Mike Bayer [Fri, 10 Sep 2021 14:07:15 +0000 (10:07 -0400)]
bulk_save->return_defaults does not populate server defaults
it fetches them but doesn't put them on the objects, which
is likely a bug, however i have no intention of working
on the bulk methods. Add more documentation to strongly
discourage these methods in favor of more explicit bulk
approaches, e.g. using an insert() construct directly.
Mike Bayer [Thu, 2 Sep 2021 18:11:38 +0000 (14:11 -0400)]
turn off deduping for col expressions
Fixed ORM issue where column expressions passed to ``query()`` or
ORM-enabled ``select()`` would be deduplicated on the identity of the
object, such as a phrase like ``select(A.id, null(), null())`` would
produce only one "NULL" expression, which previously was not the case in
1.3. However, the change also allows for ORM expressions to render as given
as well, such as ``select(A.data, A.data)`` will produce a result row with
two columns.
Daniel Stone [Mon, 30 Aug 2021 15:15:25 +0000 (11:15 -0400)]
Added loader options to session.merge, asyncsession.merge
Added loader options to :meth:`_orm.Session.merge` and
:meth:`_asyncio.AsyncSession.merge`, which will apply the given loader
options to the ``get()`` used internally by merge, allowing eager loading
of relationships etc. to be applied when the merge process loads a new
object. Pull request courtesy Daniel Stone.
Mike Bayer [Wed, 1 Sep 2021 12:58:06 +0000 (08:58 -0400)]
add asyncio.gather() example; add connection opts
while I dont like this approach very much, people will likely
be asking for it a lot, so represent the most correct and
efficient form we can handle right now.
Added missing ``**kw`` arguments to the
:meth:`_asyncio.AsyncSession.connection` method.
Mike Bayer [Tue, 31 Aug 2021 17:34:43 +0000 (13:34 -0400)]
Fix and test sequences w/ executemany in pre-exec scenarios
Fixed issue where an engine that had ``implicit_returning`` set to False
would fail to function when PostgreSQL's "fast insertmany" feature were
used in conjunction with a ``Sequence``, as well as if any kind of
"executemany" with "return_defaults()" were used in conjunction with a
``Sequence``. Note that PostgreSQL "fast insertmany" uses "RETURNING" by
definition, when the SQL statement is passed to the driver; overall, the
``implicit_returning`` flag is legacy and has no real use in modern
SQLAlchemy, and will be deprecated in a separate change.
Mike Bayer [Tue, 31 Aug 2021 20:41:58 +0000 (16:41 -0400)]
remove async session slots
1798c3cf1c added __slots__ to the base which then
caused af0824fd79 to fail. as we can't use an assignable
class variable with slots, remove slots from AsyncSession
which is how this class was working anyway.
Fixed bug where the error message for SQLite invalid isolation level on the
pysqlite driver would fail to indicate that "AUTOCOMMIT" is one of the
valid isolation levels.
Federico Caselli [Fri, 27 Aug 2021 20:45:56 +0000 (22:45 +0200)]
Allow custom sync session class in ``AsyncSession``.
The :class:`_asyncio.AsyncSession` now supports overriding which
:class:`_orm.Session` it uses as the proxied instance. A custom ``Session``
class can be passed using the :paramref:`.AsyncSession.sync_session_class`
parameter or by subclassing the ``AsyncSession`` and specifying a custom
:attr:`.AsyncSession.sync_session_class`.
Federico Caselli [Thu, 26 Aug 2021 20:00:33 +0000 (22:00 +0200)]
Handle mappings passed to ``execution_options``.
Fixed a bug in :meth:`_asyncio.AsyncSession.execute` and
:meth:`_asyncio.AsyncSession.stream` that required ``execution_options``
to be an instance of ``immutabledict`` when defined. It now
correctly accepts any mapping.
Mike Bayer [Wed, 25 Aug 2021 17:12:55 +0000 (13:12 -0400)]
ensure "sqlalchemy" info set for all considered classes
Fixed issue in mypy plugin where columns on a mixin would not be correctly
interpreted if the mapped class relied upon a ``__tablename__`` routine
that came from a superclass.
jonathan vanasco [Mon, 23 Aug 2021 20:25:21 +0000 (16:25 -0400)]
standardizing docs #6821
(redo of 2999/I5609025feee8cfdecc09b55bfbf1bd13fa2e6602)
This PR is designed to bring more clarity within the docs by renaming object
instances that may be consfusingly similar to class, method, and attribute names.
