04:51
<MikeSmith>
annevk: so the reason wattsi doesn’t report an error for <h5 id="image-map-processing-model"><span id="processing-model"></span>Processing model</h5> but does for the following:
04:51
<MikeSmith>
<h5 id="rel-alternate"><span id="link-type-alternate"></span>Link type "<dfn data-export=""
04:51
<MikeSmith>
data-dfn-for="link/rel,a/rel,area/rel" data-dfn-type="attr-value"><code
04:51
<MikeSmith>
data-x="rel-alternate">alternate</code></dfn>"</h5>
04:52
<MikeSmith>
... is that the second case has not only that <span id="link-type-alternate"></span> but also the <dfn><code data-x="rel-alternate">…
04:53
<MikeSmith>
which ends up as two IDs in the generated output
04:54
<MikeSmith>
and in that ase, wattsi is not reporting the <span id="link-type-alternate"></span> as an error — instead it’s reporting the error when it gets to that <dfn><code data-x="rel-alternate">
04:54
<MikeSmith>
...
04:55
<MikeSmith>
given all that, what do you want to do in this case?
04:56
<MikeSmith>
maybe you want it so that we report the error for that <dfn><code data-x="rel-alternate"> *only* if the preceding ID-ed element — the <span id="link-type-alternate"></span> — is non-empty?
04:57
<annevk>
Yeah that sounds reasonable
05:15
<MikeSmith>
annevk: well that turns out to be relatively hard, I think
05:15
<MikeSmith>
unless there’s something I am missing
05:15
<MikeSmith>
...because in the current code, there is nothing that’s saving any state about each node
05:16
<MikeSmith>
hmm
05:16
<MikeSmith>
well now that I say that out loud, I realize it’s obviously wrong
05:17
<MikeSmith>
...because it is at least saving the state information “this heading already has a span or dfn that results in an ID”
05:32
<MikeSmith>
annevk: also maybe you already know that, but I think the purpose of that code is solely for the TOC, and what to select for the markup output for that heading in the TOC
05:32
<MikeSmith>
anyway, I’ll get a patch together today
06:30
<MikeSmith>
annevk: so I am back to realizing is that current code intentionally does not error for <h5 id="rel-alternate"><span id="link-type-alternate">SOME TEXT CONTENT</span>Link type "<dfn data-export="" — but does for <h5 id="rel-alternate"><span id="link-type-alternate"></span>Link type "<dfn data-export=""
06:31
<MikeSmith>
that is, if the “nested” span or dfn has text content, no error; but if it has not text content, error
06:32
<MikeSmith>
it’s not an oversight or mistake; all the relevant code there was intentionally written to handle those cases in that way
06:32
<MikeSmith>
I do not understand why
06:33
<MikeSmith>
but what I was suggesting earlier above is effectively the exact opposite of what the current code was intentionally written to do
06:34
<MikeSmith>
therefore it seems pretty odd to just completely flip it opposite
06:35
<MikeSmith>
what seems less odd is to just drop the one condition/check that’s actually causing the error to be reported
06:35
<MikeSmith>
that is, what I suggested in https://github.com/whatwg/html/pull/5855#issuecomment-679079688
06:36
<MikeSmith>
in testing locally, doing that causes the rendered output for that heading in both the ToC and spec body to be exactly the same
06:36
<MikeSmith>
the target of the href remains the same
06:36
<MikeSmith>
too
06:38
<MikeSmith>
but the worry is that if we drop the check, there’s some other real problem case that check was (I guess) intended to catch that it won’t catch any more in the future
06:39
<MikeSmith>
and without having a regression test for that, it’s just kind of an act of faith that it’s not gonna be breaking something in the output later in some subtle/surprising/hard-to-catch way
06:40
<MikeSmith>
but in spite of that I think we should just go ahead and drop the check
06:41
<MikeSmith>
because it’s more important to unbreak the build than it is to worry about some potential problem that many not be a real problem for us in practice anyway
07:05
<annevk>
MikeSmith: that sounds reasonable; there's also some magic with headings and <dfn> which is why I didn't put the ID on the <dfn>, but we could do that too as an alternative
07:06
<annevk>
MikeSmith: note that master is not broken so it's not breaking the build really
07:06
<MikeSmith>
well true (about not breaking master)
07:07
<MikeSmith>
as far as the alternative, I leave that up to you — but for now, I went ahead and raised https://github.com/whatwg/wattsi/pull/133 for just dropping the error
12:29
<annevk>
domfarolino: hey, are you around by any chance?
12:29
<annevk>
If you're still in Tokyo I guess it's not the best time...
14:33
<MikeSmith>
in Bikeshed, is there a literal `event` definition type?
14:33
<MikeSmith>
I don’t see it listed at https://tabatkins.github.io/bikeshed/#dfn-types
14:33
<Domenic>
I feel like I've seen it used...
14:34
<MikeSmith>
yeah
14:34
<MikeSmith>
ah wait, I guess it’s covered under https://tabatkins.github.io/bikeshed/#dfn-for
14:35
<MikeSmith>
> "attribute", "constructor", "method", "const", "event", "serializer", "stringifier", and "iterator" definitions must define what interface they’re relative to.
14:35
<TabAtkins>
It's still missing from that first list
14:35
<TabAtkins>
But yeah, https://github.com/tabatkins/bikeshed/blob/master/bikeshed/config/dfnTypes.py#L20
14:36
MikeSmith
looks
14:37
<MikeSmith>
aha!
14:37
<MikeSmith>
cool, thanks
14:37
<MikeSmith>
always trust the actual code
14:55
<annevk>
Domenic: do you know who deals with popups at Chrome? With https://github.com/whatwg/html/issues/5872 we have the opportunity to take some more special cases away or at least finally formalize how this legacy system ought to work
14:56
<annevk>
Domenic: arai already wrote a pseudo-spec and I think they'd be willing to turn that into something more concrete
14:57
<Domenic>
I'll ask around...
14:58
<annevk>
Domenic: yeah, I was expecting that it might be tricky, but I'll remain hopeful for now 😊
15:45
<domfarolino>
annevk: hey, am around now
15:45
<domfarolino>
My location is complicated lol. Consider me in pacific time
15:47
jgraham
concludes domfarolino has a very precisely known velocity
16:41
<annevk>
domfarolino: was wondering about default referrer policy and wpt as we have some patches for that; and also the state of it as the discussion feels unresolved
16:42
<domfarolino>
annevk: I still need to review the PR. I think everyone's aligned on changing the default policy. Going further than that is I think a point of further discussion?
16:42
<domfarolino>
annevk: oh are there WPTs up to change the default?
20:17
<annevk>
domfarolino: well the discussion is in the PR; and yeah, we have some patches in Bugzilla for wpt
20:19
<domfarolino>
I think a lot of the discussion was also in the PrivacyCG thread.
20:19
<domfarolino>
annevk: Cool, if you need me to review any WPTs let me know
21:59
<hober>
Seeing this in Bikeshed: The following refs show up multiple times in their spec, in a way that Bikeshed can't distinguish between. Either create a manual link, or ask the spec maintainer to add disambiguating attributes (usually a for='' attribute to all of them)
21:59
<hober>
spec:html; type:dfn; for:/; text:browsing context
22:00
<hober>
the two definitions are apparently https://html.spec.whatwg.org/multipage/browsers.html#browsing-context and https://html.spec.whatwg.org/multipage/browsing-the-web.html#navigation-params-browsing-context
22:08
<hober>
I imagine that latter one should have a for=''