Mike Bayer [Tue, 14 Apr 2020 19:30:28 +0000 (15:30 -0400)]
Pass connection to TablesTest.insert_data()
towards the goal of reducing verbosity and repetition
in test fixtures as well as that we are moving to
connection only for execution, move the insert_data()
classmethod to accept a connection and adjust all
fixtures to use it.
Mike Bayer [Sun, 12 Apr 2020 19:18:02 +0000 (15:18 -0400)]
Enable zzzeeksphinx module prefixes
zzzeeksphinx 1.1.2 in git can now convert short
prefix names in a configured lookup to fully qualified module
names, so that
we can have succinct and portable pyrefs
that still resolve absolutely.
It also includes a formatter that will format all pyrefs
in a fully consistent way regardless of the package path,
by unconditionally removing all package tokens but always
leaving class names in place including for methods, which
means we no longer have to deal with tildes in pyrefs.
The most immediate goal of the absolute prefixes is
that we have lots of
"ambiguous" names that appear in muliple places, like select(),
ARRAY, ENUM etc. With the incoming future packages there
is going to be lots of name overlap so it is necessary
that all names eventually use absolute package paths
when Sphinx receives them.
In multiple stages, pyrefs will be converted using the
zzzeeksphinx tools/fix_xrefs.py tool so that doclinks can
be made absolute using symbolic prefixes.
For this review, the actual search and replace of symbols
is not performed, instead some general cleanup to prepare
the docs as well as a lookup file used by the tool
to do the conversion. this relatively small patch will
be backported
with appropriate changes to 1.3, 1.2, 1.1 and the tool
can then be run on each branch individually. We are shooting
for almost no warnings at all for master (still a handful
I can't figure out which don't seem to have any impact)
, very few for 1.3,
and for 1.2 / 1.1 we hope for a significant reduction
in warnings.
Overall for all versions pyrefs should
always point to the correct target, if they are in fact
hyperlinked. it's better for a ref to go nowhere and
be plain text than go to the wrong thing. Right now,
hundreds of API links are pointing to the wrong thing
as they are ambiguous names such as refresh(), insert(),
update(), select(), join(), JSON etc. and Sphinx sends these all
to essesntially random destinations among as many as five
or six possible choices per symbol. A shorthand system
that allows us to use absolute refs without having
to type out a full blown absoulte module is the only
way this is going to work, and we should ultimately
seek to abandon any use of prefix dot for lookups. Everything
should be on an underscore token so at the very least the module
spaces can be reorganized without having to search and replace
the entire documentation every time.
Mike Bayer [Mon, 13 Apr 2020 22:39:47 +0000 (18:39 -0400)]
Adjust changelog files for removed directories
unreleased_11 and unreleased_12 are gone so remove these
also there's no place for the 4065 file to go, this is an old
and very small change and it looks like the file was not published in any
case, so just remove it.
Mike Bayer [Sat, 11 Apr 2020 17:46:56 +0000 (13:46 -0400)]
Fix almost all read-level sphinx warnings
There are some related to changelog that I can't figure out
and are likely due to something in the changelog extension.
also one thing with a "collection" I can't figure out.
Mike Bayer [Fri, 10 Apr 2020 20:58:41 +0000 (16:58 -0400)]
Use define_tables for IsOrIsNotDistinctFromTest
In Ie8127ef29f1ec91e7afb88e1429538c27a321784 Mike
failed to notice that this test is build on TablesTest and
has just the one test function, which means the table is
built up by the fixture.
Mike Bayer [Fri, 10 Apr 2020 16:23:32 +0000 (12:23 -0400)]
Evaluate fixes for cx_Oracle 8 API changes
in [1] we have reported some behavioral changes in cx_Oracle
symbols. As it is unclear if these are intended changes in
cx_Oracle, test workarounds for these changes against
cx_Oracle master.
Mike Bayer [Fri, 10 Apr 2020 18:28:54 +0000 (14:28 -0400)]
Repair CTE a in b tests
In I27cac9bd265c86ff2a3381ff9f844f60ef991cfc we modernized
the default tests and converted the "a in b" CTE tests to combinations,
however apparently the existing tests were not testing all
combinations and had repeats instead. The combinations
decorator has made this much easier to spot, so use
the correct combinations that were originally intended.
Mike Bayer [Thu, 9 Apr 2020 15:15:03 +0000 (11:15 -0400)]
Repair find tables
The recipe given out on the ML for improving ABC
fails in 1.4 due to find_tables() missing out on subqueries,
which is due to the refactor of Alias into multiple sub-types.
- Remove deprecated method ``get_primary_keys` in the :class:`.Dialect` and
:class:`.Inspector` classes.
- Remove deprecated event ``dbapi_error`` and the method ``ConnectionEvents.dbapi_error`.
- Remove support for deprecated engine URLs of the form ``postgres://``.
- Remove deprecated dialect ``mysql+gaerdbms``.
- Remove deprecated parameter ``quoting`` from :class:`.mysql.ENUM`
and :class:`.mysql.SET` in the ``mysql`` dialect.
