01:22
<nox>
heycam: Great, thanks.
01:40
<TabAtkins>
Domenic: I accidentally fixed the "using [[DOM]] puts [WHATWG-DOM] in the biblio index" problem, while fixing another bug.
01:41
<TabAtkins>
(Whatever form you use in an explicit biblio ref is now used automatically by anything else that grabs the same biblio ref, whether through an aliased term or what-have-you.)
06:42
<annevk>
Is GitHub super slow today?
08:38
<annevk>
terinjokes: is https://github.com/whatwg/fetch/issues/259 something that's worth standardizing language for in the Console Standard?
09:08
<Ms2ger>
zcorpan, can I bug you about https://github.com/w3c/csswg-test/pull/1066 ?
09:42
<zcorpan>
Ms2ger: looking
09:42
<zcorpan>
what does this construct do again? https://github.com/w3c/csswg-test/pull/1066/files#diff-c8670d7412c9acf48f69da4f9aa995c0R39
09:43
<Ms2ger>
I suspect it returns matrix after calling f() on it, but do complain about it :)
10:15
<zcorpan>
ok reviewed
10:21
<Ms2ger>
Thanks
12:50
<annevk>
zcorpan: philipj: https://www.w3.org/Bugs/Public/show_bug.cgi?id=29018
13:56
<annevk>
Woohoo, less than 200 open legacy bugs
14:13
<philipj>
annevk: uh, yes, not sure what to do about that one, would need to redo testing to refresh my memory
14:36
<zcorpan>
gsnedders: jgraham: some of my <menuitem> tests fail in html5lib even though the Expected: and Received: are the same. Is that due to #errors being different, or something else?
14:38
<Ms2ger>
I think we might check the number of errors still?
14:52
<zcorpan>
https://github.com/html5lib/html5lib-tests/pull/72
14:53
<zcorpan>
happy eastern everyone! see y'all on wednesday
14:54
<wanderview>
annevk: if you get a sec, can you confirm that step 8 here should not copy the destination or type from the original request? https://fetch.spec.whatwg.org/#dom-request
14:56
<wanderview>
annevk: or maybe it should copy them... the fetch() method invokes the constructor immediately... is it intentional to lose those values if a service worker does fetch(evt.request)?
14:58
<wanderview>
I'll write an issue... you can close if I'm just confused
15:01
<wanderview>
https://github.com/whatwg/fetch/issues/262
15:10
<annevk>
wanderview: that was intentional
15:10
<annevk>
wanderview: since destination affects the CSP policy
15:11
<wanderview>
annevk: right, but doesn't that break CSP for passthrough service worker requests?
15:11
<annevk>
wanderview: I can add a note about the things we don't copy intentionally I guess
15:11
<annevk>
wanderview: I don't think so
15:12
<annevk>
wanderview: the document that initiates the fetch will get the CSP policy applied it asked for, and then in the worker you'll get the CSP policy corresponding to destination ""
15:12
<wanderview>
annevk: so this assumes the worker has a different CSP policy?
15:13
<annevk>
wanderview: a service worker will always have its own policy afaik
15:13
<annevk>
wanderview: I think all workers do
15:16
<wanderview>
annevk: hmm, ok... not sure why marcos wanted the destination/type to propagate... might need you to comment on the gecko bug at some point
15:20
<annevk>
wanderview: left a comment
15:21
<wanderview>
where?
15:22
<Ms2ger>
Under the mat before your back door
15:22
<annevk>
wanderview: I got a collision with your comment, so I think I'll leave it now since you said effectively the same
15:22
<annevk>
wanderview: I'll add the extra bit I added to GitHub instead
15:22
<wanderview>
thanks
15:34
<Domenic>
annevk: https://github.com/whatwg/html/issues/929 seems bogus to me; can you take a look? The client is the worker, right? Which is a fine place to inherit HTTPS state from.
15:44
<annevk>
Domenic: the problem is that it does not have state yet, I discussed that case earlier with jwatt
15:46
<annevk>
Domenic: gotta go for a bit, one last thing
15:46
<annevk>
Domenic: it'd be ideal for Mozilla if custom elements was all done and sorted within the next couple of weeks
15:47
<Domenic>
annevk: hmm ok, good to know. can do, but why?
