Issue 14760: logging: make setLevel() return handler itself for chained configuration (original) (raw)

Created on 2012-05-09 07:15 by techtonik, last changed 2022-04-11 14:57 by admin. This issue is now closed.

Messages (8)
msg160252 - (view)	Author: anatoly techtonik (techtonik)	Date: 2012-05-09 07:15
It would be convenient if instead of: hdlr = logging.StreamHandler() hglr.setLevel(logging.DEBUG) root.addHandler(hdlr) it would be possible to write: root.addHandler(logging.StreamHandler().setLevel(logging.DEBUG))
msg160253 - (view)	Author: Georg Brandl (georg.brandl) *	Date: 2012-05-09 07:42
-1. Attribute setters or mutating methods returning self is not a common pattern in Python. See list.sort().
msg160254 - (view)	Author: anatoly techtonik (techtonik)	Date: 2012-05-09 07:51
List is a standard type and it have None returned for a reson. Logging is a library with its own semantic and high level object. Here you don't need to return None to explicitly say that the object was modified in place.
msg160255 - (view)	Author: Georg Brandl (georg.brandl) *	Date: 2012-05-09 07:53
Well, can you find any other setter method of a "high level object" in the stdlib that returns self?
msg160256 - (view)	Author: anatoly techtonik (techtonik)	Date: 2012-05-09 07:58
Well, I can remember any other widely used high level objects from stdlib either. HTTP servers perhaps, but they are in a poor state.
msg160257 - (view)	Author: anatoly techtonik (techtonik)	Date: 2012-05-09 07:58
s/can/can't/
msg160258 - (view)	Author: anatoly techtonik (techtonik)	Date: 2012-05-09 08:02
Well, there is argparse, but it doesn't support method chaining. Whatever, it is just a proposal. If consistency inside stdlib covers calling conventions for bundled user-level libs - I am fine with that.
msg160295 - (view)	Author: Vinay Sajip (vinay.sajip) *	Date: 2012-05-09 14:18
I agree with Georg. Besides, as far as possible, the API should be consistent across versions of Python, unless something is needed to take advantage of added features. With your proposal, users could write library code which then would not work in older Python versions - and that's not something we want to encourage.