01:29
<caitp>
would you say chrome is doing the right thing here? http://jsfiddle.net/eq4La8nr/ (does not emit the `load` event when parsing svg via innerHTML's setter)
06:28
<zcorpan_>
caitp: http://www.w3.org/TR/SVG2/interact.html#LoadEvent seems to say it's window.onload. though http://www.w3.org/TR/SVG2/script.html#OnLoadEventAttribute doesn't say so. the spec seems confused
07:37
<zcorpan_>
Domenic: wait, i thought <hr role=menuitem> did override the role
07:42
<Ms2ger>
krijnhoetmer_, logs down?
07:55
<zcorpan_>
Domenic: nevermind. (it wasn't like that originally though)
07:58
<zcorpan_>
Domenic: i don't understand why you can't use the native element instead of using a custom element
07:59
<zcorpan_>
Domenic: there are also semantics that can't be expressed with ARIA in the first place (like <input type=color>)
08:03
<zcorpan_>
Domenic: maybe you should think about what aria was intended to solve. it was intended to enable accessible custom widgets without waiting for something like XBL to be implemented everywhere, using a mechanism that was available to screen readers on top of IE6 (with no acc mapping, just DOM tree)
08:03
<zcorpan_>
Domenic: this is why there is only setAttribute and no nice DOM API
08:25
<JakeA>
Any indexeddb experts awake? If I have an objectStore without out-of-line keys, how do I create an index ensuring all values are unique? store.createIndex('name', ???, {unique: true});
08:25
<JakeA>
The index should be on the value, not a property within the value
08:26
<JakeA>
This is assuming the values within 'store' are simple values (strings in this case)
08:37
<JakeA>
Answer: store.createIndex('name', '', {unique: true});
08:54
<zcorpan_>
annevk: \o/
08:56
<JakeA>
Papa Van Kesteren
09:38
<smaug____>
blinkers really try to solve all sorts of things with web components
09:39
<smaug____>
I guess "web components" is a buzz word, should be treated like HTML5
09:41
<gsnedders>
So Opera has grown by 253 full-time employees between 2Q13 and 2Q14. I wonder in what departments…
09:42
<gsnedders>
Like, I wonder if this is in the advertising side of the business or not.
09:50
<Ms2ger>
"The x-ua-compatible meta tag (or header) was meant to temporarily repair legacy sites until a long-term solution could be implemented."
09:50
<Ms2ger>
lolololol
09:50
<gsnedders>
where's that from?
09:52
<JakeA>
smaug____: I'm wary of any solution that "just does it for me". From jquery plugins to Angular to React to Web Components
09:53
<Ms2ger>
https://twitter.com/jonathansampson/status/504747785413873664
10:00
<zcorpan_>
things aren't temporary on the web :-(
10:03
<smaug____>
synx XHR is! I will kill it in 5 years
10:05
<Ms2ger>
annevk, congratulations
10:58
<zcorpan_>
Domenic: actually it's still just an authoring requirement like stevef said. "user agents MUST still use the value of the the semantic of the WAI-ARIA role when exposing the element to accessibility APIs." http://www.w3.org/TR/wai-aria/host_languages#host_general_conflict_header
11:36
<mathiasbynens>
annevk: congrats!
11:41
<hsivonen>
krijnhoetmer_: the logs for yesterday don't load for me
11:44
<jgraham>
hsivonen: It appears that the problem is that they don't exist
11:46
<hsivonen>
jgraham: I see.
11:46
hsivonen
wonders what annevk is being congratulated on
12:02
<Ms2ger>
hsivonen, see his blog
12:13
<roc>
cor
12:18
<zcorpan_>
roc: do you happen to have some time to look at https://www.w3.org/Bugs/Public/show_bug.cgi?id=25508#c16 perchance?
12:33
<hsivonen>
annevk: congratulations
12:38
<foolip>
annevk: weeee :)
13:07
<zcorpan_>
is there a good example of placeholder="" that doesn't have usability problems?
13:12
<ondras>
hm, <input type="search" placeholder="Enter search phrase..." /> ?
13:17
<zcorpan_>
ondras: i think http://www.nngroup.com/articles/form-design-placeholders/ argues that that is bad
13:17
<zcorpan_>
although a single search field is probably not so problematic as several fields in a form
13:18
<jgraham>
I would be interested in actual research showing that they perform badly; I only recall seeing theories
13:18
<ondras>
yeah. after skimming through that article, I would argue that most of those points do not apply to my one-search-input scenario
13:29
<zcorpan_>
how about <label>Email <input type=email placeholder="foo⊙bc"></label> (as part of a bigger form)? i guess it suffers from the problem that the user might think that it's already filled in, but i don't know if that's a real issue when the text is gray
13:30
<ondras>
sounds okay to me
13:30
<ondras>
but again, I am somewhat more inclined to placeholder usage, when compared to that article's author :)
13:48
<zcorpan_>
ok i've updated http://html-differences.whatwg.org wrt placeholder
13:50
<caitp>
have you ever noticed how IE will fire an input event whenever the placeholder attribute changes, like ever since they added placeholder?