15:47
<annevk>
Domenic: if there's anything I can do let me know, I'll let it take precedence over the Shadow DOM work I've been doing
15:47
<Domenic>
annevk: it'll probably just be agressive reviews as I move things to HTML.
15:48
<Domenic>
annevk: next week has some TC39 meetings that might slow things down but rest of this week and the week after next I can make it my top focus
15:49
<annevk>
Domenic: there's a large number of Mozillians in Taipei that'll be doing more Gecko work and in a couple of weeks I believe wchen will be there to help them get started with this
15:49
<annevk>
Domenic: some of them, anyway
15:49
<annevk>
Domenic: so if the spec is mostly done, that's ideal
15:49
<annevk>
Domenic: I'll do another review tomorrow then
15:50
<Domenic>
annevk: biggest missing piece is still the custom element callbacks :-/. I'll hopefully at least get a start on that today/tomorrow
16:37
<annevk>
Domenic: what timezone is TC39 in?
16:37
<Domenic>
annevk: San Francisco
17:39
<tobie>
Is there a best practice for when you should be putting the dfn within WebIDL, compared to out of it?
17:42
<TabAtkins>
If you're specifically defining something, put the dfn near it, so the IDL will link to it. You want anyone visiting the anchor to land on the definition.
17:42
<TabAtkins>
If it's more trivial, feel free to leave it in the IDL.
17:44
<tobie>
TabAtkins: ty
17:45
<tobie>
TabAtkins: asking because now my WebIDL in another spec references the dfn rather than the IDL, which is weird.
17:45
<TabAtkins>
Why weird?
17:47
<tobie>
TabAtkins: well, in interface AmbientLightSensor : Sensor {}; Sensor links to the dfn and not the IDL, which seems weird, no?
17:48
<TabAtkins>
Again, why weird?
17:50
<TabAtkins>
(Also, Sensor isn't in Bikeshed's db - you using local refs? Should I add something to Bikeshed's d?
17:51
<TabAtkins>
*db
17:51
<annevk>
tobie: that would be weird, yes
17:52
<tobie>
TabAtkins: Oh, so that might explain it.
17:52
<tobie>
annevk: OK, I thought so too.
17:52
<annevk>
tobie: class names should reference IDL
17:52
<TabAtkins>
I agree that an interface name isn't something that should be separated from its definition.
17:53
<tobie>
TabAtkins, annevk: sorry this is a combination of me not being super familiar with Bikeshed + me not being super familiar with spec editing. :)
17:53
<TabAtkins>
The most pertinent definition of "what this interface is" is the interface itself. It's just a name given an object shape, unlike methods and attributes.
17:53
<annevk>
tobie: no need to apologize, we're all here to learn
17:54
<TabAtkins>
tobie: You're talking about this dfn, right? https://w3c.github.io/sensors/#sensor
17:55
<tobie>
TabAtkins: yes.
17:56
<tobie>
TabAtkins: it's extra confusing because I have a "sensor concept" which doesn't map the Sensor interface 1:1.
17:56
<TabAtkins>
Alright, so (a) Imma go pop that into Shepherd, and (b) feel free to file a bug asking for them to change that <dfn> to an <a>.
17:56
<TabAtkins>
You can ref me and Anne saying that interface dfns should be in the IDL, as the most useful "definition" of the term. ^_^
17:58
<tobie>
well, I wrote both, so I can go fix it myself. :)
17:58
<TabAtkins>
Ah, kk. ^_^
17:58
<TabAtkins>
In that case, is the desired shortname "sensor" or "generic-sensor"?
17:59
<tobie>
the latter.
18:00
<tobie>
you might want to wait before I pushed changes to fix this, then.
18:00
<tobie>
TabAtkins: ^
18:01
<TabAtkins>
Nah, it'll pick up the differences after you push them, it's fine.
18:01
<TabAtkins>
Specs that aren't in its CI notification structure get reparsed like once a day or something.
18:02
TabAtkins
is still sincerely wondering if he should take this part of Bikeshed on his own, rather than relying on Shepherd.
