Mike Bayer [Mon, 22 Feb 2016 03:46:31 +0000 (22:46 -0500)]
- additional adjustment to the fix made in 8ad968f33100baeb3b13c7e0b724b6b79ab4277f
for ref #3657. The Oracle dialect makes more use of the "select_wraps_for"
feature than SQL server because Oracle doesn't have "TOP" for a limit-only
select, so tests are showing more happening here. In the case where
the select() has some dupe columns, these are deduped from the .c collection
so a positional match between the wrapper and original can't use .inner_columns,
because these collections wont match. Using _columns_plus_names
instead which is the deduped collection that determines the SELECT display,
which definitely have to match up.
Mike Bayer [Mon, 22 Feb 2016 01:52:43 +0000 (20:52 -0500)]
- Fixed bug in "to_list" conversion where a single bytes object
would be turned into a list of individual characters. This would
impact among other things using the :meth:`.Query.get` method
on a primary key that's a bytes object.
fixes #3660
Mike Bayer [Sun, 21 Feb 2016 01:22:38 +0000 (20:22 -0500)]
- reworked the way the "select_wraps_for" expression is
handled within visit_select(); this attribute was added in the
1.0 series to accommodate the subquery wrapping behavior of
SQL Server and Oracle while also working with positional
column targeting and no longer relying upon "key fallback"
in order to target columns in such a statement. The IBM DB2
third-party dialect also has this use case, but its implementation
is using regular expressions to rewrite the textual SELECT only
and does not make use of a "wrapped" select at this time.
The logic no longer attempts to reconcile proxy set collections as
this was not deterministic, and instead assumes that the select()
and the wrapper select() match their columns postionally,
at least for the column positions they have in common,
so it is now very simple and safe. fixes #3657.
- as a side effect of #3657 it was also revealed that the
strategy of calling upon a ResultProxy._getter was not
correctly calling into NoSuchColumnError when an expected
column was not present, and instead returned None up to
loading.instances() to produce NoneType failures; added
a raiseerr argument to _getter() which is called when we
aren't expecting None, fixes #3658.
Mike Bayer [Wed, 17 Feb 2016 22:54:43 +0000 (17:54 -0500)]
- Fixed bug which would cause an eagerly loaded many-to-one attribute
to not be loaded, if the joined eager load were from a row where the
same entity were present multiple times, some calling for the attribute
to be eagerly loaded and others not. The logic here is revised to
take in the attribute even though a different loader path has
handled the parent entity already. fixes #3431
Mike Bayer [Wed, 17 Feb 2016 21:53:01 +0000 (16:53 -0500)]
- handle parameter sets that aren't correctly formed, so that
for example an exception object made within a test suite can
still repr (error seen in Keystone)
Mike Bayer [Wed, 17 Feb 2016 18:31:29 +0000 (13:31 -0500)]
- All string formatting of bound parameter sets and result rows for
logging, exception, and ``repr()`` purposes now truncate very large
scalar values within each collection, including an
"N characters truncated"
notation, similar to how the display for large multiple-parameter sets
are themselves truncated.
fixes #2837
Mike Bayer [Fri, 12 Feb 2016 03:29:18 +0000 (22:29 -0500)]
- Fixed bug in :meth:`.Session.merge` where an object with a composite
primary key that has values for some but not all of the PK fields
would emit a SELECT statement leaking the internal NEVER_SET symbol
into the query, rather than detecting that this object does not have
a searchable primary key and no SELECT should be emitted.
fixes #3647
Mike Bayer [Thu, 11 Feb 2016 17:12:19 +0000 (12:12 -0500)]
- CTE functionality has been expanded to support all DML, allowing
INSERT, UPDATE, and DELETE statements to both specify their own
WITH clause, as well as for these statements themselves to be
CTE expressions when they include a RETURNING clause.
fixes #2551
Mike Bayer [Wed, 10 Feb 2016 15:17:45 +0000 (10:17 -0500)]
- add more documentation to MutableDict explaining
that this structure is only intended to track additions
and removals from the dictionary, not recursive tracking
of embedded changes. fixes #3646.
Mike Bayer [Wed, 10 Feb 2016 02:08:21 +0000 (21:08 -0500)]
- don't drop the ts1/ts2 databases without seeing the primary DB,
because we never log in on the ts1/ts2. races against other runs
and erases their DBs
Mike Bayer [Tue, 9 Feb 2016 23:05:40 +0000 (18:05 -0500)]
- Fixed bug in :func:`.expression.text` construct where a double-colon
expression would not escape properly, e.g. ``some\:\:expr``, as is most
commonly required when rendering Postgresql-style CAST expressions.
fixes #3644
Mike Bayer [Tue, 9 Feb 2016 22:49:38 +0000 (17:49 -0500)]
- A refinement to the logic which adds columns to the resulting SQL when
:meth:`.Query.distinct` is combined with :meth:`.Query.order_by` such
that columns which are already present will not be added
a second time, even if they are labeled with a different name.
