Comment on attachment 8568765 [details] [diff] [review] Drop the parent arg from JS_NewObjectWithGivenProto and introduce a JS_NewObjectWithGivenProtoAndParent for the few cases that still pass in a custom parent Review of attachment 8568765 [details] [diff] [review]: ----------------------------------------------------------------- ::: js/src/ctypes/CTypes.cpp @@ +909,5 @@ > MOZ_ASSERT(fnproto); > > // Set up ctypes.CType.prototype. > + RootedObject prototype(cx, JS_NewObjectWithGivenProtoAndParent(cx, &sCTypeProtoClass, > + fnproto, parent)); *evil eye* ::: js/src/jsfriendapi.h @@ +66,5 @@ > // attached to them. > extern JS_FRIEND_API(JSObject *) > JS_NewObjectWithoutMetadata(JSContext *cx, const JSClass *clasp, JS::Handle<JSObject*> proto); > > +// Like JS_NewObjectWithGivenProto but allows passing an explicit parent argument. Don't use this; it's deprecated. If you can think of a way to squirrel Deprecated into the function name, that'd be awesome. The comment's nice, but we both know most people won't read it. I'm totally fine making it JS_DeprecatedNewObjectWithGivenProtoAndParent so it's obnoxious enough people will want to use it even less, if you are!

> *evil eye* Yeah, so, I was going to try to go through those remaining consumers, at some point.... > I'm totally fine making it JS_DeprecatedNewObjectWithGivenProtoAndParent Done.

(In reply to Not doing reviews right now from comment #3) > > *evil eye* > > Yeah, so, I was going to try to go through those remaining consumers, at > some point.... This was an evil eye at the code, not at you, and not at happening to preserve the case for now. :-) Just to be clear. In general, as long as we make forward progress on a problem, even if it's incomplete forward progress, I'm satisfied. We have more good code after this, and no good code has become bad. That's a win for me.

> This was an evil eye at the code, not at you Sure, I got that. ;) You should look at bug 1136523.