18:02
<tobie>
TabAtkins: cool, so how does it work, though. Will the IDL in the AmbientLightSensor spec, automatically link to the Generic Sensor one?
18:03
<TabAtkins>
Oh, you'll need to rebuild at some point to make that work.
18:03
<TabAtkins>
But right now you're probably relying on a local <pre class=anchors> block anyway, so wtv.
18:04
<tobie>
Mmm. I'm not sure I understand all of those subtleties.
18:05
<TabAtkins>
Okay. So.
18:05
<TabAtkins>
I just added the Generic Sensor spec to Shepherd a few minutes ago. Shepherd will parse the spec and extract all its definitions, and repeat this about once a day or so iirc in case things change.
18:06
<tobie>
OK, I figured that part. :)
18:06
<TabAtkins>
Your Ambient Light Sensor spec was already linking to Generic Sensor. Since it wasn't in Shepherd yet, it wasn't relying on Bikeshed's built-in autolinking database. It was either doing a manual link (ewwwwww) or was using a <pre class=anchors> block to tell Bikeshed about some additional cross-spec refs it should know about, for the purpose of processing
18:06
<TabAtkins>
this spec.
18:07
<tobie>
OK, now I get it.
18:07
<tobie>
I guess what makes it weird is that I don't know upfront what spec is in shepherd in which isn't.
18:08
<tobie>
s/in/and/
18:08
<tobie>
So sometimes the magic works and sometimes it doesn't.
18:09
<TabAtkins>
Yeah, that's a bit of a problem.
18:09
<tobie>
For example, the permissions-API isn't in Shepherd, right?
18:09
<TabAtkins>
The interface isn't publicly available.
18:09
<TabAtkins>
Yeah, it's not.
18:09
<tobie>
TabAtkins: got it.
18:10
<TabAtkins>
You can always ask Bikeshed if it knows about a particular reference or not. Just run: `bikeshed refs` (-h to see the arguments) and it'll list everything it knows matching that ref request.
18:10
<tobie>
OK, I see.
19:01
<smaug____>
annevk: rootNode can be consider stable, right?
19:01
Domenic
is suddenly tempted to bikeshed the name
19:02
<smaug____>
ha
19:03
<smaug____>
Domenic: don't even think about or I'll start complaining about Custom Element's callback names and functionality ;)
19:03
<Domenic>
smaug____: hahaha you got me
19:07
<smaug____>
s/consider/considered
19:23
<annevk>
smaug____: yes
20:12
<jyasskin>
TabAtkins: Can Shepherd handle ReSpec editors' drafts at all? It'd be nice to get the permissions API in there even if we have to use link-defaults a lot.
20:13
<jyasskin>
TabAtkins: Also, where's Shepherd's list of specs? I'd be happy to send pull requests instead of bugging you for each new spec.
20:18
<TabAtkins>
jyasskin: Only if the ReSpec spec has explicitly added decent metadata that Bikeshed understands. The format is well-documented now, but it's rather annoying to add by hand, so ~0 specs have done so.
20:18
<TabAtkins>
jyasskin: (Or if they happen to be using patterns similar to what Shepherd is using to extract definitions from CSS/SVG/HTML/a few others.)
20:19
<TabAtkins>
(It works better than I'd expect, but not "well".)
20:19
<TabAtkins>
The list of specs is stored in Shepherd's DB, only accessible via the admin interface which is obviously not public. :/
20:19
<TabAtkins>
You can also see it in processed form in Bikeshed's data files, the bikeshed/spec-data/specs.json file.
20:20
<TabAtkins>
But there's no PR bridge to it; I have to add things manually.
20:20
<TabAtkins>
(I'm wondering how to fix it, and will bug plinss about it today.)
20:21
<TabAtkins>
Well actually, that's not true, there's a public API to read the spec data from Shepherd.
20:22
<jyasskin>
Yeah. IIUC, from ReSpec specs we'd get a lot of non-exported 'dfn' references, but they wouldn't be marked as IDL.
20:22
<TabAtkins>
https://api.csswg.org/shepherd/ is the human-readable endpoint that describes the REST api.
20:22
<TabAtkins>
jyasskin: Exactly.
