Use of print() over python's logging and warning system

[ligerlac]

Hello everyone,
hls4ml often uses print() over python's warnings and logging modules. For example, in hls4ml/converters/__init__.py:

try:
    lib = importlib.import_module(__name__ + f'.{model_type}.' + module[:-3])
    ...
except ImportError as err:
    print(f'WARNING: Failed to import handlers from {module}: {err.msg}.')

we only use the keras converter in our project and would like to suppress the warning that torch is not installed. However, this is quite cumbersome as it is a print('WARNING: ...') statement, not an actual python warning.

I can create a PR to transform these makeshift warnings into actual warnings. But I first want to discuss here if there is maybe a good reason not to do that.

Cheers

[vloncar]

We discussed this and will do it after the 1.0 release. We'll change all print statements and replace some with warnings.warn() and some with logging.warn(). Also some form of debugging level of logging will be introduced to track the transformations

[ligerlac]

Okay, sounds good. Let me know if I can help you with that.

[vloncar]

Sure, I'll let you know. If you're keen on doing some maintenance work in the meantime we can figure something else out :-)