00:17
<zewt>
are there any html parsers for python that are actually reliable? been using html5lib but apparently it throws exceptions on some escapes, and the bug has been open for over three years, which tells me i should be looking for a new library...
00:45
<MikeSmith>
zewt: you have a link to that open html5lib issue?
00:46
<MikeSmith>
anyway, why not just use the default lxml html parser?
00:47
<MikeSmith>
it seems to work pretty well, even though it’s not completely conformant to the HTML parsing algorithm (and doesn’t claim to be)
00:47
<MikeSmith>
WebIDL question: does required DOMString foo mean an instance of foo can be the empty string?
00:50
<gsnedders>
zewt: the bug with the lxml treebuilder and non-XML-respresentable trees? I wish it were easier to fix. :(
00:51
<gsnedders>
(the narrow/wide split of Python 2 makes it absolutely horrible to fix…)
00:51
<gsnedders>
(I guess we could just do some really ugly hack for element names on narrow builds, but that doesn't entirely fix the issue)
00:55
<MikeSmith>
is there actually some way in WebIDL to explicitely specify that something must be non-empty?
00:56
<zewt>
https://github.com/html5lib/html5lib-python/issues/96
00:57
<gsnedders>
zewt: yeah, that's the one
01:00
<jyasskin>
MikeSmith: "required" just means it must be present (in JS, "not undefined"). "" is a valid DOMString, so it's allowed in required dictionary elements.
01:00
<MikeSmith>
jyasskin: OK that’s what I thought, thanks
01:01
<jyasskin>
MikeSmith: WebIDL doesn't have a way to express constraints on the valid values in a field. In Bluetooth, I did it with https://webbluetoothcg.github.io/web-bluetooth/#uuids, but I also said what exception to throw in prose everywhere that uses a UUID.
01:01
<jyasskin>
The best fields are "required Type?": you can have null, but undefined is right out! ;-)
01:02
<MikeSmith>
oh
01:02
<MikeSmith>
then I guess that is what we may need to do more of in the Payment Request API spec where needed
01:03
<jyasskin>
But I picked SyntaxError in WebBT, which I think is wrong, so don't copy that part. ;)
01:03
<MikeSmith>
I am looking at https://bugs.chromium.org/p/chromium/issues/detail?id=636723 where Rouslan has “Accept any non-empty string” as the isssue title, but the Payment Request spec says nothing explicitly about it needing to be non-empty
01:04
<MikeSmith>
https://w3c.github.io/browser-payment-api/#paymentcurrencyamount
01:06
<jyasskin>
MikeSmith: I also note "MUST"s being applied to dictionary fields that are passed from user code to interface operations. You can't "MUST" a website; you have to instead say what happens when they do something you disapprove of.
01:07
<jyasskin>
It's possible the spec already does that; the MUSTs just triggered my reflexes.
01:13
<MikeSmith>
jyasskin: yeah me too. Those MUSTs should not be there, and I am actually right now at this moment writing a PR to change them to just (non-normative) “The following fields are required”
01:13
<jyasskin>
Woot
01:14
<MikeSmith>
but there are still some other places in the spec where it does not actually normatively specify what a UA should do with “invalid” instances of something
01:14
<MikeSmith>
see https://github.com/w3c/browser-payment-api/issues/235
01:15
<MikeSmith>
https://w3c.github.io/browser-payment-api/#dfn-value
01:15
<jyasskin>
Indeed. My apologies for assuming you were responsible for the current spec text. :)
01:15
<MikeSmith>
heh
01:16
<MikeSmith>
you can blame Zach Koch :)
01:16
<MikeSmith>
anyway it is a very good spec overall
01:16
<MikeSmith>
I just ran into these misses when reviewing test cases
01:42
<zewt>
don't care about how control characters are parsed (it's junk anyway), just want it to not suddenly throw errors while parsing
02:26
<astearns>
there's a few "authors must" in the CSS specs. I suppose they should change to "authors ought to"
02:26
<zewt>
"wouldn't it be nice if authors would"
02:27
<astearns>
"I'd like imagine that authors would"
02:28
<variable>
astearns: "if people read the spec and cared about what it said, they would"
02:28
<zewt>
why don't we just stop fantasizing and go with "authors don't"
02:35
<MikeSmith>
astearns: there are some “authors must” and even “documents must” requirements in the HTML, but those are not ambiguous because the HTML spec defines an explicit “conforming documents” conformance class
02:35
<MikeSmith>
https://html.spec.whatwg.org/multipage/infrastructure.html#conforming-documents
02:36
<MikeSmith>
