02:45
<MikeSmith>
isn't the plan still to integrate Streams into the EcmaScript spec itself
02:45
<MikeSmith>
if so it seems like Domenic's approach is the right one
02:48
<MikeSmith>
anyway smaug is fast asleep by now I guess
02:48
<MikeSmith>
we need an inform bot here
02:50
<MikeSmith>
Hixie: would it be OK if I had an inform bot join here?
02:50
<MikeSmith>
kind of like MsgServ but more lightweight and easier
02:50
<MikeSmith>
we could just do like this:
02:51
<MikeSmith>
foobot, inform smaug____ I think if the plan is still to integrate Streams into the EcmaScript spec itself, then Domenic's approach would seem to make a lot of sense...
02:52
<MikeSmith>
and then when smaug joined, the bot would do:
02:52
<MikeSmith>
smaug, at 2014-12-04 00:54 UTC, MikeSmith said, I think if the plan is still to integrate Streams into the EcmaScript spec itself, then Domenic's approach would seem to make a lot of sense...
02:56
<Hixie>
MikeSmith: fine by me
02:56
<MikeSmith>
k
02:57
<MikeSmith>
will try to set one up this weekend
05:25
<MikeSmith>
layout.css.vertical-text.enabled=true
05:33
<MikeSmith>
doesn't work with my test page though
05:33
<MikeSmith>
http://people.w3.org/mike/demo/melos/
05:33
<jgraham>
MikeSmith: smaug____ is probably not asleep yet
05:33
<MikeSmith>
ah yeah yall are in Portland
05:34
<MikeSmith>
glyph-orientation-vertical is what I'm using in my (old) demo. I guess I may need to update that
05:35
<MikeSmith>
"Unknown property 'glyph-orientation-vertical'. Declaration dropped." in console in Nightly
05:35
<smaug____>
not yet
05:35
<MikeSmith>
also "Unknown property 'align'. Declaration dropped"
05:36
smaug____
looks at the logs
05:37
<smaug____>
MikeSmith: oh, right, *if* Streams will be part of ecma, then sure, using whatever odd syntax ecma is using for the specs is fine
05:38
<smaug____>
but it is not clear to me why Streams should be part of ecma
05:52
<MikeSmith>
Hixie: fyi this is the bot I was talking about https://github.com/w3c/infobot
05:53
<MikeSmith>
smaug____: I personally don't know enough about Streams to have a position either way on whether it rightly should be in the ES spec or not
05:54
<MikeSmith>
smaug____: but I do vaguely recall that there was some agreement within TC39 that it should be moved into the ES spec
05:55
<smaug____>
if so, fine. I wouldn't be looking at that spec ever again :p
05:55
<MikeSmith>
but I could be misremembering that and anyway Domenic would know of course
05:55
<MikeSmith>
smaug____: well that's sad to hear
05:55
<smaug____>
(not quite true, I've looked at some ecma drafts occasionally )
05:56
<smaug____>
MikeSmith: but in practice, it would mean I'd expect js engine devs to review it
05:57
<smaug____>
not dom implementation devs
05:57
<MikeSmith>
ah OK. I was going to say that if you're not looking at a particular spec because it's hard for you to read than that could be a sign of the spec having a serious problem
05:58
<MikeSmith>
but if it's more about division of labor (between DOM and JS teams), that's a different thing
05:59
<smaug____>
I guess it is about both. If the spec is easy to read and happens to be somewhere in whatwg, it is likely that I'll read it
06:19
<MikeSmith>
smaug____: ok
07:11
<zcorpan>
Hixie: looks like it's due to https://github.com/Fyrd/caniuse/commit/2455f144fa5a261e7f15f7d03a234ee426e5432c
07:16
<Hixie>
o_O
08:14
<zcorpan>
hmm, for background-color:9.9aa gecko rounds the number so it becomes #0010aa. but for 9.9 it rejects. blink rounds both numbers and dimensions
08:16
<zcorpan>
hmm, scientific notation is interesting. 9e1
10:27
<estellevw>
does anyone want to take a look at my webcomponents talk and give me feedback tonight? My preso is tomorrow morning.
10:27
<estellevw>
(tonight as in it's 9:26 pm in AU right now)
10:27
<estellevw>
oops, wrong IRC channel.