20:23
<TabAtkins>
If people did what they were *supposed* to and actually published the *processed* ReSpec documents, instead of their source files and a JS converter, this would be easy. ReSpec could use the same sorts of shorthands and such that Bikeshed does, and mark things up well.
20:23
<jyasskin>
Yeah, I'm looking more for a place to send PRs.
20:23
<TabAtkins>
But nah, people lazy.
20:24
<jyasskin>
Presumably Shepherd could run the JS for a page and then process the result.
20:26
<TabAtkins>
Theoretically, sure.
20:26
<TabAtkins>
Would require running things in Node and hoping that jsdom is up to the task, tho. Shepherd is written in Python, so it'd need a language bridge, too.
20:27
<TabAtkins>
I'm bugging plinss now to see if we can have a GH repo be the "canonical" storage for the spec list.
21:02
<tobie>
TabAtkins: think you could take a quick look at https://github.com/w3c/ambient-light/pull/7/commits/fd7548a
21:03
<tobie>
TabAtkins: and tell me whether I should be getting auto links to SensorReading SensorOptions and the like? I'm clearly doing something wring and I have no idea what.
21:03
<tobie>
*wrong
21:04
<TabAtkins>
Will check in a few, currently walking to Starbucks to avoid the cleaners
21:04
<tobie>
TabAtkins: ty
21:07
<Domenic>
jsdom is always up to the task! <3
21:07
<Sebmaster>
^
21:14
<tobie>
TabAtkins: src file has now been merged here: https://github.com/w3c/ambient-light/blob/gh-pages/index.bs
21:14
<tobie>
TabAtkins: note the index.html is the old Respec doc, so don't be surprised if the two are disconnected for now.
21:16
<wanderview>
jsbell: can you remind me where in the chromium tree your wpt service worker tests are?
21:17
<wanderview>
I can't seem to find the right path and the source viewer doesn't seem to have a search box any more
21:17
<jsbell>
wanderview: third_party/WebKit/LayoutTests/http/tests/serviceworkers and cachestorage
21:17
<wanderview>
thanks!
21:17
<jsbell>
wanderview: https://cs.chromium.org (cs=codesearch)
21:18
<wanderview>
jsbell: ah, that helps... I followed the "browse source" link by accident
21:18
<wanderview>
instead of the search source
21:18
<jsbell>
np!
21:19
<wanderview>
jsbell: btw, I am close to upstreaming our tests... are you ok if I just do that and then we can hash out any differences in PRs?
21:19
<wanderview>
I've probably asked you this before... sorry for the repeat
21:19
<jsbell>
wanderview: yep, sounds good
21:19
<wanderview>
thanks again!
21:21
<TabAtkins>
tobie: How do you want comments, now that it's been merged?
21:21
<tobie>
TabAtkins: can't you still comment on the PR?
21:21
<TabAtkins>
I can, yeah. Just asking you whether that was good. Sounds like it is. ^_^
21:22
<tobie>
yeah, whatever's more convenient for you.
21:23
<tobie>
here's fine too, depending on what it is you need to say
21:27
<tobie>
oh!! dfn
21:27
<tobie>
TabAtkins: dfns aren't exported by default
21:27
<tobie>
Now I get it.
21:28
<TabAtkins>
dfn-type dfns.
21:28
<TabAtkins>
(Which <dfn>s are if you dont' specify a different type.)
21:28
<TabAtkins>
When fully expanded, <dfn data-dfn-type=dfn>. ^_^
21:29
<tobie>
I--
21:30
<tobie>
Maybe I should re-read the doc another 3 times.
21:30
<TabAtkins>
Yeah, I know. Badly named. Can't fix it now.
21:30
<TabAtkins>
Really annoying to talk about and I hate it.
21:31
<TabAtkins>
Something like <dfn interface> (or when expanded, <dfn data-dfn-type=interface>) *are* exported by default. It's just the "dfn"-type ones that are treated different for exporting.
21:31
<tobie>
BTW, I wasn't really asking for a review of this src file, more of an explanation as to why there wasn't auto-links to the stuff defined in the generic sensor spec.
