23602 2013-10-23 06:39:25 +0000 "optional any" is nonsense that should be disallowed 2016-08-29 13:30:41 +0000 1 1 1 Unclassified WebAppsWG WebIDL unspecified PC All NEW P2 normal --- 1 bzbarsky cam annevk mike public-script-coord public-webapps-bugzilla oldest_to_newest 95156 0 bzbarsky 2013-10-23 06:39:25 +0000 Or perhaps non-optional any is nonsense? In any case, passing "undefined" to an "any" argument should presumably simply preserve the "undefined" value, and it looks like http://www.w3.org/TR/WebIDL/#es-any in fact specifies that. Given that we no longer differentiate between "undefined" and "omitted", what should passing "undefined" to an "optional any" do? It seems weird to treat that as "omitted" when the "any" type can in fact represent the value. I would argue that we should simply not have optional vs non-optional any: have only "any", where it can always be not passed or explicitly passed as "undefined", in which case it coerces the value to "undefined". 95159 1 cam 2013-10-23 06:49:15 +0000 If we do that then we should do something similar for dictionary members, making dictionary A { any x; }; interface B { void f(A a); }; b.f({ x: undefined }); mean that x is present with the undefined value, rather than not present as is currently specced. 95160 2 bzbarsky 2013-10-23 06:53:08 +0000 Yes, I agree. Basically, either we should remove "undefined" from the allowed value set of "any" (and then all "any" arguments should be optional and undefined becomed "not present") or we should keep "undefined" in the value set and simply have "undefined" become the "undefined" value. 95168 3 annevk 2013-10-23 09:42:26 +0000 I think the latter solution of comment 2 makes the most sense. 127254 4 annevk 2016-08-29 06:32:23 +0000 Did this got resolved? I think we made some changes to undefined handling since 2013-10. 127257 5 bzbarsky 2016-08-29 13:30:41 +0000 I don't think this got resolved, no. I never got any feedback from other implementors on my proposal at https://lists.w3.org/Archives/Public/public-script-coord/2014JanMar/0176.html so I just went ahead and implemented a modified version of it in Gecko; see <https://bugzilla.mozilla.org/show_bug.cgi?id=985536>. The difference between what I implemented and the proposal on the list is whether trailing non-optional "any" affects the argc check behavior.... But in the spec right now, as far as I can tell, "optional any" can only take on the state "not passed", and can't have the state "undefined". That's what I aimed to fix in my Gecko changes; in Gecko the "optional" bit just affects .length and argc checks, but the value is always considered "passed" and may just have the value "undefined".