My previous logging to file function will be affected by add pyxcp application .The previous log.info function reported errors

[wangwei357]

My previous logging to file function will be affected by add pyxcp application . The previous log.info function reported errors
only log to file is affected , the steram out is OK

File "C:\Users\XXXXXX\AppData\Local\anaconda3\Lib\logging_init_.py", line 1113, in emit
stream.write(msg + self.terminator)
ValueError: I/O operation on closed file.

in my other coder logger config , I have code like this :

    handler = logging.FileHandler(filepath, mode='w')
    formatter = logging.Formatter('[%(asctime)s] [%(levelname)s] %(message)s')
    handler.setFormatter(formatter)
    self.log.addHandler(handler)
    console = logging.StreamHandler()
    console.setFormatter(formatter)
    if console_debug:
        console.setLevel(logging.DEBUG)
    else:
        console.setLevel(logging.INFO)
    self.log.addHandler(console)

[wangwei357]

It‘s seems that : myself logger file handler was force closed when setup the pyxcy application. and the affected from class PyXCP(Application) which inherit from traitlet lib. But I can't locate the cause code in traitlet .

[christoph2]

Yes, pyXCP uses traitlets under the hood, which is used by projects like IPython and Jupyter, so the likehood that there is a hidden bug is really small. Maybe there's a similar logging issue on Stackoverflow?
For now you may just swallow the exception:

try:
   # Offending code goes here.
except ValueError as e:
   print(f"Logging error {e!r}")

[christoph2]

The effect of closing can be easily observed...

python -c "import sys as s;s.stdout.close();print('hello')"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ValueError: I/O operation on closed file.

[wangwei357]

@christoph2 Can I disable pyxcp logger function? Now this feature applies to my use of other programs

[christoph2]

At the first sight, there are two places which could be commented out:

pyxcp/pyxcp/config/__init__.py

Line 941 in 0bcf714

# Remove any handlers installed by `traitlets`.

pyxcp/pyxcp/config/__init__.py

Line 956 in 0bcf714

self.log.addHandler(rich_handler)

If it helps, I'll add a configuration option.

[christoph2]

Also interesting would be something like

logger = logging.getLogger("my-own-logger")

as a starting point for your logging stuff.

[christoph2]

I've added a small test for a already configured root handler, if so use it.
Hope it helps.

[wangwei357]

@christoph2 Thansk for your support ! I had try your method, It can solve the problem of error, but the function of affecting the file output to the file is no longer available. I use debug to locate the problem statement from this location.
https://github.com/christoph2/pyxcp/blob/41d6fd9cc96d4c5fd830fedd30886ad818468c28/pyxcp/config/__init__.py#L970C3-L970C62

    def read_configuration_file(self, file_name: str, emit_warning: bool = True):
        self.legacy_config: bool = False

this self.read_configuration_file(file_name,False) in _read_configuration(self, file_name: str, emit_warning: bool = True) -> None
made my other logger file handler error

[wangwei357]

Fortunately, I solved the problem by adjusting the search. However, this logger is redirected to the root logger, I do not know if there will be other effects

    def start(self):
        if self.subapp:
            self.subapp.start()
            exit(2)
        else:
            # self._setup_logger()
            if logging.getLogger().hasHandlers():
                self.log = logging.getLogger()
            else:
                self._setup_logger()
            self._read_configuration(self.config_file)

[christoph2]

if logging.getLogger().hasHandlers():
    self.log = logging.getLogger()
else:
    self._setup_logger()
self._read_configuration(self.config_file)

vs.

self._read_configuration(self.config_file)
if logging.getLogger().hasHandlers():
    self.log = logging.getLogger()
else:
    self._setup_logger()

Really makes a difference?

The other thing that came to mind: I could add a use_existing_logger configuration option.

[wangwei357]

Yea，It's make difference