01:30
<MikeSmith>
in documentation in interfaces defined with Web IDL in specs, is it OK to describe instances as “objects that are instances of the Foo interface”?
01:30
<MikeSmith>
*documentation on interfaces
01:32
<TimothyGu>
MikeSmith: technically it's "objects that implement the Foo interface" per https://heycam.github.io/webidl/#implements,https://heycam.github.io/webidl/#implements, but isn't it easier to say "Foo objects"
01:33
<MikeSmith>
TimothyGu: well the specific context I am trying to deal with is a general case, so I don’t have one specific Foo object in this context.
01:34
<MikeSmith>
lemme show you the exact wording I have...
01:34
<MikeSmith>
> See _Iterator-like methods_ in _Information contained in a WebIDL file_ for information on how an iterable declaration in an interface definition causes entries(), forEach(), keys(), and values() methods to be exposed in objects that are instances of the interface.
01:35
<MikeSmith>
where _Iterator-like methods_ hypertext is a link to https://wiki.developer.mozilla.org/en-US/docs/MDN/Contribute/Howto/Write_an_API_reference/Information_contained_in_a_WebIDL_file#Iterator-like_methods
01:35
<TimothyGu>
Okay. Yeah, "objects that implement the interface" would be the technically proper wording
01:36
<MikeSmith>
OK super
01:36
<MikeSmith>
thanks much
01:37
<MikeSmith>
oh
01:37
<MikeSmith>
"exposed *from* objects that implement the interface"?
01:38
<MikeSmith>
what preposition is best there?
02:24
<Domenic>
If this is MDN documentation, I have a long-standing pet peeve to remove all uses of the term "interface", since it it is a COBOL-ism holdover and the correct thing in JavaScript is "class".
02:25
<Domenic>
I've tried suggesting this before and been convinced it will be easier to change Web IDL first before we can convince the MDN maintainers to change.
03:07
<MikeSmith>
Domenic: yeah I personally would be much happier if we could make it align with JavaScript
03:08
<MikeSmith>
that’d certainly be more clear for web developers
10:38
<annevk>
I would love to see that change in IDL
13:55
<Domenic>
I want to tackle some easier ones first, in particular https://github.com/heycam/webidl/issues/350
14:56
<MikeSmith>
if we wanted to get more changes made to the IDL spec more quickly, my impression is that it could happen if we had somebody willing to put (more) money into a contract to pay for the necessary work
14:56
<MikeSmith>
or maybe not and the real bottleneck is still just not having enough people to work on it, regardless of money
15:00
<annevk>
I think the problem is in part that the payoff is not direct and that there's a lot of cross-cutting concerns.
15:01
<MikeSmith>
yeah
15:04
<MikeSmith>
so by the way about the MDN annos, I notice a kind of bad usability/discoverability problem that happens in a small number of cases
15:04
<MikeSmith>
if you go to https://html-mdn.herokuapp.com/output/multipage/semantics-other.html#selector-read-only
15:04
<MikeSmith>
... there is actually an MDN anno for ":read-only"
15:05
<MikeSmith>
but it’s not visible right there at point of use
15:05
<MikeSmith>
not aligned there
15:06
<MikeSmith>
the reason is, if you go to https://html-mdn.herokuapp.com/output/multipage/semantics-other.html#pseudo-classes and open the second anno there
15:07
<MikeSmith>
... you’ll see it’s one giant long anno with info for every single pseudo-class in that dl/dt/dd list in the body of the spec
15:09
<MikeSmith>
but no normal reader would ever realize that they need to scroll all the way back up there to find the anno
15:10
<MikeSmith>
this is why, when I first added the MDN annos, I made them un-minimized by default — because otherwise, we run into discoverability issues like that
15:11
<MikeSmith>
so anyway if anybody has some good ideas about how we could better handle discoverability of the annos in cases like that, lemme know
15:38
<Domenic>
MikeSmith: "Pseudo-classes / No support in current engines. / No support data" seems not so good
15:39
<MikeSmith>
Domenic: yeah that’s a bug
15:39
<Domenic>
MikeSmith: interesting, I guess the problem there is that we're going up to child-of-body, which is the entire <dl>.
15:40
<MikeSmith>
yeah, that is the cause
15:40
<MikeSmith>
maybe we can special-case <dt>
15:40
<Domenic>
Yeah that's what I'm wondering
15:41
<MikeSmith>
yeah will look into that
15:41
<Domenic>
If we had a before/after diff to judge the impact i'd be comfortable with that
15:41
<Domenic>
To make sure it doesn't have worse impacts elsewhere
15:41
<MikeSmith>
right
15:41
<MikeSmith>
yeah good point
15:41
<MikeSmith>
I can do it in a way that gives us a diff
16:30
<TabAtkins>
Hm, why go up to child-of-body, rather than just nearest-ancestor-not-autoclosed-by-an-aside?
16:42
<Domenic>
That sounds more complicated :)
16:43
<Domenic>
If it was easy to know what that list was then we could probably do it
19:38
<TabAtkins>
If only there were a spec around that described the auto-closing behavior, or perhaps an editor of said spec... ^_^
22:39
<shu>
Domenic: mind elaborating the confusion on https://github.com/tc39/ecma262/issues/1524#issuecomment-611222892?
22:39
<Domenic>
shu: just confused/sad that we got to that conclusion given the arguments presented.
22:39
<shu>
AIUI the problem we want to solve was to not mislead -- currently, sometimes implementations mean a user agent, sometimes all user agents
22:40
<shu>
the compromise was to directly address that issue by enumerating all the places where implementation actually means host
22:40
<shu>
and linking to the actual host spec
22:41
<Domenic>
Sure. We've just lost a valuable word to discuss the distinction.
22:41
<shu>
that's fair -- the plan is to clarify that in the front matter somehow, non-normatively