For example, instances of the class `MetaData` are available on some objects as
`.metadata` property, and had appeared within the docs as both `meta` and
`metadata` which has confused some users in the past. By this PR, the docs now
utilize the following naming convention:
* MetaData - SQLAlchemy class
* .metadata - SQLAlchemy API attributes
* metadata_obj - developer instantiated metadata objects or references
Detailed Changes:
* standardized `meta` and `metadata` instances to `metadata_obj`. note: the docs were evenly split between 'meta' and 'metadata'.
* standardized 'cursor' to 'cursor_obj' to avoid confusion with the method.
* standardized a 'scalar_subquery = ' to 'scalar_subq' to avoid confusion with the method.
* standardized a 'cte = ' to 'cte_obj' to avoid confusion with the method
Mike Bayer [Mon, 23 Aug 2021 15:24:48 +0000 (11:24 -0400)]
qualify compile_state updates for non-current entities
Fixed issue in recently repaired ``Query.with_entities()`` method where the
flag that determines automatic uniquing for legacy ORM ``Query`` objects
only would be set to ``True`` inappropriately in cases where the
``with_entities()`` call would be setting the ``Query`` to return
column-only rows, which are not uniqued.
Mike Bayer [Fri, 20 Aug 2021 15:47:26 +0000 (11:47 -0400)]
restore statement substitution to before_execute()
Fixed issue where the ability of the
:meth:`_engine.ConnectionEvents.before_execute` method to alter the SQL
statement object passed, returning the new object to be invoked, was
inadvertently removed. This behavior has been restored.
The refactor in a1939719a652774a437f69f8d4788b3f08650089 removed this
feature for some reason and there were no tests in place to detect
it. I don't see any indication this was planned.
Jiayang [Thu, 19 Aug 2021 09:23:51 +0000 (02:23 -0700)]
mysql://user:pass@host/dbname instead of pass:host
Summary:
Reading [1 sqlalchemy.future.Engine](https://docs.sqlalchemy.org/en/14/core/future.html#sqlalchemy.future.Engine) and [2 sqlalchemy.create_engine](https://docs.sqlalchemy.org/en/14/core/engines.html#sqlalchemy.create_engine). I fond @ should be used between pass & host.
Test Plan:
I checked with sqlalchemy 1.4.22 locally and I think @ should be used.
Mike Bayer [Wed, 18 Aug 2021 14:48:16 +0000 (10:48 -0400)]
CAST Oracle table_name, owner, others to VARCHAR(128)
Added a CAST(VARCHAR2(128)) to the "table name", "owner", and other
DDL-name parameters as used in reflection queries against Oracle system
views such as ALL_TABLES, ALL_TAB_CONSTRAINTS, etc to better enable
indexing to take place against these columns, as they previously would be
implicitly handled as NVARCHAR2 due to Python's use of Unicode for strings;
these columns are documented in all Oracle versions as being VARCHAR2 with
lengths varying from 30 to 128 characters depending on server version.
Additionally, test support has been enabled for Unicode-named DDL
structures against Oracle databases.
Mike Bayer [Tue, 17 Aug 2021 19:03:57 +0000 (15:03 -0400)]
send user defined options from the current query
Revised the means by which the
:attr:`_orm.ORMExecuteState.user_defined_options` accessor receives
:class:`_orm.UserDefinedOption` and related option objects from the
context, with particular emphasis on the "selectinload" on the loader
strategy where this previously was not working; other strategies did not
have this problem. The objects that are associated with the current query
being executed, and not that of a query being cached, are now propagated
unconditionally. This essentially separates them out from the "loader
strategy" options which are explicitly associated with the compiled state
of a query and need to be used in relation to the cached query.
The effect of this fix is that a user-defined option, such as those used
by the dogpile.caching example as well as for other recipes such as
defining a "shard id" for the horizontal sharing extension, will be
correctly propagated to eager and lazy loaders regardless of whether
a cached query was ultimately invoked.
Mike Bayer [Fri, 13 Aug 2021 15:07:17 +0000 (11:07 -0400)]
rewrite _extra_criteria in selectinload; propagate correctly to Load
Fixed issue in :func:`_orm.selectinload` where use of the new
:meth:`_orm.PropComparator.and_` feature within options that were nested
more than one level deep would fail to update bound parameter values that
were in the nested criteria, as a side effect of SQL statement caching.