though we still have cases of implementors reading the spec and thinking those are meant to somehow be requirements for UAs
02:37
<MikeSmith>
I think there is a good argument to be made for not trying to specify both kinds of requirements in the same spec
02:38
<MikeSmith>
in other words, not trying to have a spec that attempts to be both for UA implementors and authors
02:38
<MikeSmith>
because in the end it just ends up being confusing for both implementors and authors
02:39
<MikeSmith>
and really in the case of the HTML spec, it’s a spec friendly for implementors in its style and content and not really friendly to authors
02:40
<MikeSmith>
I think most authors have no business trying to read the HTML spec
02:40
<MikeSmith>
or maybe we have no business continuing to imply that authors should
02:42
<MikeSmith>
I think there is a good argument to be made for a normative spec for document-conformance requirements that is written from the ground up with authors in mind
02:42
<MikeSmith>
a separate spec
02:42
<MikeSmith>
rather than what we have for authors now, which is stuff basically bolted onto the implementor spec
02:43
<variable>
MikeSmith: for the most part authors want tutorial like information, not spec like information, AFAICT
02:43
<MikeSmith>
and that bolted-on stuff is distracting and sometimes confusing to authors
02:44
<MikeSmith>
variable: right but we still need to have the normative requirements for authors defined somewhere
02:45
<caitp>
what if spec text could be both normative and accurate, and also accessible
02:45
<MikeSmith>
I really prefer the way that annevk writes specs, which are spare and clearly written for implementors only, in a way that they can implement from without distraction
02:45
<MikeSmith>
caitp: would that it could
02:45
<MikeSmith>
I think it cannot
02:46
<MikeSmith>
that is why I like annevk’s approach more
02:46
<howdoi>
would you have a div in a section or a section in a div or just use sections?
02:46
<variable>
howdoi: depends on what semantics you want
02:47
<howdoi>
From the spec, Note: The section element is not a generic container element. When an element is needed for styling purposes or as a convenience for scripting, authors are encouraged to use the div element instead. A general rule is that the section element is appropriate only if the element's contents would be listed explicitly in the document's outline.
02:47
<MikeSmith>
to me annevk’s style does not attempt to be particularly accessible to non-implementors, and that is a feature
02:47
<variable>
howdoi: you should use <section> if you care about it from an outlining PoV
02:47
<variable>
tbh outlining is not commonly used
02:47
<MikeSmith>
others who understand the specs can write guides at MDN and places that accessible to authors, and optimized for authors
02:47
<variable>
MikeSmith: I think that is a reasonable approach
02:48
<howdoi>
variable: outlining is a pretty confusing term.
02:48
<MikeSmith>
howdoi: personally I think I would just not use section at all
02:48
<variable>
howdoi: the short version is: if you think it should live in a "table of contents"
02:48
<variable>
but MikeSmith's suggestion is reasonable
02:48
<MikeSmith>
at this point IMHO there are no real benefits in practice to using section, nor article
02:49
<variable>
since outlining is basically dead :(
02:49
<MikeSmith>
yeah
02:49
<howdoi>
just the readability?
02:50
<howdoi>
The section element represents a generic section of a document or application. A section, in this context, is a thematic grouping of content, typically with a heading. 🤔 😁
02:50
<MikeSmith>
I implemented the outline algorithm in the HTML checker (validator) but I personally think authors are better of ignoring the outline algorithm and just using h1-h6 without section or even without div unless the div is needed for some other reason
02:51
<variable>
howdoi: this is the conversation that MikeSmith and I are basically having: the spec isn't great for people like you, authors
02:51
<howdoi>
variable: collusion of thoughts, it's a small world!
02:51
<variable>
MikeSmith: I wish that assistive devices were capable of using the outlining stuff, and that authors (or more accurately, framework authors) would care about this stuff
02:52
<howdoi>
we must start a session on how to read the spec
02:52
<variable>
but if the semantic web has taught us anything, its that people will not manually do stuff, ever
02:52
<howdoi>
later, how to submit a spec and convince the board :D
02:52
<howdoi>
strongly agree there.
