Alternatively, do we just want to add an idlharness.js test for:

partial interface CSSStyleDeclaration {
  [CEReactions] attribute [TreatNullAs=EmptyString] CSSOMString _camel_cased_attribute;
};

partial interface CSSStyleDeclaration {
  [CEReactions] attribute [TreatNullAs=EmptyString] CSSOMString _webkit_cased_attribute;
};

partial interface CSSStyleDeclaration {
  [CEReactions] attribute [TreatNullAs=EmptyString] CSSOMString _dashed_attribute;
};

and assume that all properties defined in CSS2 are supported CSS properties? (Plus, I guess, some -webkit- property?)

That would get us coverage of not just this but all the things that idlharness.js asserts about attributes.

I think really ideally we'd use the data in Reffy to generate those partial interface definitions as part of the IDL defined in each spec that defines CSS properties, hmm. @foolip?

I just sent a proposal to parse the CSS specs for properties and generate syntax tests (https://lists.w3.org/Archives/Public/www-style/2019May/0013.html), which I think it could be used to know which attributes to test, if that's desirable.

@annevk it does have idlharness.js tested, but there's a bunch of IDL definitions where there are attributes defined for each supported CSS property (and CSSOM doesn't enumerate them), and we don't currently create an attribute for every CSS property defined in any spec.

Ah right, it's defined in prose these days? That would indeed create this issue. 😟