Python 3, including some important decode/bytes steps.
Issues remain with BLOB types due to driver issues.
Courtesy Ben Trofatter.
- start using util.py3k, we will eventually remove the
sa2to3 fixer entirely
the creation of strong references within the Session;
an object will no longer have an internal reference cycle
created if it's in the transient state or moves into the
detached state - the strong ref is created only when the
object is attached to a Session and is removed when the
object is detached. This makes it somewhat safer for an
object to have a `__del__()` method, even though this is
not recommended, as relationships with backrefs produce
cycles too. A warning has been added when a class with
a `__del__()` method is mapped.
[ticket:2708]
the limit/offset parameter values wouldn't be taken into
account when computing the cache key. The
_key_from_query() function has been simplified to work
directly from the final compiled statement in order to get
at both the full statement as well as the fully processed
parameter list.
- Part of a longer series of fixes needed for pyodbc+
mssql, a CAST to NVARCHAR(max) has been added to the bound
parameter for the table name and schema name in all information schema
queries to avoid the issue of comparing NVARCHAR to NTEXT,
which seems to be rejected by the ODBC driver in some cases,
such as FreeTDS (0.91 only?) plus unicode bound parameters being passed.
The issue seems to be specific to the SQL Server information
schema tables and the workaround is harmless for those cases
where the problem doesn't exist in the first place.
[ticket:2355]
a rollback() before re-raising, so that the stack
trace is preserved from sys.exc_info() before entering
the rollback. This so that the traceback is preserved
when using coroutine frameworks which may have switched
contexts before the rollback function returns.
[ticket:2703]
query when refreshing an inheritance-mapped class
where the superclass was mapped to a non-Table
object, like a custom join() or a select(),
running a query that assumed a hierarchy that's
mapped to individual Table-per-class.
[ticket:2697]
the bytes() callable when run on Python 3; in particular
psycopg2 2.5 with Python 3.3 seems to now be returning
the "memoryview" type, so this is converted to bytes
before return.
handling. If a non-disconnect error occurs,
but leads to a delayed disconnect error within error
handling (happens with MySQL), the disconnect condition
is detected. The Connection can now also be closed
when in an invalid state, meaning it will raise "closed"
on next usage, and additionally the "close with result"
feature will work even if the autorollback in an error
handling routine fails and regardless of whether the
condition is a disconnect or not.
[ticket:2695]
labeled columns when apply_labels() is used; this mode
produces a SELECT where each column is labeled as in
<tablename>_<columnname>, to remove column name collisions
for a multiple table select. The fix is that if two labels
collide when combined with the table name, i.e.
"foo.bar_id" and "foo_bar.id", anonymous aliasing will be
applied to one of the dupes. This allows the ORM to handle
both columns independently; previously, 0.7
would in some cases silently emit a second SELECT for the
column that was "duped", and in 0.8 an ambiguous column error
would be emitted. The "keys" applied to the .c. collection
of the select() will also be deduped, so that the "column
being replaced" warning will no longer emit for any select()
that specifies use_labels, though the dupe key will be given
an anonymous label which isn't generally user-friendly.
[ticket:2702]
where since it considers the current class' ``__table__``, was
sensitive to when it was called. This is 0.7's behavior also,
but in 0.7 things tended to "work out" within events like
``__mapper_args__()``. :func:`.has_inherited_table` now only
considers superclasses, so should return the same answer
regarding the current class no matter when it's called
(obviously assuming the state of the superclass).
[ticket:2656]
``query(SubClass).options(subqueryload(Baseclass.attrname))``,
where ``SubClass`` is a joined inh of ``BaseClass``,
would fail to apply the ``JOIN`` inside the subquery
on the attribute load, producing a cartesian product.
The populated results still tended to be correct as additional
rows are just ignored, so this issue may be present as a
performance degradation in applications that are
otherwise working correctly. [ticket:2699]
subclass could insert the row for the "sub" table
before the parent table, if the two tables had no
ForeignKey constraints set up between them.
Also in 0.7.11. [ticket:2689]
- fix a glitch in the assertsql.CompiledSQL fixture regarding
when a multiparam compiledSQL is used within an AllOf
- add a new utility function randomize_unitofwork() which
does the function of --reversetop
passed to Table(); since we want to support external dialects
and also want to support args without a certain dialect
being installed, it only checks the format of the arg now,
rather than looking for that dialect in sqlalchemy.dialects.
- remove most use of "whereclause" and "from_obj", not done yet,
part of [ticket:2679]
- add links to method chained versions for all Select.append_XYZ()
calls
- add "method chaining" and "generative" to glossary
function syntax, renders as "SUBSTRING(x FROM y FOR z)"
when regular ``func.substring()`` is used.
Also in 0.7.11. Courtesy Gunnlaugur Por Briem.
[ticket:2676]
added, which proxies down to the ``.info`` attribute on either
the :class:`.schema.Column` object if directly present, or
the :class:`.MapperProperty` otherwise. The full behavior
is documented and ensured by tests to remain stable.
[ticket:2675]
mapper configuration; will now test for name conflicts on
superclasses and subclasses, in addition to the current mapper,
as these conflicts break things just as much. This is new for
0.8, but see below for a warning that will also be triggered
in 0.7.11.
- Improved the error message emitted when a "backref loop" is detected,
that is when an attribute event triggers a bidirectional
assignment between two other attributes with no end.
This condition can occur not just when an object of the wrong
type is assigned, but also when an attribute is mis-configured
to backref into an existing backref pair. Also in 0.7.11.
- A warning is emitted when a MapperProperty is assigned to a mapper
that replaces an existing property, if the properties in question
aren't plain column-based properties. Replacement of relationship
properties is rarely (ever?) what is intended and usually refers to a
mapper mis-configuration. Also in 0.7.11.
[ticket:2674]
construct after it's been constructed already. This is not
a pattern for normal use but we like to change the setting
for demonstration purposes in tutorials.
now accommodate an empty list as an instruction to not create/drop
any items, rather than ignoring the collection. [ticket:2664].
This is a behavioral change and extra notes to the changelog
and migration document have been added.
- create a new test suite for exercising codepaths
in engine/ddl.py
attempts to emit SQL on a Session within the after_commit()
handler, where there is not a viable transaction in progress.
[ticket:2662]
- rework how SessionTransaction maintains state, using symbols
instead.
- add lots of notes and cross-linking for session events.
- add a link to :func:`.select()` within :meth:`.FromClause.select`.
Mike Bayer