02:53
howdoi
basically no one give a $hit about semantics, until it's affecting their page rank or something similar :(
02:53
<MikeSmith>
variable: yeah for AT software to use the outline stuff, browsers would first need to implement the requirements in the spec about actually exposing the outline in the right way to a11y APIs. Which browsers don’t, so there’s not much that AT software can do in the mean time
02:55
<MikeSmith>
but that said the JAWS vendor at least never implements anything unless they absolutely have to
02:55
<MikeSmith>
case in point is language detection
02:55
<MikeSmith>
language detection is something cheap and easy to implement
02:56
<howdoi>
maybe off-topic: If you were to do a simple PWA introduction to a crowd, say a workshop, what app would you do?
02:56
<MikeSmith>
and it would be of real benefit for JAWS users to have built-in language-detection support
02:56
<MikeSmith>
but as far as I know, JAWS has no such support
02:57
<MikeSmith>
howdoi: Ilya Grigorik’s O’Reilly book
02:57
<howdoi>
MikeSmith: https://hpbn.co/?
02:58
<MikeSmith>
yeah
02:58
<MikeSmith>
works offline
02:58
<howdoi>
that's already a PWA
02:58
<MikeSmith>
automatically
02:58
<howdoi>
you mean to say, build something like that?
02:58
<MikeSmith>
oh I guess I misunderstood your quesiton
02:59
<howdoi>
Say, you were asked to do a workshop on PWA, what would you build there?
02:59
<MikeSmith>
a book I guess
02:59
<howdoi>
please don't say a weather app :D
02:59
<MikeSmith>
it is the simplest case
02:59
<MikeSmith>
book with images
02:59
<howdoi>
okies, I was planning for flickr images
02:59
<howdoi>
or maybe XKCD
02:59
<howdoi>
and some videos
02:59
<MikeSmith>
yeah
02:59
<howdoi>
not sure if I need to cover push notifications and background sync.
03:00
<MikeSmith>
yeah that would make it a lot more complicated
03:00
<variable>
MikeSmith: reminds of the html fixer
03:00
<howdoi>
okies, will ditch them then, thanks for the feedback :)
03:00
<variable>
it takes arbitrary input and produces valid HTML as output
03:00
<variable>
"rm"
03:01
<MikeSmith>
variable: not http://www.htmlfixer.com/ I guess
03:02
<variable>
MikeSmith: no, its a unix tool
03:02
<variable>
"rm"
03:03
<MikeSmith>
ah I know rm of course but I guess I’m not getting the joke
03:03
<variable>
MikeSmith: the joke is that rm will correctly produce an empty document
03:03
<variable>
which is valid HTML
03:03
<MikeSmith>
ah OK
03:04
<MikeSmith>
seems like that would be more like touch(1)
03:04
<caitp>
usually that would result in a 404, which in your typical user agent would produce results inconsistent with the "document"
05:21
<rniwa>
annevk: hello
05:41
<annevk>
rniwa: heya, how are things?
05:41
<rniwa>
annevk: coming back to custom elements
05:41
<rniwa>
annevk: I have one question for you
05:41
<annevk>
rniwa: hope you left the office
05:41
<rniwa>
annevk: the namespace is supposed to be null
05:41
<rniwa>
annevk: by default, right?
05:41
<rniwa>
annevk: not empty string
05:41
<rniwa>
annevk: this is kind of relevant in terms of attributeChangedCallback
05:41
<annevk>
rniwa: for attributes?
05:42
<rniwa>
annevk: because it adds the namespace URI as the last argument
05:42
<rniwa>
annevk: yeah
05:43
<annevk>
rniwa: hmm I guess, mostly a legacy accident but all other APIs behave that way, right?
05:43
<rniwa>
annevk: https://gist.github.com/rniwa/a2b010863e8e6f111d540f29c7d45086
05:43
<rniwa>
annevk: because attributeChangedCallback puts the namespace URI into an argument
05:43
<rniwa>
annevk: it makes this somewhat internal implementation detail of the spec exposed
05:43
<rniwa>
annevk: all existing APIs such as getAttribute has "?" qualifier
05:44
<rniwa>
annevk: so that makes the null-ness of namespace URI kind of irrelevant but this API would expose it.
