gh-117941: Fix argparse issue-Changed call method of BooleanOptionalAction class in argparse.py #117941

[jjpapa36912]

• Issue: BooleanOptionalAction does not correctly resolve arguments that are defined as starting with "--no-" #117941

[ghost]

All commit authors signed the Contributor License Agreement.

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[jjpapa36912]

I was wondering if when does it review.

[encukou]

The PR needs tests and documentation for the new behaviour.

However, I'm not convinced that we should do this; I think it would be better to deprecate adding boolean options that start with --no-. The auto-added --no-no- option itself is surprising.

[encukou]

Please see the devguide on how to write a NEWS entry.

[jjpapa36912]

The PR needs tests and documentation for the new behaviour.

However, I'm not convinced that we should do this; I think it would be better to deprecate adding boolean options that start with --no-. The auto-added --no-no- option itself is surprising.

Thank you for answer my question. I understand your explanation. But I think that to eliminating exception or abnormal situation is very important. In that sense, isn't it better to fix the code?

[encukou]

If you can fix it properly, sure. It will not be easy.
As I said above, the PR needs tests and documentation for the new behaviour. When you write the tests, you'll probably discover many cases where this PR results in unexpected behaviour.

On the other hand, if we clearly say that adding --no- options directly doesn't work, there will be fewer fewer special cases that argparse needs to handle, test and document. Less code is better for everyone ;)

[jjpapa36912]

If you can fix it properly, sure. It will not be easy. As I said above, the PR needs tests and documentation for the new behaviour. When you write the tests, you'll probably discover many cases where this PR results in unexpected behaviour.

On the other hand, if we clearly say that adding --no- options directly doesn't work, there will be fewer fewer special cases that argparse needs to handle, test and document. Less code is better for everyone ;)

Oh, thank you for your answer. I have one question. If the option string starts with "--no-" I think the current BoolOptionalAction always returns "False". What do you think about this?

[encukou]

Right; I don't think that was the intent.
But, I can see someone trying the following and finding that it works for them (with a weird auto-added --no-no-foo option, which you can ignore):

>>> import argparse
>>> parser = argparse.ArgumentParser()
>>> parser.add_argument('--no-foo', default=True, action=argparse.BooleanOptionalAction)
>>> parser.parse_args()
Namespace(no_foo=True)
>>> parser.parse_args(['--no-foo'])
Namespace(no_foo=False)

The documentation doesn't clearly say that this is wrong:

You may also specify an arbitrary action by passing an Action subclass or other object that implements the same interface. The BooleanOptionalAction is available in argparse and adds support for boolean actions such as --foo and --no-foo