Mike Bayer [Sun, 22 Dec 2013 00:51:25 +0000 (19:51 -0500)]
- alter the decorator function, and the newer public_factory function,
to use a named def instead of a lambda. this so that TypeError on wrong
arguments are more legible. [ticket:2884]
Mike Bayer [Fri, 20 Dec 2013 15:26:09 +0000 (10:26 -0500)]
- Fixed issue where a primary key column that has a Sequence on it,
yet the column is not the "auto increment" column, either because
it has a foreign key constraint or ``autoincrement=False`` set,
would attempt to fire the Sequence on INSERT for backends that don't
support sequences, when presented with an INSERT missing the primary
key value. This would take place on non-sequence backends like
SQLite, MySQL. [ticket:2896]
Mike Bayer [Thu, 19 Dec 2013 21:02:14 +0000 (16:02 -0500)]
- Fixed bug with :meth:`.Insert.from_select` method where the order
of the given names would not be taken into account when generating
the INSERT statement, thus producing a mismatch versus the column
names in the given SELECT statement. Also noted that
:meth:`.Insert.from_select` implies that Python-side insert defaults
cannot be used, since the statement has no VALUES clause. [ticket:2895]
Mike Bayer [Wed, 18 Dec 2013 23:26:15 +0000 (18:26 -0500)]
- Improvements to the system by which SQL types generate within
``__repr__()``, particularly with regards to the MySQL integer/numeric/
character types which feature a wide variety of keyword arguments.
The ``__repr__()`` is important for use with Alembic autogenerate
for when Python code is rendered in a migration script.
[ticket:2893]
Mike Bayer [Tue, 17 Dec 2013 22:46:09 +0000 (17:46 -0500)]
- make the json serializer and deserializer per-dialect, so that we are
compatible with psycopg2's per-connection/cursor approach. add round trip tests for
both native and non-native.
Mike Bayer [Tue, 17 Dec 2013 20:38:35 +0000 (15:38 -0500)]
- The :func:`.cast` function, when given a plain literal value,
will now apply the given type to the given literal value on the
bind parameter side according
to the type given to the cast. This essentially replaces what would
normally be the detected type of the literal value. This only
takes effect if the auto-detected type of the literal value is either
"nulltype" (e.g. couldn't detect)
or a type that is of the same "affinity" as the cast type.
The net change here is that the :func:`.cast` function includes more
of the functionality already present in the :func:`.type_coerce` function.
Mike Bayer [Tue, 17 Dec 2013 20:13:39 +0000 (15:13 -0500)]
- rework JSON expressions to be based off __getitem__ exclusively
- add support for "standalone" JSON objects; this involves getting CAST
to upgrade the given type of a bound parameter. should add a core-only test
for this.
- add tests for "standalone" json round trips both with and without unicode
- add mechanism by which we remove psycopg2's "json" handler in order to get
the effect of using our non-native result handlers
Mike Bayer [Tue, 17 Dec 2013 00:32:10 +0000 (19:32 -0500)]
- for [ticket:2651], leaving CheckConstraint alone, preferring to keep
backwards compatibility. A note about backslashing escapes is added.
Because the Text() construct now supports bind params better, the example
given in the code raises an exception now, so that should cover us.
The exception itself has been enhanced to include the key name of the
bound param. We're backporting this to 0.8 but 0.8 doesn't have the
text->bind behavior that raises.
Mike Bayer [Tue, 17 Dec 2013 00:17:41 +0000 (19:17 -0500)]
- An adjustment to the :func:`.subqueryload` strategy which ensures that
the query runs after the loading process has begun; this is so that
the subqueryload takes precedence over other loaders that may be
hitting the same attribute due to other eager/noload situations
at the wrong time. [ticket:2887]
John Anderson [Mon, 16 Dec 2013 20:50:10 +0000 (12:50 -0800)]
Remove terminated connections from the pool.
In pymssql, if you terminate a long running query manually
it will give you a connection reset by peer message, but this
connection remains in the pool and will be re-used.
Mike Bayer [Mon, 16 Dec 2013 20:25:48 +0000 (15:25 -0500)]
- Fixed bug when using joined table inheritance from a table to a
select/alias on the base, where the PK columns were also not same
named; the persistence system would fail to copy primary key values
from the base table to the inherited table upon INSERT.
[ticket:2885]
Mike Bayer [Mon, 16 Dec 2013 02:23:01 +0000 (21:23 -0500)]
- add "force_nocheck" as a way to turn on unicode=force without even
doing the isinstance() check - currently used only by psycopg2 + native enum + py2k.
- didn't realize psycopg2 had UNICODEARRAY extension all this time; replace _PGArray
with just using UNICODEARRAY instead.
- replace unnecessary/inconsistent __import__ in _isolation_lookup.
Mike Bayer [Thu, 12 Dec 2013 18:30:16 +0000 (13:30 -0500)]
- :func:`.composite` will raise an informative error message when the
columns/attribute (names) passed don't resolve to a Column or mapped
attribute (such as an erroneous tuple); previously raised an unbound
local. [ticket:2889]
Mike Bayer [Thu, 12 Dec 2013 17:49:57 +0000 (12:49 -0500)]
- Error message when a string arg sent to :func:`.relationship` which
doesn't resolve to a class or mapper has been corrected to work
the same way as when a non-string arg is received, which indicates
the name of the relationship which had the configurational error.