10:27
<estellevw>
sorry for the disturbance
10:27
<estellevw>
though if someone is interested, that would be great too
12:04
<MikeSmith>
Ms2ger: small assistance needed with something
12:04
<Ms2ger>
Shoot
12:04
<Ms2ger>
(I'm in class, though)
12:05
<MikeSmith>
will DM you about it since it's so highly sensitive
12:06
<gsnedders>
Ms2ger: stop procrastinating while in class!
12:08
<Ms2ger>
gsnedders, well, my prof just walked out, so... :)
12:11
<Ms2ger>
... and he's back
12:11
<MikeSmith>
infobot, tell annevk what happened to irccloud
12:12
<MikeSmith>
bah this bot sucks
12:12
<MikeSmith>
oh it's not actually here
12:42
<zcorpan>
TabAtkins: fixed hashless now https://github.com/whatwg/quirks/commit/f84b11e4abb561317c407c1a0d47ff45ad6e4fcf https://github.com/w3c/web-platform-tests/pull/1442
13:49
<JakeA>
wanderview: thanks for thinking through that partial cache stuff!
13:50
<wanderview>
JakeA: np, I've been thinking about how I would implement it for a while... streaming the partial file will be tricky for us due to how our streams infrastructure works
13:50
<wanderview>
thanks for working on the algorithms to make it work, though!
13:50
<wanderview>
I definitely think its something we want to solve
13:50
wanderview
wonders why irccloud is running slow...
14:41
<Domenic>
MikeSmith: smaug____: I was hoping to work with TC39 on making streams part of the ES standard library, but they were not willing to work with me/WHATWG really; they wanted to take over the spec themselves. So that idea kinda floundered. Still, I think it (as well as Encoding and URL, actually) should be part of a theoretical "extended ES standard library",
14:41
<Domenic>
not managed by Ecma, but still appropriate for implementation in the JS engine.
14:58
<smaug____>
well, couldn't any spec be part of the theoretical ES sl
14:58
<smaug____>
whether or not webidl is used for example
15:00
<gsnedders>
What spec it is in is relatively irrelevant.
15:01
<smaug____>
indeed. Where the spec is "published" can be somewhat relevant since it may affect to who reviews and implements it
15:01
<gsnedders>
But at the end of the day, there's some lack of desire to avoid extending the ES stdlib beyond things that cannot be implemented in JS itself.
15:01
<Domenic>
smaug____: IMO specs would need to be platform-independent, so e.g. no use of DOMException etc. And IMO (but not everyone's unfortunately) I should be able to work on JS standard library specs in WHATWG.
15:02
<gsnedders>
What makes a spec a JS stdlib spec?
15:02
<gsnedders>
(honest question)
15:02
smaug____
thinks we should make webidl more platform independent and then use it also for APIs which non-browser implementations want to have
15:03
<smaug____>
and this is mainly from practical point of view. having webidl interfaces tends to just make the spec significantly easier to read comparing to some ways to write specs
15:04
<smaug____>
if we need some other syntax to express new es features, we should change webidl
15:04
<Domenic>
I somewhat agree with that except that WebIDL is so hard to read for anyone used to JS semantics
15:04
<Domenic>
e.g. wtf is an attribute or interface
15:05
<Domenic>
thus https://streams.spec.whatwg.org/#rs-class-definition
15:06
<annevk>
Yeah, we should just rename interface to class and attribute to property
15:06
<Domenic>
I'd like something like `get/set prop` or `get prop`
15:06
<annevk>
And instead of readonly have something like get property
15:07
<Domenic>
yeah
15:07
<annevk>
Seems easy enough to fix
15:07
<Domenic>
but hard to justify given the churn
15:09
<annevk>
Given the upsides it seems worth it
15:09
<annevk>
IDL is after all a way to convey something as well and if it's doing a poor job at that we should fix it
15:12
<smaug____>
why get/set ?
