Test failures with Python 2.6

Thanks for the patch. As we depend on testtools we can use its support for those API calls rather than removing the tests entirely.

And, TestResultDecorator now includes an ExtendedToOriginalDecorator call, TestResultFilter subclasses TestResultDecorator, so AFAICT, this should be fixed in trunk. I realise you filed the bug only recently, but this was changed only recently :(.

Could you please retry with trunk, or with 0.0.8 once I release it? If its still a problem, I'll be delighted to do a 0.0.9 fixing the issue.

Results for revision 169:

$ python2.6 -m testtools.run subunit.test_suite
Tests running...
======================================================================
ERROR: subunit.tests.test_subunit_filter.TestTestResultFilter.test_fixup_expected_errors
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/subunit/python/subunit/tests/test_subunit_filter.py", line 144, in test_fixup_expected_errors
    [failure[0].id() for failure in filtered_result.expectedFailures])
AttributeError: 'TestResult' object has no attribute 'expectedFailures'
======================================================================
ERROR: subunit.tests.test_subunit_filter.TestTestResultFilter.test_fixup_expected_failures
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/subunit/python/subunit/tests/test_subunit_filter.py", line 134, in test_fixup_expected_failures
    [failure[0].id() for failure in filtered_result.expectedFailures])
AttributeError: 'TestResult' object has no attribute 'expectedFailures'
======================================================================
ERROR: subunit.tests.test_subunit_filter.TestTestResultFilter.test_fixup_unexpected_success
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/subunit/python/subunit/tests/test_subunit_filter.py", line 154, in test_fixup_unexpected_success
    [passed.id() for passed in filtered_result.unexpectedSuccesses])
AttributeError: 'TestResult' object has no attribute 'unexpectedSuccesses'
======================================================================
ERROR: subunit.tests.test_subunit_filter.TestTestResultFilter.test_time_ordering_preserved
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/subunit/python/subunit/tests/test_subunit_filter.py", line 234, in test_time_ordering_preserved
    self.assertSequenceEqual(
AttributeError: 'TestTestResultFilter' object has no attribute 'assertSequenceEqual'
======================================================================
ERROR: subunit.tests.test_subunit_filter.TestTestResultFilter.test_time_passes_through_filtered_tests
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/subunit/python/subunit/tests/test_subunit_filter.py", line 262, in test_time_passes_through_filtered_tests
    self.assertSequenceEqual(
AttributeError: 'TestTestResultFilter' object has no attribute 'assertSequenceEqual'

Ran 256 tests in 0.918s
FAILED (failures=5)

The exceptions are different:

$ python3.1 -c 'open("/dev/null", "w").read()'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
IOError: not readable
$ python3.2 -c 'open("/dev/null", "w").read()'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
io.UnsupportedOperation: not readable

The following fix works, but maybe _UnsupportedOperation should be changed.

--- python/subunit/__init__.py
+++ python/subunit/__init__.py
@@ -1280,7 +1280,7 @@
             # Read streams
             if type(stream.read(0)) is str:
                 return stream.buffer
- except _UnsupportedOperation:
+ except (_UnsupportedOperation, IOError):
             # Cannot read from the stream: try via writes
             try:
                 stream.write(_b(''))

===

The last comment was for bug #987514.

Ok, I'll have a stab at this, and comment here when I have landed a possible fix. Should be a few minutes.

ok, the assertSequenceEquals was spurious and unneeded, I've made that assertEqual. The other things are introspecting attributes that only existing in 2.7+. Seeing if there is a workaround now.

nope, nothing trivial, will take your approach.