I have some code that adds new test funtions to a TestCase class as lambda functions:

ExmplTests.myTest = lambda: 1+2

When I run the app and specify my lamba test to run, as in:

myTestProg ExmplTests.mytest

I get:

Traceback (most recent call last):
  File "C:\work\CTTS\o-ticket\scOmniCTTS\test\otfbatch_test.py", line 188, in ?
    unittest.main(defaultTest = 'suite')
  File "C:\Python24\lib\unittest.py", line 761, in __init__
    self.parseArgs(argv)
  File "C:\Python24\lib\unittest.py", line 788, in parseArgs
    self.createTests()
  File "C:\Python24\lib\unittest.py", line 794, in createTests
    self.module)
  File "C:\Python24\lib\unittest.py", line 559, in loadTestsFromNames
    suites = [self.loadTestsFromName(name, module) for name in names]
  File "C:\Python24\lib\unittest.py", line 543, in loadTestsFromName
    return parent(obj.__name__)
  File "C:\Python24\lib\unittest.py", line 211, in __init__
    raise ValueError, "no such test method in %s: %s" % \
ValueError: no such test method in <class '__main__.ExmplTests'>: <lambda>

It seems this is the case because lamba functions are always named '<lamba>'.  If I change unittest.py ln543 from:
  return parent(obj.__name__)
to:
  return parent(part)

it fixes my problem because part has the function name as a string, instead of getting it from obj, which returns '<lambda>' in my case.

Here is code that causes the problem:

import unittest
class Exmpl(unittest.TestCase):
  pass
Exmpl.testBug = lambda self: 1+2
if (__name__ == '__main__'):
  unittest.main()

Running "program.py" with no args is an error free run.  Running "program.py Exmpl.testBug" dies with the stack trace mentioned above.

This is fixed in at least 2.6 and trunk.