15:12
<smaug____>
readonly attribute is quite good abstraction
15:13
<gsnedders>
presumably because get/set is like the {get foo() { return 1; }} syntax
15:13
<Domenic>
getters and setters are what the semantics actually are
15:13
<annevk>
plus usually an internal slot
15:14
<Domenic>
attributes are not a real thing; they make people think you're talking about HTML attributes (especially because people usually read their first IDL in the HTML spec)
15:14
<smaug____>
yes. and readonly attribute is rather nice abstraction of a getter
15:14
<Domenic>
readonly sounds like non-writable which sounds like non-writable data property which is extremely different from a getter
15:14
<smaug____>
that DOM attributes vs webidl attributes case is a real issue
15:15
<smaug____>
having both getter and setter for each attribute in the interface would be rather annoying
15:16
<annevk>
I don't see how attribute is a nice abstraction of something that's actually called a property
15:16
<smaug____>
harder to read
15:16
<annevk>
smaug____: I don't think anyone is suggesting that per se
15:16
<Domenic>
yeah, I like separate getter setter but in the interests of not being annoying something like `get/set foo` seems fine
15:17
<annevk>
smaug____: although that would open up some possibilities such as accepting more in the setter than in the getter which we've had requests for
15:18
<smaug____>
attribute hides the fact that the thing is implemented using a getter/setter
15:18
<smaug____>
since most cases one shouldn't need to think about that it is a getter or setter
15:18
<gsnedders>
yeah, I find the fact that readonly attributes are actually getter/setter pairs a bit confusing
15:19
<smaug____>
(readonly attributes are getters)
15:19
<gsnedders>
(pedant :))
15:20
<annevk>
smaug____: but if you come from a JavaScript background the term attribute makes no sense and gives you no idea what's going on
15:20
<annevk>
smaug____: that seems way more important
15:20
<smaug____>
depends on the audience
15:21
<smaug____>
if you're some random js dev, you may not really use getters/setters explictly
15:22
<annevk>
but you will know about properties
15:22
<annevk>
and that there are such things as getters and setters
15:22
<annevk>
you won't have a clue about attributes other than DOM attributes which are completely different
15:23
smaug____
is not sure all the devs even know about getters/setters
15:24
<smaug____>
given all the really old documentation etc
15:24
<gsnedders>
I'm just weird coming from the JS engine side nowadays :)
15:33
<JakeA>
annevk: in the env settings object, there's no document for (shared)workers right?
15:34
<annevk>
JakeA: yup
15:35
<annevk>
Domenic: how is it explained in "web implemented in JS" idea how a random object, say <a>, gets to the internal slot of some other object, say document's base URL?
15:36
<JakeA>
annevk: so I'd have to use the creation url for workers. CurrentURL would make that easier. Will suggest that
15:43
<Domenic>
annevk: shared access to the weak map. (Can gist up an example if you like.)
15:44
<annevk>
Domenic: access is given upon creation?
15:45
<Domenic>
Or just using a shared scope
15:45
<annevk>
fair I guess
16:08
<JakeA>
annevk: if memory serves, you were keen on having ServiceWorkerClient represent window clients, and have SharedWorker / Worker represent worker clients. Have I got that right? I'm kinda uncomfortable returning a mix of objects from clients.getAll()
16:13
<JakeA>
Could have different APIs to get each type, or a type param on getAll. Or stick with ServiceWorkerClient that represents all (and provide values for visibility & focus state)
17:21
<annevk>
JakeA: so I was wrong about that
17:22
<annevk>
JakeA: APIs should be similar for each
17:22
<annevk>
JakeA: might want some things different for window vs worker still I suppose
17:35
<JakeA>
annevk: need a type property I think. Also, new ServiceWorkerClient doesn't make sense anymore as a way to create a window. clients.openWindow(url) feels better, but I guess there'll be a "but constructors" argument to have
17:35
<annevk>
JakeA: so is ServiceWorkerClient exposed outside service workers as well?
17:36
<annevk>
JakeA: otherwise we could have Client, WindowClient, WorkerClient
17:36
<annevk>
and not support new WorkerClient I suppose
17:37
<annevk>
I like how mounir announced Google left sysapps after trying to convince me for a while this was the way to go
17:37
<JakeA>
annevk: works for me.
17:38
<annevk>
JakeA: I've lost track a bit of what objects we have where unfortunately, but if this sounds reasonable, great
17:38
<annevk>
JakeA: I should probably read through it again
17:38
<annevk>
JakeA: perhaps we should make a diagram that explains the relationship between the objects
18:39
<hallvors>
annevk: have a moment? I'm where the webcompat team tends to sit (F room)
18:40
<annevk>
hallvors: yeah
19:05
<mounir>
annevk: things are never black or white
19:05
<annevk>
mounir: rainbows
19:06
<jgraham>
Black holes?