21:32
<tobie>
And it seems that's because I've using DFN's for interfaces, and those aren't exported by default
21:32
<TabAtkins>
They're the wrong type, and *also* as a result of that they aren't exported, yeah.
21:32
<TabAtkins>
The fact that they're not exported isn't why your autolinks are failing, tho, it's that they're not the right type.
21:33
<tobie>
https://github.com/w3c/sensors/blob/gh-pages/index.bs#L648
21:33
<TabAtkins>
But it looks ok in the source - Sensor is defined with an "interface" type, as it should be.
21:34
<TabAtkins>
Are you generating your spec with a local copy of Bikeshed, or the remote version?
21:34
<tobie>
local
21:35
<TabAtkins>
Have you run `bikeshed update` some time after I put Sensors in the db?
21:35
<tobie>
mmm
21:35
<tobie>
yeah-ish
21:36
<TabAtkins>
May have been too early. I just did it, and SensorReading is definitely in the db now. Update again and it should get picked up.
21:36
<tobie>
Amazing if it is.
21:36
<TabAtkins>
Why amazing?
21:37
<tobie>
Because it's 10:30pm and I'm tired. :)
21:37
<TabAtkins>
Oh, amazing as in "I'll be happy if so", not "that would be surprising"
21:38
<tobie>
yeah--sorry.
21:38
<tobie>
It works.
21:38
<tobie>
Also it's amazing because it's actually pretty cool tech.
21:38
<tobie>
:)
21:39
<TabAtkins>
Cross-spec autolinking is literally the reason I wrote Bikeshed. It is pretty cool. ^_^
21:39
<tobie>
indeed.
21:40
<tobie>
OK, so I'll need to export dfns I want to reference in the AmbientLight spec
21:40
<tobie>
And should leave the dfn with the WebIDL for the interfaces
21:40
<TabAtkins>
Yeah. If you don't you can still link them, but you need to declare that you really do mean to link that spec, usually through a link-defaults block.
21:40
<tobie>
And then it'll be all good.
21:41
<TabAtkins>
Which is annoying if the term is *meant* to be generally useful for other specs.
21:41
<tobie>
OK. Got it.
21:41
<tobie>
Thanks so much for your help.
21:42
<TabAtkins>
np
21:43
<TabAtkins>
Glad to help the ecosystem grow.
22:24
<Krinkle>
Woo. https://w3c.github.io/webstorage/#dom-localstorage points to https://html.spec.whatwg.org/multipage/webstorage.html
22:34
<tobie>
TabAtkins: curious as to why https://w3c.github.io/sensors/#high-vs-low-level [GEOLOCATION-API] ends up as [geolocation-API] in the ref section.
22:34
<tobie>
TabAtkins: https://w3c.github.io/sensors/#biblio-geolocation-api
22:36
<tobie>
TabAtkins: sure, the shortname is mixed-cased, but my local ref isn't.
23:29
<TabAtkins>
tobie: Because the ref index uses the casing from SpecRef if it's mixed-case, in case that's important.
23:29
<TabAtkins>
(It uppercases things that are all-lowercase.)
23:30
<tobie>
TabAtkins: oh, right, because you're not using the specref service
23:30
<TabAtkins>
Well, I'm calling SpecRef for its data.
23:30
<tobie>
TabAtkins: the service is case-insensitive and generates an alias on the fly
23:31
<TabAtkins>
duuuuude, store the casing you want *in the db*.
23:32
<tobie>
TabAtkins: well... depending on the use case, having the actual shortname casing is important
23:32
<TabAtkins>
Then don't complain when I preserve it. ^_^
23:32
<TabAtkins>
(Bikeshed is also case-insensitive for *looking up* refs. It just preserves the casing for display if it surmises that it's important, such as when the casing is mixed.)
23:33
<tobie>
Well, sticking references in a spec isn't one of those use cases. :P
23:33
<TabAtkins>
Mang I can't tell that.
23:35
<tobie>
It's not that important tbh, but I do think the casing displayed in the ref section should match that used in the body of the spec
23:35
<tobie>
i.e. those references are internal
23:36
<TabAtkins>
I can probably do that, given the way I'm handling "preferred" names now.