[selectors-4] :nth-child and friends are either underdefined or defined wrong

[SimonSapin]

All of these

• https://drafts.csswg.org/selectors/#the-nth-of-type-pseudo
• https://drafts.csswg.org/selectors/#the-nth-last-child-pseudo
• https://drafts.csswg.org/selectors/#the-nth-of-type-pseudo
• https://drafts.csswg.org/selectors/#the-nth-last-of-type-pseudo

have a definition that includes “the An+Bth element” with no details on what that means. This is either wrong (since this is using “the element” singular, where as multiple elements could match) or at least under-defined. A and B are from https://drafts.csswg.org/css-syntax/#anb, but what is n?

This is a regression from Selectors Level 3, which has a precise definition:

https://drafts.csswg.org/selectors-3/#nth-child-pseudo

The :nth-child(an+b) pseudo-class notation represents an element that has an+b-1 siblings before it in the document tree, for any positive integer or zero value of n, and has a parent element.

Notable edge cases that should be well-defined include n<0 and A=0.

[SelenIT]

Isn't the 'An + B' notation described by reference to the CSS Syntax module for all these definitions?

The An+B notation defines an integer step (A) and offset (B), and represents the An+Bth elements in a list, for every positive integer or zero value of n, with the first element in the list having index 1 (not 0).

The edge cases like A=0 are described in the next (though non-normative) section 6.1.

[SimonSapin]

Ah, I was misled by assuming that “CSS Syntax Module” only described the syntax of An+B. Either way, I think it wouldn’t hurt for Selectors 4 to repeat that.

The edge cases like A=0 are described in the next (though non-normative) section 6.1.

I think the part you quoted is sufficient. With A=0, An+B simplifies to B regardless of n, and so at most one element (the Bth) is represented.

The A=0 case would only be problematic if division by A was involved in the definition (as it is in my implementation, which therefore requires a code path dedicated to A=0).

[Loirooriol]

I think the meaning is understandable, but nitpicking I agree it's not perfect because CSS Syntax says the notation "represents the An+Bth elements in a list", so it's a set of elements, but CSS Selectors says “the An+Bth element”. Instead, in Selectors I would say something like “any one of the An+Bth elements” or maybe “an An+Bth element”.

[FremyCompany]

@Loirooriol make a pull request :)

[Loirooriol]

More nitpicking, what I said above is something like

E ∈ subList(A, B, inclusiveSiblings(E) ∩ match(S))

I don't like it much neither, because E is defined as belonging to the sublist, but the sublist depends on E. Not wrong because we know it's a sublist of elements, but I would prefer defining E in its domain and then impose the restriction: (if someone is interested, see Russell's paradox)

E ∈ Elements | E ∈ subList(A, B, inclusiveSiblings(E) ∩ match(S))

In English I would say

The :nth-child(An+B [of S]?) pseudo-class notation represents an element among its An+Bth inclusive siblings that match the selector list S.

[tabatkins]

Okay, I just pushed be7492c which attempts to fix this.

[SimonSapin]

Looks good, thanks Tab.

[Loirooriol]

There is a repeated sentence. See #1308 and be7492c#commitcomment-21954890