Merge lp:~anybox/openerp-command/trunk-pdb-tests-poc into lp:openerp-command

[Georges Racinet]

This is a proof-of-concept implementation of post-mortem pdb for failures and errors.

The pdb related options are directly set on the TestResult subclass, which is a bit harsh and could be surprising for further developments, but that's good enough to demonstrate that it works.

A cleaner way would be to subclass TextTestRunner as well to pass options to PdbTextTestResult.__init__(). That kind of subclassing would not lead to much duplication.

If you'd prefer it that way, just tell me.

Cheers,

[Vo Minh Thu (thu)]

This seems nice (I didn't try it).

Actually in the case of a failure, the stack wouldn't be much interesting (the stack would lead to the assertion error, not to the code at fault).

So this makes it possible to write it a bit more cleanly: just check a single --post-mortem flag, if it true, use your PdbTextTestResult, otherwise use the regular TextTestResult. (So no need to pass aroung the flag through class variables.)

Thanks!

[Vo Minh Thu (thu)]

(I mean, the `if flag` in the PdbTextTestResult can be removed as the class is used only when the only remaining flag is True.)

Unmerged revisions

182. By Georges Racinet <email address hidden> on 2012-11-04

[IMP] options for pdb post-mortem debugging on failures and errors

=== modified file 'openerpcommand/run_tests.py'
--- openerpcommand/run_tests.py	2012-10-09 12:34:39 +0000
+++ openerpcommand/run_tests.py	2012-11-04 12:26:19 +0000
@@ -91,6 +91,7 @@
 
 def run(args):
     import unittest2
+    from .test_result import PdbTextTestResult
 
     import openerp
 
@@ -157,7 +158,10 @@
         suite = unittest2.TestSuite()
         for test_module in test_modules:
             suite.addTests(unittest2.TestLoader().loadTestsFromModule(test_module))
-        unittest2.TextTestRunner(verbosity=2).run(suite)
+        PdbTextTestResult.pdb_errors = bool(args.pdb)
+        PdbTextTestResult.pdb_failures = bool(args.pdb_failures)
+        unittest2.TextTestRunner(resultclass=PdbTextTestResult,
+                                 verbosity=2).run(suite)
     else:
         print 'Test modules:'
         for test_module in test_modules:
@@ -187,4 +191,10 @@
     parser.add_argument('--dry-run', action='store_true',
         help='do not run the tests')
 
+    # options for post-mortem pdb, named as in nosetests
+    parser.add_argument('--pdb', action='store_true',
+        help='Invoke post mortem pdb on errors')
+    parser.add_argument('--pdb-failures', action='store_true',
+        help='Invoke post mortem pdb on failures')
+
     parser.set_defaults(run=run)

=== added file 'openerpcommand/test_result.py'
--- openerpcommand/test_result.py	1970-01-01 00:00:00 +0000
+++ openerpcommand/test_result.py	2012-11-04 12:26:19 +0000
@@ -0,0 +1,20 @@
+import pdb
+from unittest2.runner import TextTestResult
+
+class PdbTextTestResult(TextTestResult):
+    """A subclass with post-mortem pdb capability for errors and failures.
+    """
+
+    pdb_failures = False
+
+    pdb_errors = False
+
+    def addFailure(self, test, err):
+        super(PdbTextTestResult, self).addFailure(test, err)
+        if self.pdb_failures:
+            pdb.post_mortem(err[2])
+
+    def addError(self, test, err):
+        super(PdbTextTestResult, self).addError(test, err)
+        if self.pdb_errors:
+            pdb.post_mortem(err[2])