06:10
<howdoi>
is there a stage-4 proposal list?
06:29
<annevk>
KevinMarks_: why should it call out ETag?
07:10
<KevinMarks>
Well, conditional get applies to both etag and last modified
07:11
<annevk>
KevinMarks: the conditional request logic is mostly based on request headers
07:11
<annevk>
KevinMarks: the only reason Last-Modified is mentioned is because that header is safelisted for CORS
07:12
<annevk>
KevinMarks: not entirely sure why ETag isn't, but I'm not sure it matters much
09:14
<zcorpan>
Ms2ger: do you have an opinion on https://github.com/whatwg/html/pull/1449#issuecomment-228803521 ?
09:15
<zcorpan>
since there is machinery already for reflecting enumerated attributes for both impl and testing i don't think making this case-sensitive is necessarily simpler to implement and test
09:19
<annevk>
zcorpan: stepping on the value space seems a bigger problem
09:20
<annevk>
zcorpan: and the long term confusion that causes, e.g., with HTTP methods
09:22
<zcorpan>
annevk: yeah i agree that that is a problem
10:01
<kochi>
zcorpan: quick question: related to the recent change of offsetParent to unclosed node
10:03
<kochi>
zcorpan: if we traverse up the layout tree and find a fixed position element, which is "closed" to the context object, and if it end up that the nearest unclosed one is <body>
10:04
<kochi>
zcorpan: then it should return null, not <body>, right?
10:07
<kochi>
hmm, https://github.com/w3c/csswg-drafts/issues/159 has some follow-up question.
10:11
kochi
is reading the thread
10:35
<zcorpan>
kochi: possibly the spec is not yet what we want it to be. i also don't understand shadow dom well enough yet to be able to say what does and doesn't make sense :-)
11:38
<kochi>
zcorpan: I'm not a frontend engineer so I'm not sure whether the offsetParent usages of what @platosha put in the thread is popular in the wild or not, but my gut feeling is that the spec still needs some tweaks.
11:45
<zcorpan>
kochi: ok. happy to tweak the spec as necessary
11:47
<kochi>
zcorpan: the problem is that what is the expected answer for web developers :)
11:47
<zcorpan>
yeah
12:17
<Ms2ger>
> we can remove the "gone async" concept
12:17
<Ms2ger>
\o/
12:17
<annevk>
Actually making the patch is the hard part
12:17
<annevk>
Well, maybe thinking it all through is too
12:18
<annevk>
Since I keep hitting new things
12:19
<kochi>
zcorpan: I posted comments on the thread. In "the nearest ancestor unclosed element", does "ancestor" mean box tree based one? (I'm not familiar with CSS terminology yet)
12:19
<kochi>
if so, an element with fixed position does not have an ancestor, right?
12:21
<zcorpan>
kochi: i think the walk is supposed to be on the DOM tree per spec
12:23
<kochi>
oh, then returning document.body for the case sounds correct, if we follow the spec strictly.
12:25
<zcorpan>
nox: SimonSapin: what about calc(10% - 10px) ? (which quirk?)
12:26
<nox>
zcorpan: Line height calculation quirk.
12:26
<SimonSapin>
zcorpan: https://quirks.spec.whatwg.org/#the-line-height-calculation-quirk "
12:26
<SimonSapin>
The border-right-width, border-left-width, padding-right and padding-left properties have a used value of zero.
12:26
<SimonSapin>
"
12:26
<nox>
This is 0 if containing block is 100px wide.
12:26
<nox>
But should that trigger the quirk?
12:27
<zcorpan>
i suppose so, per the current spec :-) but i'm happy to disable the quirk when calc() is used
12:27
<SimonSapin>
nox: we should (by which I mean do you wanna?) write test cases and see what other impls do
12:28
<nox>
zcorpan: I think we could just say "have a computed value that is definitely zero", if current browsers don't trigger quirk in such a case.
12:28
<nox>
SimonSapin: We should indeed. I may do it when I'm done with my webrender stuff.
12:28
<Ms2ger>
As long as "is definitely zero" is well-defined
12:29
<nox>
Ms2ger: Yes of course.
12:29
<SimonSapin>
s/definitely zero/0px or 0%/
12:29
<zcorpan>
http://w3c-test.org/quirks-mode/percentage-height-calculation.html has tests for this quirk but they do not match the current spec, they match a previous version of the spec. i had to change the spec to make <body> fill the viewport
12:29
zcorpan
bbiab
12:30
<nox>
SimonSapin: Fair enough.
13:55
<zcorpan>
SimonSapin: the line-height calculation quirk should maybe take writing modes into account
14:20
<SimonSapin>
zcorpan: maybe... if impls do
14:21
<SimonSapin>
If impls need to change I'd rather remove the quirk than make it nicer
14:49
<zcorpan>
SimonSapin: i think these two quirks are ones that can't be removed. note they are also present in limited-quirks which is probably a bigger set of pages than those that are in quirks mode
14:57
<Ms2ger>
SimonSapin, we're talking about this because even Servo hit webcompat issues from not implementing the quirk, right?
15:36
<Domenic>
mathiasbynens: did you ever propose some way to get the length of a JS string in code points?
