fix tests on python 3.2

[Mike Bayer]

3.2 allowed non-attribute exceptions to propagate from hasattr(). a couple of tests which are specifically testing our workarounds for the old behavior choke on this.

[Former user]

Update that removes 3 errors in running tests when using 3.2

[Mike Bayer]

I think due to the usage of __getattr__ on the _Group class we are stuck raising InvalidRequestError and a behavioral change with Py3.2.

I would add this test to SelectTest.test_attribute_sanity as well, using the conditional approach:

assert not hasattr(select([table1.c.myid](table1.c.myid)).as_scalar(), 'columns')

and also rename py3_2k to py32.

[Former user]

(original author: taavi) Reworked patch with zzzeek's suggestions, and (imo) some more explicit exception handling

[Former user]

(original author: taavi) Note that even with 2088-take2.patch there are still 8 failures when running py3.2 sqlite: * {{{test.aaa_profiling.test_pool.QueuePoolTest.test_first_connect}}} (55 vs 72) * {{{test.aaa_profiling.test_pool.QueuePoolTest.test_second_connect}}} (23 vs 32) * {{{test.aaa_profiling.test_pool.QueuePoolTest.test_second_samethread_connect}}} (7 vs 6) * {{{test.engine.test_pool.QueuePoolTest.test_max_overflow}}} ({{{assert not pool._refs}}}) * {{{test.engine.test_pool.QueuePoolTest.test_mixed_close}}} ({{{assert not pool._refs}}}) * {{{test.engine.test_pool.QueuePoolTest.test_no_overflow}}} ({{{assert not pool._refs}}}) * {{{test.engine.test_pool.QueuePoolTest.testqueuepool_close}}} ({{{assert not pool._refs}}}) * {{{test.engine.test_pool.QueuePoolTest.testqueuepool_del}}} ({{{assert not pool._refs}}})

[Former user]

(original author: taavi) 2088-take2.patch applied with f9e58cad3bca4bd141ceb5ab4846c2258279ae34 (thanks lsblakk!)

[Mike Bayer]

the remaining should be trivial, there's a 3.2 build on jenkins and current local output is http://paste.pocoo.org/show/362703/

[Mike Bayer]

we're good on this thanks for the patches !

[Mike Bayer]

Removing milestone: 0.7.0 (automated comment)