05:44
<annevk>
rniwa: it should match mutation observers, can look up in 20 what that is
05:46
<rniwa>
annevk: that's a good point
05:47
<rniwa>
annevk: we DO use null in that case
05:53
<annevk>
rniwa: so yeah, https://dom.spec.whatwg.org/#concept-element-attributes-change does exactly the same for mutation observers and custom elements
05:53
<rniwa>
annevk: yeah
05:53
<rniwa>
annevk: cool
05:53
<rniwa>
annevk: it looks like this is just a bug in Chrome's implementation
05:53
<annevk>
rniwa: and https://dom.spec.whatwg.org/#dom-attr-namespaceuri returns that field literally too so I think we're good
05:53
<rniwa>
annevk: ah, okay
05:53
<annevk>
Oh Chrome does empty string? Geez...
05:53
<rniwa>
annevk: yeah
05:54
<rniwa>
annevk: well, null / empty string different is somewhat subtle due to historical nullability/undefined implicit coercion :(
05:54
<rniwa>
annevk: cdumez has been doing an amazing job updating our impl
05:54
<annevk>
Yeah, this should never have allowed null
05:54
<rniwa>
annevk: but there is a lot of crazy edge cases in WebIDL around that area
05:55
<annevk>
cdumez is great
05:57
<rniwa>
yeah
06:03
<rniwa>
sigh... our impl. of sequence conversion is completely busted :(
06:04
<annevk>
the finer details of sequences are hard due to all the subtle hooks in ECMAScript
06:05
<annevk>
bz explained it to me once and I promptly forgot 😟
06:06
<rniwa>
annevk: it's not as bad as all the things Proxy can do
06:07
<rniwa>
annevk: e.g. you can trap `Get(constructor, "prototype")` with Proxy
06:07
<annevk>
I'm still a bit mystified as to why we have those
06:07
<rniwa>
annevk: and create an infinitely recursive prototype chain :(
06:07
<rniwa>
annevk: the part that sucks most about Proxy is that you can't do it per property
06:07
<rniwa>
annevk: so all bets are off for optimizations :(
06:08
<annevk>
I really hope the only justification wasn't just to explain the hooks a browser is using to implement some HTML features
06:09
<rniwa>
annevk: which is still strictly less worse than https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Symbol/isConcatSpreadable
06:10
<rniwa>
one of my colleagues spent one entire month optimizing this feature to not regress the perf in regular case :(
06:10
<rniwa>
I literally can't think of a situation in which one would want to use this feature instead of simply calling push...
06:17
<annevk>
That TC39 doesn't let itself influence by performance concerns is rather frustrating
06:17
<annevk>
But one thing Apple could help with is sending people to those meetings (I know they can be dull) and object to these kind of things
06:18
<rniwa>
annevk: oh i think we've been sending people to every TC39 meeting these days
06:18
<annevk>
Implementers seem somewhat underrepresented at times
06:18
<annevk>
Ah good
06:18
<annevk>
And yeah, this happend a while ago, as did freeze() et al
06:21
<rniwa>
annevk: yeah
06:21
<rniwa>
annevk: ES6 is quite a massive latency regression for us so we had to spend a lot of time optimizing various parts of the engine
06:42
<hsivonen>
I need a demo string that encodes without unmappables into a legacy multibyte encoding. Does anyone happen to know what's used in programming examples in Chinese, Japanese or Korean where "Hello, World!" is used in English?
06:46
<hsivonen>
Wikipedia suggests ハロー・ワールド in Japanese. Unclear if idiomatic.
06:46
<hsivonen>
I suppose that's good enough
06:49
<hsivonen>
oh. it looks like that's a phonetic rendering of English
07:57
<zewt>
hsivonen: phonetic renderings of English seems like exactly what the japanese would use
08:38
<hsivonen>
zewt: ok
08:50
<MikeSmith>
hsivonen: yeah for Japanese what zewt said I think though all examples in https://ja.wikipedia.org/wiki/Hello_worldプログラムの一覧 just use literal latin “Hello world”
08:51
<MikeSmith>
hmm except for https://ja.wikipedia.org/wiki/Hello_worldプログラムの一覧#KQ
08:52
<MikeSmith>
I wonder what KQ is
08:52
<MikeSmith>
oh man
08:52
<MikeSmith>
https://github.com/fumieval/LazyKQ
08:52
<MikeSmith>
that is pretty cool
08:56
<MikeSmith>
man the obnoxious of that is brilliant
08:57
<MikeSmith>
I will never again say that half-width katakana is useless
08:59
<MikeSmith>
botie, inform rniwa hey somebody finally found an appropriate use for half-width katakana https://github.com/fumieval/LazyKQ
08:59
<botie>
will do
09:19
<zewt>
as far as i can tell the main use of halfwidth katakana is to make my IME have an extra state to toggle through
09:34
<MikeSmith>
heh
09:34
<MikeSmith>
yup
09:34
<foolip>
annevk: how do I "sync html" in whatwg/xref?