- Remove deprecated function ``comparable_property``. and function
``comparable_using`` in the declarative extension.
- Remove deprecated function ``compile_mappers``.
- Remove deprecated method ``collection.linker``.
- Remove deprecated method ``Session.prune`` and parameter ``Session.weak_identity_map``.
This change also removes the class ``StrongInstanceDict``.
- Remove deprecated parameter ``mapper.order_by``.
- Remove deprecated parameter ``Session._enable_transaction_accounting`.
- Remove deprecated parameter ``Session.is_modified.passive``.
- Remove deprecated class ``Binary``. Please use :class:`.LargeBinary`.
- Remove deprecated methods ``Compiled.compile``, ``ClauseElement.__and__`` and
``ClauseElement.__or__`` and attribute ``Over.func``.
- Remove deprecated ``FromClause.count`` method.
- Remove deprecated parameter ``Table.useexisting``.
- Remove deprecated parameters ``text.bindparams`` and ``text.typemap``.
- Remove boolean support for the ``passive`` parameter in ``get_history``.
- Remove deprecated ``adapt_operator`` in ``UserDefinedType.Comparator``.
Mike Bayer [Tue, 7 Apr 2020 21:37:14 +0000 (17:37 -0400)]
Use dot-separated name resolution for relationship target
The string argument accepted as the first positional argument by the
:func:`.relationship` function when using the Declarative API is no longer
interpreted using the Python ``eval()`` function; instead, the name is dot
separated and the names are looked up directly in the name resolution
dictionary without treating the value as a Python expression. However,
passing a string argument to the other :func:`.relationship` parameters
that necessarily must accept Python expressions will still use ``eval()``;
the documentation has been clarified to ensure that there is no ambiguity
that this is in use.
Mike Bayer [Tue, 7 Apr 2020 23:21:01 +0000 (19:21 -0400)]
Gracefully skip isolation level if no row returned
Fixed issue in MySQL dialect when connecting to a psuedo-MySQL database
such as that provided by ProxySQL, the up front check for isolation level
when it returns no row will not prevent the dialect from continuing to
connect. A warning is emitted that the isolation level could not be
detected.
Federico Caselli [Tue, 24 Mar 2020 20:06:04 +0000 (21:06 +0100)]
Add length parameter in `Enum`
The `Enum` type now supports the parameter `Enum.length`
to specify the length of the VARCHAR column to create when using
non native enums by setting `Enum.native_enum` to `False`
Mike Bayer [Fri, 3 Apr 2020 20:00:22 +0000 (16:00 -0400)]
Key subqueryloaders on the property object, not string key
Fixed bug in :func:`.orm.selectinload` loading option where two or more
loaders that represent different relationships with the same string key
name as referenced from a single :func:`.orm.with_polymorphic` construct
with multiple subclass mappers would fail to invoke each subqueryload
separately, instead making use of a single string-based slot that would
prevent the other loaders from being invoked.
Add github action workflow to run tests on master and on pr to master
Trying the pr to check if it works right away
### Description
<!-- Describe your changes in detail -->
### Checklist
<!-- go over following points. check them with an `x` if they do apply, (they turn into clickable checkboxes once the PR is submitted, so no need to do everything at once)
-->
This pull request is:
- [ ] A documentation / typographical error fix
- Good to go, no issue or tests are needed
- [x] A short code fix
- please include the issue number, and create an issue if none exists, which
must include a complete example of the issue. one line code fixes without an
issue and demonstration will not be accepted.
- Please include: `Fixes: #<issue number>` in the commit message
- please include tests. one line code fixes without tests will not be accepted.
- [ ] A new feature implementation
- please include the issue number, and create an issue if none exists, which must
include a complete example of how the feature would look.
- Please include: `Fixes: #<issue number>` in the commit message
- please include tests.
Mike Bayer [Fri, 3 Apr 2020 00:45:44 +0000 (20:45 -0400)]
Run autoflush for column attribute load operations
The "autoflush" behavior of :class:`.Query` will now trigger for nearly
all ORM level attribute load operations, including when a deferred
column is loaded as well as when an expired column is loaded. Previously,
autoflush on load of expired or unloaded attributes was limited to
relationship-bound attributes only. However, this led to the issue
where column-based attributes that also depended on other rows, or even
other columns in the same row, in order to express the correct value,
would show an effectively stale value when accessed as there could be
pending changes in the session left to be flushed. Autoflush
is now disabled only in some cases where attributes are being unexpired in
the context of a history operation.
Mike Bayer [Wed, 1 Apr 2020 22:31:16 +0000 (18:31 -0400)]
Repair caching / traversals for values
The test suite wasn't running the copy_internals most fixtures,
enable that and try to get all cases working.
Set up selectable.values to do tuple conversion within compilation
step. at the same time, disable caching for selectable.values
for the moment and make it equivalent to dml_multi_values.
fix cache / compare / copy cases for dml_values and dml_multi_values
which weren't fully tested or covered.
Trying to see if by making the cache key memoized, we
still can have the older "identity" form of caching
which is the cheapest of all, at the same time as the
newer "cache key each time" version that is not nearly
as cheap; but still much cheaper than no caching at all.