Regardless of this change, the extra columns added to the SQL have
never been returned in the final result, so this change only impacts
the string form of the statement as well as its behavior when used in
a Core execution context. Additionally, columns are no longer added
when the DISTINCT ON format is used, provided the query is not
wrapped inside a subquery due to joined eager loading.
fixes #3641
Mike Bayer [Tue, 9 Feb 2016 03:57:02 +0000 (22:57 -0500)]
- Testing reveals that we have *no* weak references to any cx_Oracle connections at all,
yet cx_Oracle still has open sessions that cannot be killed until process dies.
Oracle wins! Add a completely separate DB reaper script that runs
after py.test is done.
Mike Bayer [Mon, 8 Feb 2016 23:05:51 +0000 (18:05 -0500)]
- fully hyperlink the docstring for make_transient
- establish make_transient and make_transient_to_detached as special-use,
advanced use only functions
- list all conditions under make_transient() under which an attribute
will not be loaded and establish that make_transient() does not attempt
to load all attributes before detaching the object from its
session, fixes #3640
Mike Bayer [Mon, 8 Feb 2016 18:13:39 +0000 (13:13 -0500)]
- adjust to allow dblink tests to work with multiprocess users.
unfortunately the synonym doesn't work for SQL statements here
when the dblink is on a different user, testing this is not really
critical so just removed it.
Mike Bayer [Thu, 4 Feb 2016 18:36:45 +0000 (13:36 -0500)]
- add an impl for Enum to Oracle which has subclassing requirements
on unicode. Enum would be better as a TypeDecorator
at this point but then that becomes awkward with native enum
types (Interval works that way, but we don't need the bind_processor for
native interval...)
Mike Bayer [Tue, 2 Feb 2016 18:00:19 +0000 (13:00 -0500)]
- add changelog and migration notes for new Enum features,
fixes #3095, #3292
- reorganize enum constructor to again work with the MySQL
ENUM type
- add a new create_constraint flag to Enum to complement that of
Boolean
- reinstate the CHECK constraint tests for enum, these already
fail /skip against the MySQL backend
- simplify lookup rules in Enum, have them apply to all varieties
of Enum equally
Mike Bayer [Tue, 2 Feb 2016 17:07:39 +0000 (12:07 -0500)]
- D rules seem to have been added overnight to the flake8 plugin
I'm using and many of these seem contradictory and/or undesirable.
block all until individual rules can be approved.
Mike Bayer [Tue, 2 Feb 2016 15:15:40 +0000 (10:15 -0500)]
- Fixed issue where inadvertent use of the Python ``__contains__``
override with a column expression (e.g. by using ``'x' in col``)
would cause an endless loop in the case of an ARRAY type, as Python
defers this to ``__getitem__`` access which never raises for this
type. Overall, all use of ``__contains__`` now raises
NotImplementedError.
fixes #3642
Mike Bayer [Fri, 29 Jan 2016 16:44:58 +0000 (11:44 -0500)]
- The ``sqlalchemy.dialects.postgres`` module, long deprecated, is
removed; this has emitted a warning for many years and projects
should be calling upon ``sqlalchemy.dialects.postgresql``.
Engine URLs of the form ``postgres://`` will still continue to function,
however.
Mike Bayer [Thu, 28 Jan 2016 20:01:31 +0000 (15:01 -0500)]
- revert the change first made in a6fe4dc, as we are now generalizing
the warning here to all safe_reraise() cases in Python 2.
- Revisiting :ticket:`2696`, first released in 1.0.10, which attempts to
work around Python 2's lack of exception context reporting by emitting
a warning for an exception that was interrupted by a second exception
when attempting to roll back the already-failed transaction; this
issue continues to occur for MySQL backends in conjunction with a
savepoint that gets unexpectedly lost, which then causes a
"no such savepoint" error when the rollback is attempted, obscuring
what the original condition was.
The approach has been generalized to the Core "safe
reraise" function which takes place across the ORM and Core in any
place that a transaction is being rolled back in response to an error
which occurred trying to commit, including the context managers
provided by :class:`.Session` and :class:`.Connection`, and taking
place for operations such as a failure on "RELEASE SAVEPOINT".
Previously, the fix was only in place for a specific path within
the ORM flush/commit process; it now takes place for all transational
context managers as well.
fixes #2696
Mike Bayer [Wed, 27 Jan 2016 22:39:48 +0000 (17:39 -0500)]
- experimental - try out a very open ended design for the factors, assume
many more combinations of factors are possible and basically make them
more akin to command options at the tox level.
Mike Bayer [Wed, 27 Jan 2016 19:23:33 +0000 (14:23 -0500)]
- restore the use of PYTHONNOUSERSITE that was removed
in df3f125bd84fc7ec5d45592c5774daf3a39d9bc9, this flag is
explicitly checked within conftest.py and we need to continue to use
it, otherwise a tox build inside of .tox that isn't usedevelop
is ignored, including C extensions
- rework the whole system of running with coverage, so that
with coverage, we *are* using usedevelop, but also make sure
we rm the .so files for nocext, make sure we --cov-append, etc.