09:34
<foolip>
the html.py script takes a while and does nothing
09:35
<annevk>
foolip: you need to manually add entries to html.json
09:35
<foolip>
how many specs are still using this? it's not great :)
09:35
<annevk>
foolip: it's sad
09:35
<annevk>
foolip: only anolis, they need to become bikeshed at some point
09:35
<foolip>
I assume you'd be happy to merge a bikeshed PR for fullscreen?
09:35
<annevk>
foolip: I'm mostly putting it off since it's a lot of work and the regressions from last couple of times haven't been great
09:36
<annevk>
foolip: yeah
09:36
<MikeSmith>
zewt: for the iPhone Apple dropped support for halfwidth katakana input in the IME a few years ago but then subsequently had to add it back
09:36
<foolip>
which others are still left on anolis?
09:36
<annevk>
foolip: Fetch, XHR I guess
09:36
<foolip>
mkay
09:36
<foolip>
oh well, not today
09:36
<foolip>
I can just push directly to xref I guess?
09:37
<annevk>
foolip: yeah
09:38
<foolip>
annevk: enjoy https://github.com/w3c/permissions/issues/121
09:38
<foolip>
bah, not that one
09:38
<foolip>
https://github.com/whatwg/fullscreen/pull/46
09:39
<annevk>
ta
09:39
<foolip>
I couldn't figure out the line wrapping rules so I went with inconsistent :)
09:42
<annevk>
Oh, I should have looked at that more carefully, oh well
09:42
<annevk>
We can fix that when we bikeshed it
09:42
<annevk>
Or rather, after
09:43
<annevk>
Best not to change line endings as part of the conversion as it'll make review too hard
09:47
<foolip>
I use git diff --word-diff -w a lot when reviewing spec changes because never changing line wrapping is hard
09:58
<MikeSmith>
oh wow
09:58
<MikeSmith>
I never heard of --word-diff
09:58
<MikeSmith>
this is great
10:04
<annevk>
https://github.com/w3c/webpayments-method-identifiers/issues/9 is all kinds of horrible
10:04
<annevk>
Payments WG is basically in back in the dark ages with nothing learned about anything
10:07
<annevk>
"I'll just fix @annevk"
10:18
<wilsonpage>
Domenic ping
10:37
<smaug____>
interesting. Chrome uses HTMLDocument for svg images
10:38
<smaug____>
but if one uses document.write for example, the document goes into some odd state without element
10:38
<smaug____>
s
10:49
<MikeSmith>
annevk: sorry to have roped you into that payments URL nonsense
10:50
<MikeSmith>
anyway I think the actual browser Payment Request API spec is a quite good spec
10:51
<MikeSmith>
it is the only thing that’s actually implemented at this point so the rest of that stuff is not really affecting anything anyway
11:55
<mathiasbynens>
annevk: can you make javascript.spec.whatwg.org 301 to https://github.com/tc39/ecma262/labels/web%20reality? I can set up a <meta>/JS-based redirect for now, but hey…
12:20
<annevk>
mathiasbynens: done
12:35
<mathiasbynens>
annevk: thanks! and woohoo
12:40
<MikeSmith>
http://stackoverflow.com/questions/38917631/why-is-it-impossible-to-gather-parent-window-location-from-iframe-when-different
12:44
<MikeSmith>
if the WebIDL for some field in a dictionary defines it as DOMString and my script sets it to null or undefined, then per WebIDL, the UA should throw?
12:45
<MikeSmith>
lacking the [TreatNullAsEmpty] annotation (or whatever teh exact name is)
12:48
<annevk>
youennf made really great Fetch tests
12:48
<annevk>
MikeSmith: required DOMString or just DOMString?
12:49
<annevk>
MikeSmith: either way, null should turn into "null" I think
12:49
<MikeSmith>
annevk: required DOMString
12:49
<annevk>
MikeSmith: then undefined should throw
12:49
<MikeSmith>
annevk: OK but then null should not?
12:49
<annevk>
MikeSmith: null should stringify
12:49
<MikeSmith>
or you mean it turns into the string "null"?
