04:17
<MikeSmith>
so I am planning to spend some time next week doing an experimental implementation of a language-detection feature for the HTML checker
04:18
<MikeSmith>
but I am wondering what methodology to use for getting the best sample of text from a document to check in the normal case
04:19
<MikeSmith>
general case
04:19
<MikeSmith>
it seems clear I want to begin but only looking at some text from some point in the body
04:19
<MikeSmith>
ignoring the head
04:20
<MikeSmith>
and ignoring any text nodes from <script>
04:21
<MikeSmith>
but if I do the naive thing of just looking at the first couple thousand characters of a given document, it many cases that is just going to amount to looking at at Table of Contents or some such
04:23
<MikeSmith>
which will actually probably work fine in practice as far as identifying the language of the document, but then my implementation amounts to being a more specific language-identification-by-only-looking-at-front-matter tool
04:24
<MikeSmith>
so I am wondering if there are better heuristics for detecting when I have hit the actual main text flow of a document
04:26
<MikeSmith>
btw this needs to not introduce a requirement significantly increasing the amount of memory the checker uses, so I can’t do the really simple thing of just sucking the whole document and running the language detector on the whole doc
04:28
<MikeSmith>
one simple idea is just to not start collecting any characters to check until I hit the first <p>
04:31
<MikeSmith>
another is to not do it until I hit the first h2 or h3 or h4
10:09
<howdoi>
wasup with Array.prototype.values?
10:09
<howdoi>
String.prototype.contains also had issues right?
10:10
<mathiasbynens>
howdoi: yeah, so it was renamed to `includes`
10:10
<howdoi>
mathiasbynens: reason being?
10:11
<howdoi>
not a security thinge right?
10:11
<mathiasbynens>
howdoi: not web-compatible https://esdiscuss.org/topic/having-a-non-enumerable-array-prototype-contains-may-not-be-web-compatible
10:11
<howdoi>
it broke some other code AFAIR
10:11
<howdoi>
MooTools! Thanks I remember reading that mathiasbynens
10:11
<howdoi>
that was fast!
10:12
<howdoi>
and with values, what's the issue?
10:12
<howdoi>
we don't even need .values(), for-in does a good job
10:13
<howdoi>
or even Symbol.iterator
10:13
<mathiasbynens>
howdoi: see https://bugs.chromium.org/p/v8/issues/detail?id=4247#c12 which points to https://bugs.chromium.org/p/chromium/issues/detail?id=615873
10:16
<howdoi>
mathiasbynens: Array.values would have been better?
10:34
<mathiasbynens>
howdoi: who knows
10:34
<howdoi>
:{
12:29
<Greg_>
Hello
16:02
<AutomatedTester>
annevk: is there a list of idl attributes that are not element attributes anywhere?
16:08
<Ms2ger>
"all of them"? :)(
16:09
<Domenic>
AutomatedTester: assuming you mean "IDL attributes that don't reflect element attributes", there's not yet, but you could get reasonably close by excluding those with [CEReactions]
16:10
<Domenic>
That will exclude some that should be included, since [CEReactions] really means "can modify the DOM in certain ways". But everything that is a reflected attribute will have [CEReactions].
16:11
<AutomatedTester>
Domenic: ahh cool
16:11
<AutomatedTester>
Domenic: I can see those in the html spec
16:11
<AutomatedTester>
I will make a list
16:12
<Domenic>
One day we will spec [Reflect]
16:12
<AutomatedTester>
Domenic: wrt to WebComponents, how would Reflect work
16:12
<Domenic>
AutomatedTester: what do you mean exactly
16:13
<Domenic>
[Reflect] would just be a way of expressing in IDL the spec's current "The target IDL attribute must reflect the content attribute of the same name."
16:13
<AutomatedTester>
Domenic: normal reflective attributes do not need to be reflective in web components iirc
16:13
<Domenic>
It's fairly unrelated to web components
16:13
<Domenic>
Web components can create whatever API they want
16:13
<Domenic>
Good ones will, IMO, have APIs that reflect their valid attributes
16:13
<Domenic>
but that'd be done manually
20:13
<TabAtkins>
Mek: Looks like the problem should be fixed? Domenic, you too.
20:14
<TabAtkins>
lmk if there are still refs missing.
20:39
<Mek>
TabAtkins: yeah, looks fixed
21:22
<wanderview>
love the broken emoji in this tweet https://twitter.com/webkit/status/746441587811508227
21:22
<wanderview>
at least on my machine/platform
21:23
<nox>
wanderview: It works on Apple computer, it's a private use code point.
21:23
<nox>
wanderview: It's Apple's logo.
21:23
<wanderview>
yea, thats what I was guessing
21:23
<wanderview>
just drives the point home, I guess