Also needed is a per-execution update of _keymap when
we invoke from a cached select, so that Column objects
that are anonymous or otherwise adapted will match up.
this is analogous to the adaption of bound parameters
from the cache key.
Adds test coverage for the keymap / construct_params()
changes related to caching. Also hones performance
to a large extent for statement construction and
cache key generation.
Also includes a new memoized attribute
approach that vastly simplifies the previous approach
of "group_expirable_memoized_property" and finally
integrates cleanly with _clone(), _generate(), etc.
no more hardcoding of attributes is needed, as well
as that most _reset_memoization() calls are no longer
needed as the reset is inherent in a _generate() call;
this also has dramatic performance improvements.
Mike Bayer [Wed, 1 Apr 2020 17:51:35 +0000 (13:51 -0400)]
Fix databases -> dialects
The change in I5343a2d270ed5a8c654b9fe13dff40cdf54649ed
left the name "databases" present in one spot, which is usually
not reached unless tests are being run on only partial files
where not every dialect has been imported.
Mike Bayer [Mon, 30 Mar 2020 15:04:24 +0000 (11:04 -0400)]
Remove ORDER BY pk from subqueryload, selectinload
Modified the queries used by subqueryload and selectinload to no longer
ORDER BY the primary key of the parent entity; this ordering was there to
allow the rows as they come in to be copied into lists directly with a
minimal level of Python-side collation. However, these ORDER BY clauses
can negatively impact the performance of the query as in many scenarios
these columns are derived from a subquery or are otherwise not actual
primary key columns such that SQL planners cannot make use of indexes. The
Python-side collation uses the native itertools.group_by() to collate the
incoming rows, and has been modified to allow multiple
row-groups-per-parent to be assembled together using list.extend(), which
should still allow for relatively fast Python-side performance. There will
still be an ORDER BY present for a relationship that includes an explicit
order_by parameter, however this is the only ORDER BY that will be added to
the query for both kinds of loading.
Mike Bayer [Sun, 29 Mar 2020 18:24:39 +0000 (14:24 -0400)]
Add a third labeling mode for SELECT statements
Enhanced the disambiguating labels feature of the
:func:`~.sql.expression.select` construct such that when a select statement
is used in a subquery, repeated column names from different tables are now
automatically labeled with a unique label name, without the need to use the
full "apply_labels()" feature that conbines tablename plus column name.
The disambigated labels are available as plain string keys in the .c
collection of the subquery, and most importantly the feature allows an ORM
:func:`.orm.aliased` construct against the combination of an entity and an
arbitrary subquery to work correctly, targeting the correct columns despite
same-named columns in the source tables, without the need for an "apply
labels" warning.
The existing labeling style is now called
LABEL_STYLE_TABLENAME_PLUS_COL. This labeling style will remain used
throughout the ORM as has been the case for over a decade, however,
the new disambiguation scheme could theoretically replace this scheme
entirely. The new scheme would dramatically alter how SQL looks
when rendered from the ORM to be more succinct but arguably harder
to read.
The tablename_columnname scheme used by Join.c is unaffected here,
as that's still hardcoded to that scheme.
Federico Caselli [Tue, 24 Mar 2020 21:55:46 +0000 (22:55 +0100)]
String compiler can now literal compile datetime objects
Add ability to literal compile a :class:`DateTime`, :class:`Date`
or :class:"Time" when using the string dialect for debugging purposes.
This change does not impact real dialect implementation that retain
their current behavior.
Mike Bayer [Sun, 29 Mar 2020 16:06:29 +0000 (12:06 -0400)]
Repair queries from #5134 to ORDER BY outside the subquery
ORDER BY has to be on the outermost query to guarantee ordering
as these tests were failing for SQL Server. Also add
new tests that don't use from_self() as this is also going to
be removed in 2.0.
Also propose some from_self() alternatives. References #5221
Federico Caselli [Sat, 28 Mar 2020 10:04:44 +0000 (11:04 +0100)]
Fix typo in resultproxy.c and test compatibility with python 3.5
- Fix typo in resultproxy.c that would error on windows.
- add -Wundef to C flags when linux is detected so that undefined
symbols emit a warning
- a few adjustments for tests to succeed on python 3.5
- note minimum version still documented here as 3.4 but this should
move to at least 3.5 if not 3.6 for SQLAlchemy 1.4
Mike Bayer [Thu, 26 Mar 2020 21:26:57 +0000 (17:26 -0400)]
introduce "autobegin" concept for Connection
because engine.connect() and engine.begin() should feature
identical internal behavior, with the sole exception that
one rolls back and the end and the other commits at the end,
while also supporting execution options like transaction isolation level
at the connection level, include that engine.connect() will
return a connection that uses autobegin in the same way as the
session will. This is solely to support the "begin" event
noting that a transaction is begun which is tracked on the
connection. Behavior and design should be very similar to that
of the ORM session and "Transaction" should no longer be a very
explicit object.
projects / thirdparty / sqlalchemy / sqlalchemy.git / log
