| 13:35 | <tobie> | annevk: re https://github.com/heycam/webidl/issues/240, I'm struggling to understand this wrt dictionaries, too. |
| 13:36 | <tobie> | annevk: you have an extra category in there which is dictionary Foo {}, no? Which you don't (yet) have with records. |
| 13:36 | <annevk> | tobie: I think we should not have a dfn for the thing ending in type |
| 13:38 | <tobie> | annevk: well, it certainly makes things easier. |
| 13:38 | <annevk> | tobie: in your example Foo would be marked up as Bikeshed dictionary I think |
| 13:38 | <annevk> | tobie: for xref purposes |
| 13:38 | <tobie> | annevk: oh. |
| 13:39 | <tobie> | annevk: right. |
| 13:39 | <annevk> | tobie: records would have that only through typedef I guess |
| 13:39 | <tobie> | annevk: that's the example TabAtkins mentioned, right? |
| 13:39 | <tobie> | annevk: <dfn>record<Foo></dfn> |
| 13:40 | <tobie> | (pardon the absence of HTML escaping.) |
| 13:41 | <annevk> | tobie: well, you can't really name a record other than through a typedef I think |
| 13:41 | <tobie> | annevk: I guess I think of dictionary Foo when someone talks of a dictionary type. So I'm not sure what it means for records |
| 13:41 | <tobie> | annevk: to your point |
| 13:43 | <annevk> | So you have records, dictionaries, sequences as concepts / IDL instances |
| 13:43 | <tobie> | annevk: I would say: {{Foo}} is a [=dictionary type=]. But record<DOMString, DOMString> is a [=record=]. |
| 13:43 | <annevk> | They are also types in that you can distinguish them |
| 13:44 | <tobie> | annevk: mmm. true. |
| 13:44 | <annevk> | I think both of those are types of sorts |
| 13:44 | <annevk> | Since they are not instances |
| 13:45 | <annevk> | And I guess another aspect is that there is hierarchy of sorts |
| 13:45 | <annevk> | There is a general dictionary type, but also more specific variants such as Foo |
| 13:46 | <tobie> | I guess that makes sense. |
| 13:46 | <annevk> | Though you can't type-test for Foo, so maybe there is only generic |
| 13:47 | <tobie> | I think what you're suggesting is there's no reason to distinguish in markup between: [=dictionary type=] and [=dictionary=] [=type=]. |
| 13:48 | <tobie> | (or even [=dictionary=] type) |
| 13:49 | <annevk> | Yeah, I'm leaning towards the latter myself |
| 13:50 | <annevk> | That is the convention the platform has used thus far |
| 13:50 | <annevk> | For "concepts" |
| 13:50 | <annevk> | We should maybe get more formal about types too though |
| 13:51 | <annevk> | "Type(x) is URL" might be interesting, but not sure and it has not been needed thus far |
| 23:09 | <zcorpan> | MikeSmith: how about we check the usage statistics of the doctype again in a few years? |
| 23:11 | <zcorpan> | MikeSmith: you would have annoyed 400 people in two days had you changed the validator already, right? :-) |
| 23:12 | <MikeSmith> | zcorpan: yup |
| 23:13 | <MikeSmith> | thanks much for the data |
| 23:13 | <MikeSmith> | I’ll keep the use counter on in the checker |
| 23:13 | <zcorpan> | np |
| 23:14 | <MikeSmith> | so yeah we can look again in few years but I suspect we are always going to have some number like that as long as we still have XSLT etc |
| 23:18 | <zcorpan> | i guess so |