For now we'll just support having either non-Chrome-only or Chrome-only constructors, not both at the same time. We'd have to generate separate constructors if we ever do (and make Xrays support them etc).

Comment on attachment 832150 [details] [diff] [review] v1 Review of attachment 832150 [details] [diff] [review]: ----------------------------------------------------------------- ::: dom/bindings/Codegen.py @@ +2223,2 @@ > """ > object is the name of a JSObject* Maybe document context too

Comment on attachment 832150 [details] [diff] [review] v1 > + }""" % GetAccessCheck(self.descriptor, "cx", "obj") I think you want a linebreak before the """ here. r=me with the comments ms2ger asked for. What actually ends up preventing having both a [Constructor(), ChromeConstructor()] situation, by the way?

https://hg.mozilla.org/integration/mozilla-inbound/rev/e36faba57f46 (In reply to Boris Zbarsky [:bz] from comment #2) > What actually ends up preventing having both a [Constructor(), > ChromeConstructor()] situation, by the way? We check for differing extended attributes on overloads.

Peter, can you please document this at https://developer.mozilla.org/en/Mozilla/WebIDL_bindings ?