Implementation adds a new step that rewrites the parameters
inside of all _extra_criteria when invoking selectinload
as well as subqueryload. Additionally, changed how Load()
gets "extra_criteria", in that it pulls it from
UnboundLoad._extra_criteria instead of re-fetching it from the
path elements, which are not updated by this new step.
This patch also builds upon the removal of lambda queries
for use in loader strategies in #6889. lambdas made this issue
much more difficult to diagnose. An attempt to reintroduce
lambdas here after finally identifying the "extra_criteria"
issue above showed that lambdas still impact the
assertsql fixture, meaning we have a statement structure that
upon calling .compile() still delivers stale data due to lambdas,
even if caching is turned off, and the non-cached test was still
failing due to stale data within the lambdas.
This is basically the complexity that #6889 fixes and as there's
no real performance gain to using lambdas in these strategies
on top of the existing statement caching that does most of the
work, it should be much less likely going forward to have as many
deeply confusing issues as we've had within selectinload/lazyload
in the 1.4 series.
Mike Bayer [Sun, 15 Aug 2021 22:12:42 +0000 (18:12 -0400)]
honor NO_CACHE in lambdas
Fixed issue in lambda caching system where an element of a query that
produces no cache key, like a custom option or clause element, would still
populate the expression in the "lambda cache" inappropriately.
This was discovered as part of :ticket:`6887` but is a separate
issue.
Mike Bayer [Mon, 16 Aug 2021 21:20:48 +0000 (17:20 -0400)]
remove lambda caching from loader strategies
Adjusted ORM loader internals to no longer use the "lambda caching" system
that was added in 1.4, as well as repaired one location that was still
using the previous "baked query" system for a query. The lambda caching
system remains an effective way to reduce the overhead of building up
queries that have relatively fixed usage patterns. In the case of loader
strategies, the queries used are responsible for moving through lots of
arbitrary options and criteria, which is both generated and sometimes
consumed by end-user code, that make the lambda cache concept not any more
efficient than not using it, at the cost of more complexity. In particular
the problems noted by :ticket:`6881` and :ticket:`6887` are made
considerably less complicated by removing this feature internally.
Fixed an issue where the :class:`_orm.Bundle` construct would not create
proper cache keys, leading to inefficient use of the query cache. This
had some impact on the "selectinload" strategy and was identified as
part of :ticket:`6889`.
Added a Select._create_raw_select() method which essentially
performs ``__new__`` and then populates ``__dict__`` directly,
with no coercions. This saves most of the overhead time that
the lambda caching system otherwise seeks to avoid.
Includes removal of bakedquery from
mapper->_subclass_load_via_in() which was overlooked from
the 1.4 refactor.
Mike Bayer [Sun, 15 Aug 2021 16:21:13 +0000 (12:21 -0400)]
fix linter JOIN logic; fix PostgreSQL ARRAY op comparison
Adjusted the "from linter" warning feature to accommodate for a chain of
joins more than one level deep where the ON clauses don't explicitly match
up the targets, such as an expression such as "ON TRUE". This mode of use
is intended to cancel the cartesian product warning simply by the fact that
there's a JOIN from "a to b", which was not working for the case where the
chain of joins had more than one element.
this incurs a bit more compiler overhead that comes out in profiling
but is not extensive.
Added the "is_comparison" flag to the PostgreSQL "overlaps",
"contained_by", "contains" operators, so that they work in relevant ORM
contexts as well as in conjunction with the "from linter" feature.
Mike Bayer [Mon, 26 Jul 2021 22:06:41 +0000 (18:06 -0400)]
limit greenlet dependency to pypi-listed platforms
The setup requirements have been modified such ``greenlet`` is a default
requirement only for those platforms that are well known for ``greenlet``
to be installable and for which there is already a pre-built binary on
pypi; the current list is ``x86_64 aarch64 ppc64le amd64 win32``. For other
platforms, greenlet will not install by default, which should enable
installation and test suite running of SQLAlchemy 1.4 on platforms that
don't support ``greenlet``, excluding any asyncio features. In order to
install with the ``greenlet`` dependency included on a machine architecture
outside of the above list, the ``[asyncio]`` extra may be included by
running ``pip install sqlalchemy[asyncio]`` which will then attempt to
install ``greenlet``.
Additionally, the test suite has been repaired so that tests can complete
fully when greenlet is not installed, with appropriate skips for
asyncio-related tests.
