Question

I have an application the sets up the logging using:

logging.basicConfig(level=logging_level, format=format_string, filename=log_file, filemode='a')

then call

logging.debug("My Message")

etc. to log messeages. This work fine in most of my application, but then for a particular module I get this error

File "C:\path\to\my\module\MyModule.py", line 53, in __init__
    logging.debug("__init__ called")
  File "C:\Python26\Lib\logging\__init__.py", line 1481, in debug
    root.debug(*((msg,)+args), **kwargs)
  File "C:\Python26\Lib\logging\__init__.py", line 1035, in debug
    if self.isEnabledFor(DEBUG):
  File "C:\Python26\Lib\logging\__init__.py", line 1242, in isEnabledFor
    return level >= self.getEffectiveLevel()
  File "C:\Python26\Lib\logging\__init__.py", line 1230, in getEffectiveLevel
    while logger:
IndexError: list index out of range

Does anyone have any ideas on what could be causing this? Or where else to look to read up on it. I have already read through the logging module code and the python reference pages

Answer 1

Something seems very wrong with your installation – it’s hard to see how a line like

while logger:

would generate an IndexError.

So, delete all .pyc and .pyo files from your system (including in the stdlib folders) and try again. Ensure that none of your modules have the same name as any modules in the standard library.

Also, what version of Python are you using, and on which platform?

Update: If you are using embedded in a C++ program, or with C extensions, it’s quite possible that some C or C++ code is clobbering memory, leading to the IndexError in a most unexpected place.

Can you reproduce in a pure-Python environment? If not, I fear the problem might be in the C/C++ code. It’s also easier to clobber stuff in multi-threaded environments 🙁