13:53
<zcorpan_>
caitp: did you see my message about <svg onload>? (i think you weren't here when i sent it)
13:53
<caitp>
was probably sleeping, I missed that
13:53
<zcorpan_>
caitp: http://www.w3.org/TR/SVG2/interact.html#LoadEvent seems to say it's window.onload. though http://www.w3.org/TR/SVG2/script.html#OnLoadEventAttribute doesn't say so. the spec seems confused
13:54
<caitp>
hum
13:55
<zcorpan_>
i think what happened was that an earlier version of svg required a load event when </svg> was parsed or some such, but browsers didn't like that and then the svg wg wanted to align with <body onload> instead
13:56
<caitp>
makes sense
14:01
<caitp>
thanks for the summary, I would have spent hours digging through that and probably not found what I was looking for :>
14:03
<zcorpan_>
filed https://www.w3.org/Bugs/Public/show_bug.cgi?id=26688
14:08
<zcorpan_>
would it be useful with more examples in html-differences?
14:44
<Domenic>
zcorpan_: yeah it looks like I was mega-wrong in a few ways.
14:45
<Domenic>
zcorpan_: regardless the project I am trying to solve is "reimplement the usual HTML elements as custom elements" so using the native element instead of the custom element would be beside the point
14:48
<caitp>
it only seems to be worth it for non-trivial elements, because for trivial stuff you end up writing more, and more complicated code (from what I've seen)
15:28
<Domenic>
caitp: here is pretty much all of the code it takes to implement HTMLDivElement https://github.com/dglazkov/html-as-custom-elements/blob/master/src/idl/CustomHTMLDivElement.idl
15:30
<caitp>
defining getters/setters in JS ends up being more code than letting the bindings generator do the hard work and just implementing quick getters/setters in C++
15:30
<caitp>
so for trivial elements where it's like, 1 reflected content attribute + HTMLelement, it just ends up not being worth it imo
15:51
<Domenic>
Yes, that is why you define it in IDL and have the IDL generate the JS
15:52
<Domenic>
That is why I said that the IDL file is all the code it takes to implement HTMLDivElement
15:53
<caitp>
ah, so that's a bit more than the privatescript generator is doing currently
15:54
<caitp>
unless something nice landed that I didn't notice
18:45
<Hixie>
annevk: you around?
18:46
<Hixie>
zcorpan: do you have any news on https://www.w3.org/Bugs/Public/show_bug.cgi?id=24005 ?
18:46
<Hixie>
zcorpan: or a link for where i can look myself for the news
18:47
<zcorpan>
Hixie: ask fs⊙oc
18:48
<Hixie>
k
18:48
<Hixie>
thanks
20:13
<Domenic>
So what are the UA-facing parts of HTML's ARIA mapping? Just implicit default ARIA roles? The strong native semantics and restrictions are only for authors?
20:13
<Domenic>
s/implicit default ARIA roles/implicit default ARIA roles, properties, and states/
20:49
<Hixie>
Domenic: yeah more or less
20:49
<Domenic>
Hixie: OK. FWIW I think that could be signposted better.
20:49
<Hixie>
i'm kinda hoping that whole section could be spun off into its own spec
20:50
<Domenic>
I feel like SteveF would be happy to take it off your hands :P
20:51
<Hixie>
i suggested it but he declined
21:23
<annevk>
Hixie: yeah now for a bit
21:23
<annevk>
(thanks for all the congrats everyone :-))
21:28
<TabAtkins>
annevk: Well, your contribution was minimal, but your partner deserves plenty of congrats. ^_^
21:28
<annevk>
heh, true story, she did great
21:29
<Hixie>
annevk: i was just gonna ask about what we should do with https://www.w3.org/Bugs/Public/show_bug.cgi?id=12230 , but it sounds like you may have more important things to worry about right now :-)
21:30
<jgraham>
annevk: I hope it all goes well from here :)
21:31
<annevk>
Hixie: I'd really like browsers to fix things; seems unfortunate to have this for just two objects or so
21:31
<Hixie>
yeah
21:31
<Hixie>
couldn't agree more
21:32
<annevk>
I'm also happy to wait longer just like I'm doing with mutation events
21:32
<annevk>
maybe I should add a XXX in the spec though to call out this bug
21:32
<Hixie>
well i don't think anything's gonna change unless we put pressure on something to change
21:36
<jgraham>
The thing that has to change is that someone has to collect data
21:36
<jgraham>
Or you have to spec it I guess
21:37
<jgraham>
Or you wait a few years and see if Servo has to implement it
21:37
<jgraham>
I don't recommend the last option though
21:37
<Hixie>
spec what? the feature we want dropped?
