Parked: design issues that would need a major #357

These issues all describe the same kind of change: the AST or the parsing model would have to change in a way that cannot ship in a minor. None of them breaks working code today, and none has attracted a single reaction in the years they have been open, so they are collected here rather than kept individually open.

• "combinators" found in nth-child expressions #14, :nth-child(2n + 1) parses the + as a combinator. The An+B of S microsyntax would need its own representation.
• API to return whether a selector has non-standard syntax. #135, an API to report whether a selector uses non-standard syntax, which also implies a strict parsing mode.
• invalid/non stable ast for combinator with comment #189, a comment between two compound selectors lands in raws rather than in the AST, so h2 /*c*/ h4 and h2/*c*/h4 disagree.
• broken ast for :foo() {} #202, :foo() with empty arguments produces a questionable child shape.
• broken ast #204, // comments reach this parser as selector text. Arguably postcss-scss should strip them before they get here.
• Add a selector equality check #228, a semantic equality check, so that :is(a, b) and :is(b, a) compare equal.

There is no major planned. This is a parking list, not a roadmap. The package is maintained for the tooling that depends on it, and the bar for a breaking AST change is a concrete problem that cannot be solved any other way.

If one of these blocks you, comment on it with a real use case and it can be taken back out.