See #2032 for some answers.

I'm not sure this is all that confusing, it's always been pretty obvious which kind of ellipsis is used for what.

• Pack expansion in code uses code ....
• A textual ellipsis in body text to mark a trailing-off thought is not something we should really have in the standard, but if we did, it'd use (text-mode) \ldots. The closest we come to this is an omission ellipsis in example code, for which I made the \commentellip macro.
• In maths/symbolic contexts, the ellipsis matches the operator that's being folded: the comma uses \dotsc ("c" for "comma") from AmsMath (or, less desirably but equivalently, \ldots), and other binary operators like + and - use \dotsb ("b" for "binary operator") from AmsMath (or, less desirably but equivalently, \cdots). I would strongly suggest AmsMath's consistent family of ellipses rather than the lower-level LaTeX builtins.
• Diagonal and vertical ellipses exist, too, for various matrix and table situations.

Documented in the wiki at: https://github.com/cplusplus/draft/wiki/Specification-Style-Guidelines#Formatting

the comma uses \dotsc ("c" for "comma") from AmsMath

In contexts involving both packs and ellipses, using \dotsc / \ldots is confusing and borderline unreadable; using \cdots is much clearer.

For example, [temp.variadic]p8:

The instantiation of a pack expansion that is neither a sizeof... expression nor a fold-expression produces a list of elements E1, E2, · · · , EN, where N is the number of elements in the pack expansion parameters.

The choice to use \cdots here was intentional to avoid the ellipsis being confused with a pack expansion ellipsis.

@zygoloid, thanks. I've added this corner of the cases to the wiki page. Is there any reason to keep this issue open, then?

The guidance in the wiki leaves me with one question: what about ellipses outside of math/code? I recently replaced a bunch of those in my fork of the document because cxxdraft-htmlgen's sentence splitting can get confused by them.

So for me, it would be great if such raw ellipses outside math/code were also systematically replaced with some command, because it would be one less patch to maintain.

@Eelis: There should be no raw "..." in plain prose text. I'm preparing a patch.

@zygoloid: OK, fair enough, but I'd say that's a corner case that represents a lose-lose-lose situation. I'm sympathetic to your chosen solution, but I'd say that it doesn't affect the general rules I outlined above.