Through a small piece of stupid code, I found a bug in the Jython 2.0
compiler.
In a function (in a class, in a module if it matters) I had the
following code :
del some_var
Originally some_var existed (a temporary local), but though some
modifications some_var no longer existed. Trying to run the app
yielded the following (partial) result :
./run.bash Main.py
Traceback (innermost last):
File "Main.py", line 33, in ?
File "SessionTests\__init__.py", line 25, in ?
java.lang.NullPointerException
at org.python.compiler.CodeCompiler.Name(Compiled Code)
at org.python.parser.SimpleNode.visit(Compiled Code)
at org.python.compiler.CodeCompiler.del_stmt(CodeCompiler.java:470)
at org.python.parser.SimpleNode.visit(Compiled Code)
at org.python.compiler.CodeCompiler.suite(Compiled Code)
at org.python.parser.SimpleNode.visit(Compiled Code)
at org.python.compiler.CodeCompiler.parse(Compiled Code)
... (very long stack trace)
at org.python.core.PyTableCode.call(Compiled Code)
at org.python.core.Py.runCode(Py.java:1055)
at org.python.core.__builtin__.execfile(__builtin__.java:288)
at org.python.core.__builtin__.execfile(__builtin__.java:292)
at org.python.util.PythonInterpreter.execfile(PythonInterpreter.java:155)
at org.python.util.jython.main(jython.java:159)
java.lang.NullPointerException: java.lang.NullPointerException
Obviously the error was on my part (can't del a var that DNE), but the
compiler should give a more informative error message (;-)). Maybe it
should wait and give a NameError during exection, since it is possible
that I created the name through exec() or something. (I didn't use
any fancy dynamism in this case)
I just checked against CPython and found this:
>>> def foo( ):
... del dne
...
>>> foo()
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "<stdin>", line 2, in foo
UnboundLocalError: Local variable 'dne' referenced before assignment
>>>
All of this is on a Win2k box. Version 2.0 for both python
interpreters, jdk1.1.8 for the jvm.
If you want any more info just yell ;-).
-D
|
