Jason Kirtland [Tue, 31 Mar 2009 19:49:13 +0000 (19:49 +0000)]
- Added support for create_engine(isolation_level=...); postgres &
sqlite initially [ticket:443]
- Dialects gained visit_pool
- Pools gained a first_connect event
Patch from Adam Lowry. Thank you Adam!
Mike Bayer [Tue, 31 Mar 2009 17:00:02 +0000 (17:00 +0000)]
- rollback everything before dropping tables. PG + jython is a rough combo
- zxJDBC totally not returning rowcounts correctly
- some dict ordering for jython
Mike Bayer [Sat, 21 Mar 2009 19:34:45 +0000 (19:34 +0000)]
- connection initialize moves to a connection pool event [ticket:1340]
- sqlite doesn't support schemas. not sure if some versions do, but marking those as unsupported for now.
- added a testing.requires callable for schema support.
- standardized the "extra schema" name for unit tests as "test_schema" and "test_schema_2".
- sqlite needs description_encoding (was some other version of pysqlite tested here ?)
- other test fixes.
Arc Riley [Wed, 18 Mar 2009 00:13:39 +0000 (00:13 +0000)]
Set the set in __builtins__ check to Py2-only, set is always available in Py3
I believe some other test may be overriding __builtins__ to be a dict instead of a module only on Py3, but this is the easiest fix.
Mike Bayer [Fri, 30 Jan 2009 19:22:43 +0000 (19:22 +0000)]
- sql round trips are coming on line for sqlite, pg8000
- pg8000 returns col.description as bytes, heh
- absolute module imports play havoc with "python test/foo/test.py" - start using -m
Mike Bayer [Fri, 30 Jan 2009 15:12:41 +0000 (15:12 +0000)]
use sqlalchemy.util functions here. the previous implementation is
not compatible with py3k. I can run all tests or any combination of tests
by using sqlalchemy.util so it's not clear to me why sqlalchemy.util
reportedly cannot be imported at this phase.
Mike Bayer [Sun, 25 Jan 2009 00:48:15 +0000 (00:48 +0000)]
- the 'expire' option on query.update() has been renamed to 'fetch', thus matching
that of query.delete()
- query.update() and query.delete() both default to 'evaluate' for the synchronize
strategy.
- the 'synchronize' strategy for update() and delete() raises an error on failure.
There is no implicit fallback onto "fetch". Failure of evaluation is based
on the structure of criteria, so success/failure is deterministic based on
code structure.
Mike Bayer [Sat, 17 Jan 2009 07:03:36 +0000 (07:03 +0000)]
- added a compiler extension that allows easy creation of user-defined compilers,
which register themselves with custom ClauseElement subclasses such that the compiler
is invoked along with the primary compiler. The compilers can also be registered
on a per-dialect basis.
This provides a supported path for SQLAlchemy extensions such as ALTER TABLE
extensions and other SQL constructs.
Mike Bayer [Fri, 16 Jan 2009 19:31:28 +0000 (19:31 +0000)]
- more or less pg8000 support. has a rough time with non-ascii data.
- removed "send unicode straight through" logic from sqlite, this becomes
base dialect configurable
- simplfied Interval type to not have awareness of PG dialect. dialects
can name TypeDecorator classes in their colspecs dict.
Mike Bayer [Wed, 14 Jan 2009 17:52:10 +0000 (17:52 +0000)]
- mysql+pyodbc working for regular usage, ORM, etc. types and unicode still flaky.
- updated testing decorators to receive "name+driver"-style specifications