[ticket:2888]
Mike Bayer [Thu, 12 Dec 2013 00:48:27 +0000 (19:48 -0500)]
- The :class:`.ForeignKey` class more aggressively checks the given
column argument. If not a string, it checks that the object is
at least a :class:`.ColumnClause`, or an object that resolves to one,
and that the ``.table`` attribute, if present, refers to a
:class:`.TableClause` or subclass, and not something like an
:class:`.Alias`. Otherwise, a :class:`.ArgumentError` is raised.
[ticket:2883]
Mike Bayer [Wed, 11 Dec 2013 19:30:18 +0000 (14:30 -0500)]
- The :class:`.exc.StatementError` or DBAPI-related subclass
now can accomodate additional information about the "reason" for
the exception; the :class:`.Session` now adds some detail to it
when the exception occurs within an autoflush. This approach
is taken as opposed to combining :class:`.FlushError` with
a Python 3 style "chained exception" approach so as to maintain
compatibility both with Py2K code as well as code that already
catches ``IntegrityError`` or similar.
nathan [Tue, 10 Dec 2013 15:01:51 +0000 (10:01 -0500)]
sqlalchemy/dialects/postgresql/pgjson:
- Fixed reference to HSTORE
- Corrected spelling of SQLAlchemy
sqlalchemy/dialects/postgresql/psycopg2:
- Added psycopg2 specific wrapper type for JSON which uses inherent json deserialization facilities
- Added code to detect and utilize the JSON wrapper if psycopg2 >= 2.5
test/dialect/postgresql/test_types:
- removed reference to use_native_hstore
nathan [Mon, 9 Dec 2013 16:46:36 +0000 (11:46 -0500)]
sqlalchemy/dialects/postgresql/__init__.py:
- Added import references to JSON class
sqlalchemy/dialects/postgresql/base.py:
- Added visitor method for JSON class
sqlalchemy/dialects/postgresql/pgjson (new):
- JSON class, supports automatic serialization and deserialization of json data, as well as basic json operators.
Mike Bayer [Sat, 7 Dec 2013 23:38:15 +0000 (18:38 -0500)]
- The :func:`.engine_from_config` function has been improved so that
we will be able to parse dialect-specific arguments from string
configuration dictionaries. Dialect classes can now provide their
own list of parameter types and string-conversion routines.
The feature is not yet used by the built-in dialects, however.
[ticket:2875]
Mike Bayer [Sat, 7 Dec 2013 22:20:05 +0000 (17:20 -0500)]
- A DBAPI that raises an error on ``connect()`` which is not a subclass
of dbapi.Error (such as ``TypeError``, ``NotImplementedError``, etc.)
will propagate the exception unchanged. Previously,
the error handling specific to the ``connect()`` routine would both
inappropriately run the exception through the dialect's
:meth:`.Dialect.is_disconnect` routine as well as wrap it in
a :class:`sqlalchemy.exc.DBAPIError`. It is now propagated unchanged
in the same way as occurs within the execute process. [ticket:2881]
- add tests for this in test_parseconnect, but also add tests in test_execute
to ensure the execute() behavior as well
Mike Bayer [Sat, 7 Dec 2013 00:57:19 +0000 (19:57 -0500)]
- The :class:`.QueuePool` has been enhanced to not block new connection
attempts when an existing connection attempt is blocking. Previously,
the production of new connections was serialized within the block
that monitored overflow; the overflow counter is now altered within
it's own critical section outside of the connection process itself.
[ticket:2880]
Mike Bayer [Fri, 6 Dec 2013 20:53:59 +0000 (15:53 -0500)]
- Made a slight adjustment to the logic which waits for a pooled
connection to be available, such that for a connection pool
with no timeout specified, it will every half a second break out of
the wait to check for the so-called "abort" flag, which allows the
waiter to break out in case the whole connection pool was dumped;
normally the waiter should break out due to a notify_all() but it's
possible this notify_all() is missed in very slim cases.
This is an extension of logic first introduced in 0.8.0, and the
issue has only been observed occasionally in stress tests.
Mike Bayer [Fri, 6 Dec 2013 20:53:59 +0000 (15:53 -0500)]
- Made a slight adjustment to the logic which waits for a pooled
connection to be available, such that for a connection pool
with no timeout specified, it will every half a second break out of
the wait to check for the so-called "abort" flag, which allows the
waiter to break out in case the whole connection pool was dumped;
normally the waiter should break out due to a notify_all() but it's
possible this notify_all() is missed in very slim cases.
This is an extension of logic first introduced in 0.8.0, and the
issue has only been observed occasionally in stress tests.
Mike Bayer [Fri, 6 Dec 2013 00:03:31 +0000 (19:03 -0500)]
- The precedence rules for the :meth:`.ColumnOperators.collate` operator
have been modified, such that the COLLATE operator is now of lower
precedence than the comparison operators. This has the effect that
a COLLATE applied to a comparison will not render parenthesis
around the comparison, which is not parsed by backends such as
MSSQL. The change is backwards incompatible for those setups that
were working around the issue by applying :meth:`.Operators.collate`
to an individual element of the comparison expression,
rather than the comparison expression as a whole. [ticket:2879]
projects / thirdparty / sqlalchemy / sqlalchemy.git / log
