Commits

Show all
Author Commit Message Labels Comments Date
Antonio Cuni
w_None is a valid value to assign to __traceback__
Branches
py3k
Antonio Cuni
kill sys.exc_clear(). Also kill OperationError.clear, which seems to be no longer used anywhere else now. I hope not to be wrong :-)
Branches
py3k
Antonio Cuni
bah, confusion between applevel and interplevel None
Branches
py3k
Antonio Cuni
bump the pyc magic number; this should have been checked in with 3648ec4ef989
Branches
py3k
Antonio Cuni
bah, we need to support both py3k and py2 bytecodes, for the flow objspace. Add a flag to the space, and check whether or not we need to support POP_EXCEPT
Branches
py3k
Antonio Cuni
use the official way to check whether an object is a valid traceback, and add a test
Branches
py3k
Antonio Cuni
kill XXX after discussion with armin on IRC, we don't care :-)
Branches
py3k
Antonio Cuni
implement support for __traceback__ when raising exceptions. test_raise fully passes now :-)
Branches
py3k
Antonio Cuni
fix test_with_reraise_2: when we pop a WithBlock, the exception must be considered already handled, and thus we don't want to restore it (which is different than FinallyBlock)
Branches
py3k
Antonio Cuni
two more tests from cpython's test suite. The first passes, the second is failing
Branches
py3k
Antonio Cuni
a passing test, from cpython's test suite
Branches
py3k
Antonio Cuni
emit and implement POP_EXCEPT, which is needed for lexical exception handlers. This is equivalent to part of the patch at http://bugs.python.org/issue3021
Branches
py3k
Antonio Cuni
add a more useful repr for OperationError, and add a warning when using pycode.dump()
Branches
py3k
Antonio Cuni
add two tests which fails because we don't emit/implement POP_EXCEPT
Branches
py3k
Antonio Cuni
make sure that .dump() works again
Branches
py3k
Antonio Cuni
in python3, when we enter the finally: block because of an exception sys.exc_info() returns the current one, as it happens for except: blocks. In python2, the current exception was updated only inside the latters
Branches
py3k
Antonio Cuni
simplify the code in RAISE_VARARGS now that we no longer support the form with three arguments
Branches
py3k
Antonio Cuni
improve the raises() function which is used inside -A tests. Now test_raises -A passes on cpython3
Branches
py3k
Antonio Cuni
we cannot catch '42' in py3. Not sure whether it's essential for the point of the test, removing this except clause does not change anything even on the default branch
Branches
py3k
Antonio Cuni
again, we need to subclass Exception to raise/catch exceptions
Branches
py3k
Antonio Cuni
kill this test as well, we can only raise subclasses of BaseException nowadays
Branches
py3k
Antonio Cuni
kill this test, we can no longer raise tuples
Branches
py3k
Antonio Cuni
try to refactor this test to pass on python3 with -A
Branches
py3k
Antonio Cuni
f_exc_type no longer exists. Rewrite the test using exc_info(), and also the type of the exception is different
Branches
py3k
Antonio Cuni
python3 changed the behavior in case we raise an exception from within a finally block: in python2 it was ignored and the main exception went through, in python3 the new exception is raised and the old one is set as __context__. Fix two tests to account for this change, they now pass with -A (but still fail on py.py)
Branches
py3k
Antonio Cuni
- allow to pass string to raises when we run tests with -A
Branches
py3k
Antonio Cuni
python3 no longer supports the form 'raise Type, value, tb'. Instead, we can use __traceback__. Adapt the test to the new semantics: it passes with -A but still fails on py.py
Branches
py3k
Antonio Cuni
adapt the syntax to py3k, and kill some outdated tests about the 'raise Type, args' form which is no longer valid
Branches
py3k
Antonio Cuni
kill this test: python3 no longer supports the "raise with traceback" form: we
Branches
py3k
Antonio Cuni
save the source of applevel direct tests in a temporary file: this way, we get nicer tracebacks
Branches
py3k
  1. Prev
  2. Next