Allow deprecate_nonkeyword_arguments to not specify version, show class name

[MarcoGorelli]

xref #41511 (comment)

as request, ping @jreback @simonjayhawkins

I've removed the ability to pass an integer to allowed_arguments as it added complexity, was only used once in the codebase anyway, and wasn't even covered by tests

Now, if version is None, then it'll just print "In a future version of pandas", it'll show the class name (e.g. Series.drop_duplicates instead of just drop_duplicates) and won't print 'self' in the warning message

[simonjayhawkins]

nbd, but this and the line above could be moved to the outer scope.

[MarcoGorelli]

indeed - have fixed, thanks!

[simonjayhawkins]

same for the message creation

[jorisvandenbossche]

Now, if version is None, then it'll just print "In a future version of pandas"

Personally, I think it is good practice to mention the version when it will be removed/changed, so forcing us to do it here is maybe actually fine.

[jreback]

very minor point (can be in a followup).

[jreback]

optional

[jreback]

Now, if version is None, then it'll just print "In a future version of pandas"

Personally, I think it is good practice to mention the version when it will be removed/changed, so forcing us to do it here is maybe actually fine.

disagree this locks us in, and we don't do this anywhere now. this should be a deliberate decision.

[jorisvandenbossche]

disagree this locks us in, and we don't do this anywhere now. this should be a deliberate decision.

1) we are already doing this for all other cases of deprecated positional keyword arguments, and 2) this has been decided in our version policy

[MarcoGorelli]

Hi @jorisvandenbossche - deprecated positional keyword arguments are quite rare in pandas, there's not been many so far, and most other FutureWarnings don't mention specific versions, e.g.

            warn(
                "The `inplace` parameter in pandas.Categorical."
                "set_categories is deprecated and will be removed in "
                "a future version. Removing unused categories will always "
                "return a new Categorical object.",
                FutureWarning,
                stacklevel=2,
            )

It's true that the version policy says

Deprecations will only be enforced in major releases. For example, if a
behavior is deprecated in pandas 1.2.0, it will continue to work, with a
warning, for all releases in the 1.x series. The behavior will change and the
deprecation removed in the next major release (2.0.0).

though it doesn't say that warnings should specify versions.

I have no preference here though, am happy to go with whatever you all want