Mike Bayer [Tue, 26 Jan 2016 22:28:40 +0000 (17:28 -0500)]
- unfortunately we need to match within join where col.key does not match
what's given so we need to use a set() here. contains_column is not within
any performance paths
Mike Bayer [Tue, 26 Jan 2016 21:41:26 +0000 (16:41 -0500)]
- rework ColumnCollection to no longer persist "all_col_set"; we don't
need this collection except in the extend/update uses where we
create it ad-hoc. simplifies pickling. Compatibility with 1.0
should be OK as ColumnColleciton uses __getstate__ in any case
and the __setstate__ contract hasn't changed.
- Fixed bug in :class:`.Table` metadata construct which appeared
around the 0.9 series where adding columns to a :class:`.Table`
that was unpickled would fail to correctly establish the
:class:`.Column` within the 'c' collection, leading to issues in
areas such as ORM configuration. This could impact use cases such
as ``extend_existing`` and others. fixes #3632
Mike Bayer [Tue, 26 Jan 2016 20:41:56 +0000 (15:41 -0500)]
- The workaround for right-nested joins on SQLite, where they are rewritten
as subqueries in order to work around SQLite's lack of support for this
syntax, is lifted when SQLite version 3.7.16 or greater is detected.
fixes #3634
- The workaround for SQLite's unexpected delivery of column names as
``tablename.columnname`` for some kinds of queries is now disabled
when SQLite version 3.10.0 or greater is detected.
fixes #3633
Mike Bayer [Sun, 24 Jan 2016 23:28:09 +0000 (18:28 -0500)]
- add a new qualifier "native" vs. "dbapi" unicode to avoid
profiling problems here
- add extras_require to setup.py for the most common DBAPIs
- rework tox.ini to use extras, specify a test matrix built in
Mike Bayer [Thu, 21 Jan 2016 20:21:33 +0000 (15:21 -0500)]
- documenation updates to clarify specific SQLite versions
that have problems with right-nested joins and UNION column keys;
references #3633 references #3634. backport from 1.1 to 0.9
announcing 1.1 as where these behaviors will be retired based
on version-specific checks
- fix test_resultset so that it passes when SQLite 3.10.0 is
present, references #3633
Mike Bayer [Wed, 20 Jan 2016 22:56:04 +0000 (17:56 -0500)]
- Added new parameter :paramref:`.orm.mapper.passive_deletes` to
available mapper options. This allows a DELETE to proceed
for a joined-table inheritance mapping against the base table only,
while allowing for ON DELETE CASCADE to handle deleting the row
from the subclass tables.
fixes #2349
Mike Bayer [Wed, 20 Jan 2016 21:32:27 +0000 (16:32 -0500)]
- The unsupported Sybase dialect now raises ``NotImplementedError``
when attempting to compile a query that includes "offset"; Sybase
has no straightforward "offset" feature. fixes #2278
Mike Bayer [Tue, 19 Jan 2016 22:08:50 +0000 (17:08 -0500)]
- the order in which _generate_backref() for different mappers is called
is random; therefore it may be called against the subclass mapper first, so
need to check .concrete on both sides, references #3630
Mike Bayer [Tue, 19 Jan 2016 21:47:16 +0000 (16:47 -0500)]
- calling str() on a core sql construct has been made more "friendly",
when the construct contains non-standard sql elements such as
returning, array index operations, or dialect-specific or custom
datatypes. a string is now returned in these cases rendering an
approximation of the construct (typically the postgresql-style
version of it) rather than raising an error. fixes #3631
- add within_group to top-level imports
- add eq_ignore_whitespace to sqlalchemy.testing imports
Mike Bayer [Tue, 19 Jan 2016 18:34:42 +0000 (13:34 -0500)]
- Fixed issue where two same-named relationships that refer to
a base class and a concrete-inherited subclass would raise an error
if those relationships were set up using "backref", while setting up the
identical configuration using relationship() instead with the conflicting
names would succeed, as is allowed in the case of a concrete mapping.
fixes #3630
Mike Bayer [Tue, 19 Jan 2016 17:44:42 +0000 (12:44 -0500)]
- The ``str()`` call for :class:`.Query` will now take into account
the :class:`.Engine` to which the :class:`.Session` is bound, when
generating the string form of the SQL, so that the actual SQL
that would be emitted to the database is shown, if possible. Previously,
only the engine associated with the :class:`.MetaData` to which the
mappings are associated would be used, if present. If
no bind can be located either on the :class:`.Session` or on
the :class:`.MetaData` to which the mappings are associated, then
the "default" dialect is used to render the SQL, as was the case
previously. fixes #3081
Mike Bayer [Mon, 18 Jan 2016 22:35:44 +0000 (17:35 -0500)]
- Fixed bug where some exception re-raise scenarios would attach
the exception to itself as the "cause"; while the Python 3 interpreter
is OK with this, it could cause endless loops in iPython.
fixes #3625
- add tests for reraise, raise_from_cause
- raise_from_cause is the same on py2k/3k, use just one function