytest snarkiness

11 years ago- mutablemapping adds compiler overhead, so screw it
Mike Bayer [Fri, 15 Aug 2014 22:39:26 +0000 (18:39 -0400)] 
- mutablemapping adds compiler overhead, so screw it

11 years ago- port the _collect_insert_commands optimizations from ticket_3100
Mike Bayer [Fri, 15 Aug 2014 22:33:42 +0000 (18:33 -0400)] 
- port the _collect_insert_commands optimizations from ticket_3100

11 years ago- The :class:`.IdentityMap` exposed from :class:`.Session.identity`
Mike Bayer [Fri, 15 Aug 2014 19:13:13 +0000 (15:13 -0400)] 
- The :class:`.IdentityMap` exposed from :class:`.Session.identity`
now returns lists for ``items()`` and ``values()`` in Py3K.
Early porting to Py3K here had these returning iterators, when
they technically should be "iterable views"..for now, lists are OK.

11 years ago- TIL that dict.keys() in py3K is not an iterator, it is an iterable
Mike Bayer [Fri, 15 Aug 2014 18:57:29 +0000 (14:57 -0400)] 
- TIL that dict.keys() in py3K is not an iterator, it is an iterable
view.  So copy collections.OrderedDict and use MutableMapping to set up
keys, items, values on our own OrderedDict.

Conflicts:
lib/sqlalchemy/engine/base.py

11 years ago- clean up provision and keep sqlite on memory DBs if thats what we start with
Mike Bayer [Fri, 15 Aug 2014 18:38:33 +0000 (14:38 -0400)] 
- clean up provision and keep sqlite on memory DBs if thats what we start with

11 years ago- don't add the parent attach event within _on_table_attach
Mike Bayer [Fri, 15 Aug 2014 18:27:12 +0000 (14:27 -0400)] 
- don't add the parent attach event within _on_table_attach
if we already have a table; this prevents reentrant calls and
we aren't supporting columns/etc being moved around between different parents

11 years ago- other test fixes
Mike Bayer [Fri, 15 Aug 2014 04:19:57 +0000 (00:19 -0400)] 
- other test fixes

11 years ago- modify how class state is tracked here as it seems like things
Mike Bayer [Fri, 15 Aug 2014 04:19:32 +0000 (00:19 -0400)] 
- modify how class state is tracked here as it seems like things
are a little more crazy under xdist mode

11 years ago- UPDATE statements can now be batched within an ORM flush
Mike Bayer [Fri, 15 Aug 2014 00:47:49 +0000 (20:47 -0400)] 
- UPDATE statements can now be batched within an ORM flush
into more performant executemany() call, similarly to how INSERT
statements can be batched; this will be invoked within flush
to the degree that subsequent UPDATE statements for the
same mapping and table involve the identical columns within the
VALUES clause, as well as that no VALUES-level SQL expressions
are embedded.
- some other inlinings within persistence.py

11 years agopep8
Mike Bayer [Fri, 15 Aug 2014 00:37:28 +0000 (20:37 -0400)] 
pep8

11 years agopep8
Mike Bayer [Fri, 15 Aug 2014 00:01:59 +0000 (20:01 -0400)] 
pep8

11 years ago- The string keys that are used to determine the columns impacted
Mike Bayer [Fri, 15 Aug 2014 00:00:35 +0000 (20:00 -0400)] 
- The string keys that are used to determine the columns impacted
for an INSERT or UPDATE are now sorted when they contribute towards
the "compiled cache" cache key.   These keys were previously not
deterministically ordered, meaning the same statement could be
cached multiple times on equivalent keys, costing both in terms of
memory as well as performance.
fixes #3165

11 years ago- Removing (or adding) an event listener at the same time that the event
Mike Bayer [Thu, 14 Aug 2014 18:40:28 +0000 (14:40 -0400)] 
- Removing (or adding) an event listener at the same time that the event
is being run itself, either from inside the listener or from a
concurrent thread, now raises a RuntimeError, as the collection used is
now an instance of ``colletions.deque()`` and does not support changes
while being iterated.  Previously, a plain Python list was used where
removal from inside the event itself would produce silent failures.
fixes #3163

11 years agopep8
Mike Bayer [Thu, 14 Aug 2014 18:21:03 +0000 (14:21 -0400)] 
pep8

11 years agoon second thought we need to prioritize what really needs to be
Mike Bayer [Thu, 14 Aug 2014 18:15:46 +0000 (14:15 -0400)] 
on second thought we need to prioritize what really needs to be
here and what's just in changelog

11 years ago- updates
Mike Bayer [Thu, 14 Aug 2014 17:25:09 +0000 (13:25 -0400)] 
- updates

11 years ago- repair against use here
Mike Bayer [Thu, 14 Aug 2014 04:03:03 +0000 (00:03 -0400)] 
- repair against use here

11 years ago- The ``info`` parameter has been added to the constructor for
Mike Bayer [Wed, 13 Aug 2014 23:45:34 +0000 (19:45 -0400)] 
- The ``info`` parameter has been added to the constructor for
:class:`.SynonymProperty` and :class:`.ComparableProperty`.
- The ``info`` parameter has been added as a constructor argument
to all schema constructs including :class:`.MetaData`,
:class:`.Index`, :class:`.ForeignKey`, :class:`.ForeignKeyConstraint`,
:class:`.UniqueConstraint`, :class:`.PrimaryKeyConstraint`,
:class:`.CheckConstraint`.

fixes #2963

11 years ago- The :meth:`.InspectionAttr.info` collection is now moved down to
Mike Bayer [Wed, 13 Aug 2014 23:20:44 +0000 (19:20 -0400)] 
- The :meth:`.InspectionAttr.info` collection is now moved down to
:class:`.InspectionAttr`, where in addition to being available
on all :class:`.MapperProperty` objects, it is also now available
on hybrid properties, association proxies, when accessed via
:attr:`.Mapper.all_orm_descriptors`.
fixes #2971