12:49
<MikeSmith>
hai
12:49
<annevk>
hai
12:50
<MikeSmith>
thanks
12:50
<annevk>
We use hai here at home all the time, such a great word
12:50
<MikeSmith>
haha
12:50
<MikeSmith>
yeah it’s pretty useful
12:50
<zcorpan>
mathiasbynens: \o/
13:01
<Domenic>
mathiasbynens: you should write a blog post for blog.whatwg.org about sunsetting the JavaScript Standard! It's a fun success story.
13:02
<mathiasbynens>
Domenic: sounds fun, I’ll write something up next week!
13:06
<bradleymeck>
uggg please no more drama
13:08
<annevk>
?
13:10
<bradleymeck>
annevk: was referring to the idea of making a blog post about bugs w/ JS spec
13:10
<Domenic>
???
13:10
<Domenic>
who suggested that?
13:11
<bradleymeck>
"mathiasbynens: you should write a blog post for blog.whatwg.org about sunsetting the JavaScript Standard! It's a fun success story." must have misread?
13:11
<annevk>
Yeah, I think you did
13:11
<bradleymeck>
then I am confused, w/e
13:13
<bradleymeck>
don't see how it could be read diff w/ scrollback though
13:15
<Domenic>
We're talking about https://github.com/whatwg/javascript/issues/29. You appear to be talking about a completely separate topic, bugs in ECMA262?
13:16
<bradleymeck>
yea, diff standard
13:17
<bradleymeck>
thx for the link, was very confused
13:52
<zcorpan>
dang I was for a moment looking forward to some JS drama :-)
14:29
<wanderview>
JakeA: annevk: have we considered exposing self.clients Clients interface on window or other workers? except for claim() it seems like it would work just as well from those globals
14:30
<JakeA>
"controlled" doesn't make as much sense from a window, but yeah, I think it'd be fine to have in other contexts
14:31
<wanderview>
JakeA: you mean for ClientQueryOptions?
14:31
<JakeA>
wanderview: yeah
14:32
<wanderview>
JakeA: it would be kind of nice if we could get people using Clients.openWindow() instead of window.open()
14:33
<JakeA>
wanderview: although… if you postmessage a client, would it still land on navigator.serviceWorker?
14:33
<wanderview>
JakeA: oh... I forgot about that part... thats kind of icky
14:37
<wanderview>
JakeA: it seems if we did this we would need to move .claim() into a specialization for service worker globals anyway... we could specialize postMessage() the same way
14:37
<JakeA>
Agreed
14:37
<wanderview>
I guess I could write an issue
14:38
<wanderview>
work, work
14:38
<JakeA>
There isn't a way to postmessage to SharedWorkers though
14:38
<JakeA>
other than navigator.serviceWorker
14:38
<JakeA>
self.onmessage doesn't exist
14:38
<wanderview>
JakeA: self.clients.onmessage?
14:39
<wanderview>
I mean.. we could hang it off of there for Client API postMessaging...
14:39
<JakeA>
wanderview: In retrospect, that's a better place for it than navigator.serviceWorker
14:39
<wanderview>
but sometimes it feels like we try to see how many messaging paths we can create
14:40
<wanderview>
JakeA: which reminds me... we talked about BroadcastChannel in service worker... would those messages get our new fangled ExtendableMessageEvent with waitUntil()?
14:41
<JakeA>
wanderview: I don't think so, since you'd need to be in another event to create the channel
14:41
<wanderview>
JakeA: I meant for receiving BroadcastChannel messages from a window, etc
14:42
<wanderview>
but maybe thats what you meant
14:42
<wanderview>
I see
14:42
<JakeA>
wanderview: are you suggesting a new top-level thing like self.onbroadcastchannelmessage?
14:42
<wanderview>
JakeA: no... I was just wondering if BroadcastChannel is easily usable in SW... or if it has problems due to waitUNtil() being missing... but you correctly point out you already need to be in a place with a waitUntil
14:43
<wanderview>
JakeA: I wonder how long until people start putting `self.waitUntil = evt.waitUntil` at the top of all their event handlers
14:43
<wanderview>
I guess we need async waitUntil for that to be more useful
14:44
wanderview
may not be thinking right today...
14:44
<JakeA>
It'd still fall foul of overall limits right?
14:44
<wanderview>
JakeA: you mean the timeout with a waitUntil() holding things alive too long?
14:45
<JakeA>
yeah
14:45
<wanderview>
JakeA: it should I think
14:45
<wanderview>
JakeA: although if you keep getting functional events then I think the timer woudl be reset, of course
14:53
<wanderview>
JakeA: https://github.com/slightlyoff/ServiceWorker/issues/955
14:54
<JakeA>
Cheers!