João Sampaio [Wed, 11 Aug 2021 14:36:49 +0000 (11:36 -0300)]
Fix example with deprecated code
Just a few lines above, it is said that creating a mapping with `mapper()` is deprecated in favor of `mapper_registry.map_imperatively()`. This PR fixes an example in the documentation that still uses the old function.
Mike Bayer [Mon, 9 Aug 2021 18:43:53 +0000 (14:43 -0400)]
create concise + deterministic cache key for unboundload.options
Fixed issue in loader strategies where the use of the
:meth:`_orm.Load.options` method, particularly when nesting multiple calls,
would generate an overly long and more importantly non-deterministic cache
key, leading to very large cache keys which were also not allowing
efficient cache usage, both in terms of total memory used as well as number
of entries used in the cache itself.
Eric Masseran [Fri, 6 Aug 2021 08:11:09 +0000 (04:11 -0400)]
Dispatch independent ctes on compound select
Fix issue in :class:`_sql.CTE` where new :meth:`_sql.HasCTE.add_cte` method
added in version 1.4.21 / :ticket:`6752` failed to function correctly for
"compound select" structures such as :func:`_sql.union`,
:func:`_sql.union_all`, :func:`_sql.except`, etc. Pull request courtesy
Eric Masseran.
Mike Bayer [Fri, 23 Jul 2021 20:07:50 +0000 (16:07 -0400)]
add columns_clause_froms and related use cases
Added new attribute :attr:`_sql.Select.columns_clause_froms` that will
retrieve the FROM list implied by the columns clause of the
:class:`_sql.Select` statement. This differs from the old
:attr:`_sql.Select.froms` collection in that it does not perform any ORM
compilation steps, which necessarily deannotate the FROM elements and do
things like compute joinedloads etc., which makes it not an appropriate
candidate for the :meth:`_sql.Select.select_from` method. Additionally adds
a new parameter
:paramref:`_sql.Select.with_only_columns.maintain_column_froms` that
transfers this collection to :meth:`_sql.Select.select_from` before
replacing the columns collection.
In addition, the :attr:`_sql.Select.froms` is renamed to
:meth:`_sql.Select.get_final_froms`, to stress that this collection is not
a simple accessor and is instead calculated given the full state of the
object, which can be an expensive call when used in an ORM context.
Additionally fixes a regression involving the
:func:`_orm.with_only_columns` function to support applying criteria to
column elements that were replaced with either
:meth:`_sql.Select.with_only_columns` or :meth:`_orm.Query.with_entities` ,
which had broken as part of :ticket:`6503` released in 1.4.19.
Mike Bayer [Sat, 7 Aug 2021 15:02:59 +0000 (11:02 -0400)]
dont qualify literal_binds with literal_execute
this appears to be unnecessary and prevents end-user
literal_binds case from working.
Fixed issue where the ``literal_binds`` compiler flag, as used externally
to render bound parameters inline, would fail to work when used with a
certain class of parameters known as "literal_execute", which covers things
like LIMIT and OFFSET values for dialects where the drivers don't allow a
bound parameter, such as SQL Server's "TOP" clause. The issue locally
seemed to affect only the MSSQL dialect.
Mike Bayer [Thu, 5 Aug 2021 23:17:07 +0000 (19:17 -0400)]
accommodate for untracked boundparam lambda in offline_string
Fixed an issue in the ``CacheKey.to_offline_string()`` method used by the
dogpile.caching example where attempting to create a proper cache key from
the special "lambda" query generated by the lazy loader would fail to
include the parameter values, leading to an incorrect cache key.
Mike Bayer [Thu, 29 Jul 2021 14:10:28 +0000 (10:10 -0400)]
accommodate for cloned bindparams w/ maintain_key
Fixed issue where a bound parameter object that was "cloned" would cause a
name conflict in the compiler, if more than one clone of this parameter
were used at the same time in a single statement. This could occur in
particular with things like ORM single table inheritance queries that
indicated the same "discriminator" value multiple times in one query.
Deprecate scoped_session usage with async sessions
Deprecate usage of :class:`_orm.scoped_session` with asyncio drivers.
When using Asyncio the :class:`_asyncio.async_scoped_session` should
be used instead.
Mike Bayer [Tue, 27 Jul 2021 15:35:41 +0000 (11:35 -0400)]
accommodate plain core textual statements
Fixed issue where the horizontal sharding extension would not correctly
accommodate for a plain textual SQL statement passed to
:meth:`_orm.Session.execute`.
projects / thirdparty / sqlalchemy / sqlalchemy.git / log