21:37
<jgraham>
Yes
21:37
<Hixie>
that doesn't seem likely to lead to the change we're looking for :-)
21:38
<jgraham>
If no one thinks the value of dropping it is equal to the effort of adding telemetry, then it seems like a very mild want
21:39
<Domenic>
what's this now http://www.whatwg.org/specs/url/2014-07-30/
21:39
<jgraham>
OTOH whenever the spec is out of line with what actually has to be implemented to match other browsers, that's a serious problem
21:40
<Hixie>
jgraham: well, _I_ think the value of dropping it is equal to the effort of adding telemetry... far greater than, even
21:40
<jgraham>
Hixie: so why haven't you added telemetry?
21:41
<annevk>
Domenic: a non-publicised experiment, I suspect we might do a blog post thing around it at some point
21:41
<Hixie>
jgraham: i'm not a browser vendor
21:42
<jgraham>
Hixie: Gecko accepts patches
21:42
<jgraham>
Blink too
21:42
<annevk>
jgraham: not sure that would be effective usage of Hixie's time
21:42
<Hixie>
i don't think spec writers should be in the business of writing code for browsers. it leads to a conflict of interest.
21:43
<jgraham>
I don't see how adding code to run experiments to see what needs to be specced is either a) a waste of a spec author's time or b) a conflict of interest
21:45
<Hixie>
the conflict of interest comes from when i inevitably become interested in doing more than just adding experiments
21:47
<jgraham>
So your argument is that if you add code to test specific hypotheses about which features can be dropped, there is an inevitable slippery slope to you adding more code to actually implement features the way you want?
21:49
<Hixie>
jgraham: yeah. i don't have good impulse control. i'm good at setting boundaries when they're discontinuous, but not when they're a judgement call.
21:50
<Hixie>
(for similar reasons, i've stopped reading books entirely.)
21:51
<jgraham>
OK, I guess
21:52
<jgraham>
Then I guess you need to expend the effort you think removing the feature is worth getting someone else to do the implementation for you
21:52
<jgraham>
(implementation of the telemetry)
21:52
<Hixie>
my concern is that if i have to do work to convince people to do it, then i'm probably wrong that it's worth doing.
21:53
<Hixie>
i don't like it when other spec writers go out and advocate for their dumb features
21:54
<Hixie>
i feel like specs should win or lose purely on their merits rather than on how much one person was able to convince other people to do stuff (especially since often the "other people" are poorly-supervised interns, new employees, or people who got confused by fancy words like "accessibility")
21:58
<jgraham>
Well I agree that making specs a PR battle isn't a good approach. But convincing someone to run an experiment is not morally equivalent to argument-by-exhaustion
21:59
<Hixie>
certainly
21:59
<Hixie>
who should i be trying to convince, any idea?
22:01
<jgraham>
I don't know how easy it is to run that kind of experiment in gecko, but smaug____ might be an obvious person. For Blink I don't know; maybe ask zcorpan
22:03
<smaug____>
hmm
22:05
<smaug____>
ah, about bug 12230
22:06
<smaug____>
last time I was doing something with that stuff, some pages relied on synthetic click to trigger links
22:06
<smaug____>
(and it wasn't just click())
22:06
<smaug____>
but sure, time to re-evaluate
22:32
<smaug____>
bpoulain's response to bz in www-style is hilarious :)
22:38
<Domenic>
what is hilarious is https://groups.google.com/a/chromium.org/forum/#!topic/blink-dev/2-R4XziFc7A
22:38
<Domenic>
(but ok smaug____, I read your thing too, and it is pretty good.)
22:51
<Hixie>
did bpoulain just tell bz to learn about browsers
22:51
<Hixie>
because that is pretty funny indeed
22:52
<tantek>
wow what? ok that's enough to make me click on a forum link.
22:53
<Hixie>
http://lists.w3.org/Archives/Public/www-style/2014Aug/0403.html is the e-mail smaug____ is referring to, i assume
22:54
<smaug____>
yup
22:55
<tantek>
zoms
22:56
<tantek>
and cites dbaron for "a little introductory article". this is gold.
23:48
<zcorpan>
Hixie: ask philipj about adding use counters
23:48
<Hixie>
already did
23:48
<Hixie>
(in the bug)
23:49
<zcorpan>
Hixie: ah ok. (he removes blink features for breakfast)
23:50
<Hixie>
yeah, no kidding
23:50
<Hixie>
guy is scary :-P
23:53
<gsnedders>
anyone that tall is scary