Sorry, I don't think it's an improvement.

Old sentence is clear: we return a rational number with value numerator/denominator.
But original numerator and denominator may have any sign, may be not integers, etc.

I agree with@skirpichev. We can leave "equals", but "where the denominator is positive" is incorrect or at least ambiguous. "Denominator" refers to the argument of theFraction constructor, not thedenominator property.

I also agree. It's as if we requiredenomintor to be positive (as an argument) which is not the case. So I suggest to revert this PR.

I opened pr to fix docstrings in Rational class:#136800 It's a accompanying pr to#122464, which fix sphinx docs (and stuck in review).

But maybe it's ok to combine both and include reversion of this pr? All that only affects docs and IMO - fine for backporting up to 3.13.

What do you think?

@serhiy-storchaka@skirpichev we discussed this at the sprint and decided that*denominator* (italics) referred to the argument, butdenominator (normal/roman) in the context of the sentence referred to the new argument. Regardless, it simplifies the arguments, so it is not always thedenominator argument that is set as the class attribute.

A

It says: "a new :class:Fraction instance with value equal tonumerator/denominator". "denominator" refers to argument, not to the property. Otherwise this sentence does not make sense. Then you say that in "where the denominator is positive" "denominator" refers to something other. But it clearly refers to the same thing as in previous expression. Your interpretation requires interpreting the same word in two different ways simultaneously. Even if it makes sense, it is very unclear and confusing. And to me, it just looks wrong.