test_decimal fails on Mac OS X 10.3 when using Python 2.5.2c1. I haven't
tested other python versions. The same test passes on OSX 10.4 and 10.5
(both on Intel and PPC machines), this was done using the same binaries
(the Python.org build for 2.5.2c1).

The end of the output of "python test_decimal.py" is:

======================================================================
FAIL: test_normalize (main.DecimalTest)
----------------------------------------------------------------------

Traceback (most recent call last):
  File "test_decimal.py", line 1351, in <lambda>
    tester = lambda self, f=filename: self.eval_file(directory + f)
  File "test_decimal.py", line 213, in eval_file
    self.fail('Exception "'+exception.__class__.__name__ + '" raised on 
line '+line)
AssertionError: Exception "Clamped" raised on line nrmx218 normalize  
0.001E-999       ->   0         Inexact Rounded Subnormal Underflow

Ran 189 tests in 35.947s

FAILED (failures=1)
Traceback (most recent call last):
  File "test_decimal.py", line 1376, in <module>
    test_main(arith=True, verbose=True)
  File "test_decimal.py", line 1357, in test_main
    run_unittest(*test_classes)
  File 
"/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/test/te
st_support.py", line 451, in run_unittest
    run_suite(suite, testclass)
  File 
"/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/test/te
st_support.py", line 436, in run_suite
    raise TestFailed(err)
test.test_support.TestFailed: Traceback (most recent call last):
  File "test_decimal.py", line 1351, in <lambda>
    tester = lambda self, f=filename: self.eval_file(directory + f)
  File "test_decimal.py", line 213, in eval_file
    self.fail('Exception "'+exception.__class__.__name__ + '" raised on 
line '+line)
AssertionError: Exception "Clamped" raised on line nrmx218 normalize  
0.001E-999       ->   0         Inexact Rounded Subnormal Underflow

It looks like you've got a stale normalize.decTest file in
Lib/test/decimaltestdata. This file shouldn't be present. Can you do a
fresh svn checkout and see if the problem persists?

I did an upgrade install on the OSX 10.3. Doing a clean install removes
the problem.

I'm therefore closing this issue as invalid. Sorry about the noise.

Hmm. But others may run into this problem as well.

Is there any way to force the upgrade install to remove this file?

OSX installer is lame. The only way to work around this is to have pre-
and/or post-install hooks that remove the stdlib (but not site-
packages).

As this only affects the unittests I don't think this is worth the
problem right now. Especially because I have rather limited time
available until the 2.5.2 release.

A cheap and easy fix would be to update normalize.decTest so that's it's
an exact copy of reduce.decTest. This would result in all the
normalization tests being run twice, but that's probably better than
seeing failing tests.

(What happened here is that Mike Cowlishaw fixed some buggy tests in
normalize.decTest and also renamed normalize.decTest to reduce.decTest,
and renamed the tests within the file.)

But I agree that it's not a disaster if this doesn't get fixed. It's
the tests that are buggy, not the Decimal module.

That's a good idea. I'll try to get this into the installer before 2.5.2
is actually released.

I've applied the workaround in revision 60903 (after testing that it
actually works).