NavigatorOnline.onLine IDL attribute is not exported to bikeshed's database

[dbaron]

NavigatorOnline.onLine is not exported to bikeshed's database as an IDL attribute. (It's exported as the navigator.online dfn, though.) This means that a reasonable way to reference it in bikeshed like {{NavigatorOnline/onLine|navigator.onLine}} or {{Navigator/onLine|navigator.onLine}} doesn't work. See the following test bikeshed document:

<pre class="metadata">
Title: Web Platform Design Principles
Group: tag
Shortname: design-principles
Repository: w3ctag/design-principles
Status: ED
Level: none
ED: https://w3ctag.github.io/design-principles/
Editor: Sangwhan Moon, Invited Expert, https://sangwhan.com
Abstract: This document contains a set of design principles to be used when designing Web Platform technologies. These principles have been collected during the Technical Architecture Group's discussions in <a href="https://github.com/w3ctag/design-reviews/">reviewing</a> developing specifications. We encourage specification designers to read this document and use it as a resource when making design decisions.
</pre>

<div class="example">
If a JavaScript task has accessed the {{NavigatorOnline/onLine|navigator.onLine}} property,
and browser's online status changes,
the property won't be updated until the next task runs.
</div>

which produces the output:

LINK ERROR: No 'idl' refs found for 'onLine'.
{{NavigatorOnline/onLine|navigator.onLine}}
 ✔  Successfully generated, with 1 linking errors

We worked around this in w3ctag/design-principles#242.

[annevk]

@plinss is this something you could help with? I thought IDL was taken care of automatically.

[domenic]

The intent is to mark up all our <dfn>s explicitly, so that we can transition away from the fallible heuristics Shepherd uses. It looks like this was missed in #5916. #5957 doesn't cover it either because the <dfn> is outside the IDL fragment. /cc @dontcallmedom.

[dontcallmedom]

https://raw.githubusercontent.com/w3c/webref/master/ed/dfns/html.json correctly identifies it with the proper dfn-type. and as a result #5956 would fix this I believe.

Also, once speced/bikeshed#1761, the right dfn type will be exposed to bikeshed and respec specs independently of whether or when #5956 gets merged.

[dontcallmedom]

hmm... looking into #5956 it would probably need to be tweaked since right now the <dfn> would try to expose the full navigator.onLine as the attribute, rather than just onLine - will flag it there.

[plinss]

Shepherd does parse IDL and uses that to classify existing anchors where names match, but it doesn't generate dfns that don't exist. Having explicit markup on dfns is preferred. Bikeshed uses the same IDL parser and similar heuristics to generate the proper markup on the output spec. Ideally wattsi should do something similar.

[annevk]

@dontcallmedom FWIW, I would not mind dropping navigator. from the text. If that helps I can prepare a PR. That's an old-style leftover.

[dontcallmedom]

I have just suggested dropping navigator. in https://github.com/whatwg/html/pull/5956/files#r503712458

[annevk]

Okay, that also works.