GeezerVison

Many other PEPs have revised the typing spec since PEP 484

• Any is now its own class, rather than being an instance of _SpecialForm
• Numerous other _SpecialForms have now been added. Rather than attempting to list them all (which would just go out of date again), I've just added "and others" at the end of an abbreviated list.
• Instances of ForwardRef aren't actually accepted by type checkers as type annotations, as far as I know, so that was somewhat confusing here.

Many other public helper functions have been added that aren't listed here, so I just made it an abbreviated list and put "and others" at the end

Guido mentioned in #101827 that he didn't like this style of code example, where it's not really a REPL example, but also isn't something you'd ever put in an executable .py file either. I agree that using an explicit assert is better for this kind of code example.

If AttributeError or TypeError is encountered, the __final__ attribute won't be set.

This is a partial revert of #104945.

#104945 removed any mention of the functional syntax from this docstring. I agree with removing the mention of Python 3.5 from the docstring (also done in #104945), but I disagree with removing any mention of the functional syntax. The functional syntax is supported by type checkers, and still quite common in many codebases.

However, I do think we should remove any mention of the keyword-argument syntax, since this way of creating NamedTuples isn't supported by any type checker that I know of.

All points except this one have examples. Is it intentional?

Yeah, they're deprecated ;p