19:06
<annevk>
ClampBetween(0, 255) Domenic, that's called ToUint8
19:07
<rubys>
the sun is white: http://blog.chron.com/sciguy/2013/12/what-color-is-the-sun/#28933101=0
19:08
<TabAtkins>
rubys: That's not what my crayon drawings say.
19:09
<rubys>
TabAtkins: tl;dr: the sun is yellow because the sky is blue.
19:09
<TabAtkins>
Yeah, I know. ^_^
19:10
<TabAtkins>
A combination of blue light getting scattered across the sky, so the leftover is more yellowish, and the opponency of our visual pathway contrasting blue against yellow.
19:21
<Domenic>
annevk: unclear whether it's EnsureBetween or ClampBetween IMO.
19:21
<Domenic>
annevk: also unclear why you would use ClampBetween(0, 255) more often than ClampBetween(0, 100) or whatever
19:25
<Domenic>
annevk: exciting to see element.closest() starting to ship. It's almost as if we can actually make the DOM better by adding things to standards :)
19:26
<annevk>
Domenic: but are they stable?!
19:26
<TabAtkins>
Domenic: Because clamping behavior in APIs is almost always dictated by the size of the int used to store it.
19:27
<TabAtkins>
Also: ClampBetween(0, 65536) is obnoxious. ^_^
19:27
<TabAtkins>
65535, rather.
19:27
<TabAtkins>
See?!?
19:27
<Domenic>
TabAtkins: but how often do you actually want to clamp on the web
19:27
<TabAtkins>
Whenever you're dealing with bytes.
19:27
<Domenic>
TabAtkins: colors, *maybe*, otherwise unsure when
19:28
<Domenic>
how many APIs actually deal with bytes instead of just dealing with arraybuffers
19:28
<Domenic>
annevk: made it to Chrome beta... only 6 weeks left
19:28
<annevk>
Domenic: it was a lame joke about the standard
19:28
<Domenic>
annevk: oh lollll
19:32
<Domenic>
so for longdesc, is it productive to just vote against publishing and say "we agree with the previous formal objection"? or is that a lost cause?
19:32
<TabAtkins>
Continue doing minimal-effort objecting.
19:33
<TabAtkins>
I mean, yes, it's a lost cause, but we can at least put forth a modicum of effort showing that everything is bullshit forever.
19:33
<Domenic>
also lol apparently browser extensions count for the purposes of interoperable implementations https://w3c.github.io/test-results/html-longdesc/cr-report.html
19:34
<Domenic>
*also* apparently my TAG status lets me fill out the voting form despite not being the AC rep... I'll quietly close that window I guess and talk to my AC rep.
19:35
<Domenic>
(Oh nevermind: all the form field controls are clickable but there's a "You are allowed to read the questionnaire, but you are not authorized to answer with your current credentials!" at the top of the page)
19:50
<annevk>
imagine if all that time put into longdesc and alt was put into making things accessible
19:51
<tantek>
annevk - have you spoke with dbolter about that while you're in the same room?
19:52
tantek
looks over and sees them at the same table, with only marcosc between them!
19:56
<annevk>
(we talked about getComputedRole())
21:01
<tnelis>
hello, i'm seeking help clarifying the removal of a paragraph in XHR, is this the right place to ask?
21:03
<caitp>
here or bugmail is proabably a good place for it
21:05
<tnelis>
alright let's try; xhr 4.5.6 (the send method), the following paragraph is in f18fc52 and removed in c99000a
21:05
<tnelis>
"For 304 Not Modified responses that are a result of a user agent generated conditional request the user agent must act as if the server gave a 200 OK response with the appropriate content. The user agent must allow author request headers to override automatic cache validation (e.g. If-None-Match or If-Modified-Since), in which case 304 Not Modified responses must be passed through. [HTTP]"
21:06
<tnelis>
i essentially question the intent behind the removal and wonder if this behavior can still be relied on (i would assume so, there'd be some serious breakage otherwise)
21:07
<tnelis>
(also note that it's still in the w3 draft)
21:20
<tnelis>
nvm, found it reworded, just had to look closer
21:22
<annevk>
zcorpan: "If the element does not have a srcset attribute specified and it does not have a parent or it has a parent but it is not a picture element, and it has a src attribute specified and its value is not the empty string, let selected source be the value of the element's src attribute, and selected pixel density be 1.0." is really hard to parse
21:23
<tantek>
that's quite a long sentence
21:27
<annevk>
following that algorithm :-(