logger_setup: Logging configuration used by accessible apps

This repository contains a collection of logging functions used mainly by applications from accessible apps.

While you are more than welcome to include this module standalone, it was meant specifically for use with continuum projects and might be considered somewhat uncompromising otherwise.

Features

• A CrashLogger to handle unplanned application crashes
• Initialization of debug and error logs
• Optionally logs all unhandled tracebacks for later analysis
• Support for remote logging to a syslog server

Usage

Call setup_logging during application startup on the main thread:

import logging
from logger_setup import setup_logging

setup_logging(error_log="errors.log", debug_log="debug.log")
logger = logging.getLogger(__name__)
logger.info("Application started")

The function returns Python's root logger and sets its level to DEBUG. Application and library loggers that propagate to the root use its handlers. Importing the package does not install handlers or replace sys.excepthook.

error_log receives WARNING and higher, despite its name. debug_log receives DEBUG and higher. Both filenames are optional, open in write mode, and require existing parent directories. Opening a file replaces previous contents, even if a later setup step fails; file contents are not rolled back.

The console handler writes to stderr at INFO by default. Disable it with console_level=None. A packaged application that discards stderr also discards that console output; configure an error file for retained diagnostics.

remote_address=(host, port) enables the standard SysLogHandler, with an ERROR threshold by default. Supply application_name and application_version alongside the address. Configuring this handler does not confirm remote message delivery.

Exception coverage and lifetime

log_unhandled_exceptions=True installs sys.excepthook to log uncaught main-thread Python exceptions and then call sys.__excepthook__. The delegate is resolved when the exception occurs. app_framework's later debug bootstrap uses this to install its debugger delegate.

With error_log and log_crashes=True, faulthandler writes fatal traceback dumps directly to the error-log file descriptor. These are raw dumps without the normal logging format. This diagnoses crashes; it does not recover from them. Keep the stream open while faulthandler uses it, and disable or retarget fault handling before independently closing or replacing the stream.

The package does not install a thread exception hook or handle every GUI callback. app_framework logs its executor and wx CallAfter failures separately. RepeatingTimer also catches and logs callback exceptions. Those mechanisms are distinct from sys.excepthook and fatal traceback dumps.

Configure logging before starting worker threads. Explicit process logging shutdown is available as logger_setup.logger_setup.shutdown(); it invokes Python's process-wide logging.shutdown, which closes unrelated handlers too. It is not a private per-package cleanup API.

Use through app_framework

app_framework supplies identity and paths under its application data directory, preserves the preceding error log as errors.log.old, enables debug logging for source/debug runs, and passes the application's optional remote address. Those policies belong to the framework; logger_setup does not rotate or retain logs. The standard application startup path configures logging once.