16:20
<mathiasbynens>
Domenic: no actual proposal, but I may have started an es-discuss thread about it
16:20
<Domenic>
seems more useful than .at() ;P
16:20
<mathiasbynens>
Domenic: agreed
16:20
<mathiasbynens>
Domenic: https://esdiscuss.org/topic/how-to-count-the-number-of-symbols-in-a-string
16:21
<mathiasbynens>
(…and `codePointAt`, hah!)
16:22
<mathiasbynens>
Domenic: really depends on the use case though — some systems count code points, others grapheme clusters
16:22
<Domenic>
right yeah
16:22
<Domenic>
I guess I was mostly thinking about this HTML issue https://github.com/whatwg/html/issues/1467
16:22
<mathiasbynens>
the latter is probably more useful in general, but then again the “new” ES2015 APIs all deal with code points
16:23
<Domenic>
If we changed textarea[maxlength] to validate code point length then having something in the JS API you could compare it to would make sense
16:23
<mathiasbynens>
I see
16:23
<Domenic>
[...string].length lol
16:23
<mathiasbynens>
well, there’s always [...string].length
16:23
<mathiasbynens>
GMTA
16:23
<Domenic>
probably shorter than anything we come up with
16:24
<Domenic>
https://esdiscuss.org/topic/how-to-count-the-number-of-symbols-in-a-string#content-11 is fairly compelling too
16:26
<mathiasbynens>
yeah
16:32
<zcorpan>
Domenic: thx for reviewing as="". will check again later tonight or tomorrow. :-)
16:32
<Domenic>
:)
16:56
<SimonSapin>
zcorpan: Ms2ger: yeah, the removing bit is just wishful thinking. It's for the writing mode sensitivity we should write test cases
17:19
<annevk>
https://github.com/whatwg/html/commit/b063361829de68dc0907ecc0764e2582c105cf05 has some of the navigate restructuring I'm planning < Manishearth
17:20
<annevk>
Needs a lot of polish still
17:20
<annevk>
And even that is just a rather minor step of a whole lot of changes that are needed 😕
19:06
<wanderview>
Domenic: is step 9.b.i of ReadableByteStreamControllerEnqueue() correct? https://streams.spec.whatwg.org/#readable-byte-stream-controller-enqueue
19:06
<wanderview>
Domenic: seems like it should be asserting its a default reader and not that the reader is unlocked?
19:08
<Domenic>
wanderview: so it is definitely a default reader; an assert might help. now trying to figure out if the assertion of it being unlocked is false...
19:08
<wanderview>
Domenic: the style in other places has been to assert the default reader there
19:08
<wanderview>
other places in the streams spec
19:09
<Domenic>
wanderview: no, I think it's correct. there are 3 cases: default reader, byob reader, no reader
19:09
<wanderview>
Domenic: like step 6.a of https://streams.spec.whatwg.org/#readable-byte-stream-controller-commit-pull-into-descriptor
19:09
<Domenic>
wanderview: step 8 => default reader. 9a => byob reader. 9b => no reader
19:09
<wanderview>
Domenic: oh, I see... ok... thanks!
19:09
<Domenic>
I think there should be some de-indenting and maybe an additional assert
19:09
<Domenic>
I will do a clarifying PR
19:15
<wanderview>
Domenic: what is the purpose of the SameRealmTransfer in ReadableByteStreamControllerEnqueue? I thought we were trying to avoid copying the buffer
19:16
<Domenic>
wanderview: transfer, not copy!
19:16
<wanderview>
Domenic: where else is this concept used?
19:16
<Domenic>
wanderview: in HTML a bunch
19:17
<Domenic>
wanderview: postMessage
19:17
<wanderview>
hmm
19:17
<wanderview>
Domenic: the whole goal here is to make the original thing that script has a hold of detached?
19:18
<Domenic>
wanderview: exactly, yeah
19:18
<Domenic>
https://blog.domenic.me/reading-from-files/ kind of goes over it, in a more abstract setting
19:19
<wanderview>
ok, thanks
19:19
wanderview
leaves a big TODO comment there for now.
19:19
<Domenic>
^_^
19:21
<wanderview>
Domenic: the js engine guys pointed me to a serialize/deserialize thing... but I don't think thats what we want here
19:21
<wanderview>
since thats a copy
19:21
<Domenic>
agree
19:21
<Domenic>
this should probably be pretty simple; Mozilla's lth was leading the charge on an ArrayBuffer.transfer method proposal
19:22
<Domenic>
I guess it's in nightlies https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/ArrayBuffer/transfer
19:22
<Domenic>
Ah that wasn't lth though
19:22
<wanderview>
Domenic: this is only needed for ArrayBuffers in streams spec?
19:23
<Domenic>
wanderview: in the streams spec, only ArrayBuffers are transferred, if that's what you're asking
19:23
<wanderview>
yes
19:49
<wanderview>
Domenic: should step 8.b here say "set ready" instead of "let ready"? since ready was already declared further up
19:49
<wanderview>
https://streams.spec.whatwg.org/#readable-byte-stream-controller-fill-pull-into-descriptor-from-queue
19:49
<Domenic>
wanderview: yep
19:52
<wanderview>
Domenic: want me to write an issue or do you have time to just fix it?
19:52
<Domenic>
wanderview: fixed :)
19:53
<wanderview>
thanks