improve setting global options to None

[mwageringel]

The __call__ method of global options is used as getter and setter. This ticket improves this method to properly handle values of None.

For example:

sage: from sage.structure.global_options import GlobalOptions
sage: class config(GlobalOptions):
....:     size = dict(default=42,
....:                 description='integer or None',
....:                 checker=lambda val: val is None or val >= 0)
sage: config.size(None)  # this should set the option to `None`
42

(See #30755 for an actual example where None is used as a value.)

This can be used as a workaround instead:

sage: config.size = None
sage: config.size() is None
True

This ticket makes config.size(None) work as a setter. Additionally, a deprecation is added for the keyword-argument syntax config.size(value=None), as it is not usually used and this will simplify the implementation a little.

Component: misc

Author: Markus Wageringel

Branch/Commit: 4c08256

Reviewer: Vincent Delecroix

Issue created by migration from https://trac.sagemath.org/ticket/30763

[mwageringel]

Author: Markus Wageringel

[mwageringel]

Commit: 4c08256

[mwageringel]

New commits:

4c08256

30763: allow setting global options to None

[mwageringel]

Branch: u/gh-mwageringel/30763

[mkoeppe]

comment:2

Wouldn't it be simpler to just use a unique object instead of None as the default argument? Then no change of the interface is necessary

[mwageringel]

comment:3

Well, yes, but I think that it is common to use None as a default value/no preferred choice. For example, in Sage, this is the case for the preferences of the display manager (though, these are distinct from GlobalOptions).

Note that, in the example from the description, config.size() still keeps working as a getter, so the interface does not really change.

[mkoeppe]

comment:4

Setting new milestone based on a cursory review of ticket status, priority, and last modification date.

[mkoeppe]

comment:5

Setting a new milestone for this ticket based on a cursory review.

[mkoeppe]

comment:6

Stalled in needs_review or needs_info; likely won't make it into Sage 9.5.

[videlec]

comment:7

The proposed change is reasonable and simplifies (actually will simplify in the future) the implementation of __call__.

[videlec]

Reviewer: Vincent Delecroix

[vbraun]

Changed branch from u/gh-mwageringel/30763 to 4c08256