Hello,

Thank you for all the hard work on Pyomo!

We are having some trouble with pyomo not respecting our custom logging configuration, and unconditionally emitting its logs to stdout in a hardcoded format.

Consider, for example, how pyomo ignores a Python logging configuration that sends all logs into a StringIO buffer with a custom format message:

>>> import logging
>>> from io import StringIO
>>> import pyomo.environ
>>> 
>>> string_io = StringIO()
>>> handler = logging.StreamHandler(stream=string_io)
>>> logging.basicConfig(format='custom:%(message)s', handlers=[handler])
>>> 
>>> logging.getLogger("pyomo").warning("pyomo warning")
WARNING: pyomo warning
>>> logging.warning("example warning")
>>> string_io.getvalue()
'custom:pyomo warning\ncustom:example warning\n'
>>> 

Unfortunately, the "pyomo" logger emits its log to both stdout and the string_io buffer...Additionally, the log it emits to stdout also does not respect the custom format configured in the logging.basicConfig call---the message pyomo emits to stdout doesn't have custom as its prefix.

I think the root cause of this issue is that pyomo adds a custom handler to itself in common/log.py:

This custom handler, evidently, is responsible for the log messages that always wind up in stdout in the wrong format irrespective of the users's choices. The Python docs advise against adding a handler in a library's logger because it means that developers are unable to configure logging to suit their needs:

Note It is strongly advised that you do not add any handlers other than NullHandler to your library’s loggers. This is because the configuration of handlers is the prerogative of the application developer who uses your library. The application developer knows their target audience and what handlers are most appropriate for their application: if you add handlers ‘under the hood’, you might well interfere with their ability to carry out unit tests and deliver logs which suit their requirements.

In our application, for example, we want all log messages to have a custom JSON format and be sent to stderr. But, we wind up with many logs from pyomo in stdout without our custom JSON formatting.

We can work around this problem by preemptively running import pyomo.environ and then clobbering all of the handlers on the "pyomo" logger, but---of course---that solution is pretty brittle.

Thanks for your help.