15:10
<Domenic>
wilsonpage: pong
15:12
<wilsonpage>
Domenic got a tricky situation with my fastdom library regarding multiple versions running on the same page
15:12
<wilsonpage>
annevk pointed me your way :)
15:12
<wilsonpage>
https://github.com/wilsonpage/fastdom/issues/90
15:12
<wilsonpage>
Domenic I imagine the problem has been solved someplace beforebefore
15:13
<Domenic>
Hmm not sure exactly what makes me qualified to help here, but I guess I can weigh in on option A/B/C...
15:15
<wilsonpage>
Domenic that would be a mega help :)
15:16
<Domenic>
Goodness this is a tough one :(
15:16
<Domenic>
Most libraries don't have this kind of coordination problem
15:17
<wilsonpage>
Domenic yeah most the options I came up with are pretty lame
15:17
<wilsonpage>
But I'm all out
15:40
<Domenic>
Today I learned about <meta http-equiv="default-style">. Whyyyy
15:56
<wilsonpage>
Domenic great response, thanks mate!
17:44
<TabAtkins>
astearns: No, "authors must" is intentional and should be kept. It's just a different conformance class that we can't actually test. ^_^ It still, of course, has zero influence on what the implementors have to do.
17:52
<TabAtkins>
MikeSmith: As an implementor / fellow spec editor, I'm actually not a fan of how extreme Anne takes his "spare, details-only" speccing. Makes it very difficult to tell what the *intent* is of something, or sometimes what it even *means*.
17:53
<TabAtkins>
Despite the occasional implementor trying to interpret an informative detail as if it had normative consequences, I find it's worthwhile overall to give understandable context (to the point that a clueful web author could understand), as it's caught a lot of algorithm bugs when someone was able to go "huh, that result doesn't seem to match what it claims it
17:53
<TabAtkins>
should be doing...".
18:05
<astearns>
"conformance class that we can't actually test" doesn't make any sense to me. The instances I saw were more like, "If you don't do this, you're gonna have a bad time."
18:15
<smaug____>
igrigorik: what does Math.random() r 2 - r; mean in requestIdleCallback spec?
18:17
<wanderview>
JakeA: Domenic: unfortunately I have to stop working streams API stuff in firefox for the time being... trying to find someone else to pick it up
18:17
<Domenic>
wanderview: awww sad! You seemed to be making a lot of progress! What happened?
18:17
<Mek>
smaug____: looks like some kind of markdown processing gone bad (replacing * r * with <em>r</em>
18:18
<wanderview>
Domenic: we need to fix a number of service worker issues in order to enable multiple renderer processes in firefox... thats a high priority thing at the moment
18:18
<Domenic>
wanderview: ah well that's pretty understandable.
18:18
<wanderview>
Domenic: also, I was not making very good progress because I don't really know the js engine stuff very well
18:19
<wanderview>
Domenic: but with it shipped in both edge and chrome (and likely coming soon in safari) I think the pressure is on to implement it
18:19
<smaug____>
you see, if it was a normal webidl based spec it would all be easier and understandabe ;)
18:19
<smaug____>
understanable
18:19
<wanderview>
hoping we can get one or two people with better knowledge on it
18:21
<TabAtkins>
Mek: That's precisely why Bikeshed doesn't have inline-markdown support turned on by default. :/
18:21
<wanderview>
Domenic: if not, then I will circle back to it after we have multi-e10s working
18:43
<Domenic>
Anyone remember the bugzilla bug about radio button names and unicode case matching?
18:52
<annevk>
Domenic: https://www.w3.org/Bugs/Public/show_bug.cgi?id=16970
18:58
<zewt>
TabAtkins: think we've had the conversation a few times before, but I disagree on "authors must"; it's just a confusingly poor choice of terminology
19:00
<botie>
rniwa, at 2016-08-12 08:59 UTC, MikeSmith said: hey somebody finally found an appropriate use for half-width katakana https://github.com/fumieval/LazyKQ
19:04
<Domenic>
annevk: thanks!
19:17
<rniwa>
MikeSmith: that's some crazy language.
20:43
<smaug____>
igrigorik: ping
20:43
<smaug____>
igrigorik: what should timeRemaining() return if JS keeps IdleDeadline after the callback?
20:46
<smaug____>
filed a bug