00:36
<rniwa>
TabAtkins, Hixie: yt?
05:08
<mustang89>
holy crap i didn't even know mirc was around still looks like it did in 99
05:08
<mustang89>
almost
06:47
<zcorpan>
should the whitespace after the image in the second row be collapsed per spec? http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1417
06:48
<Hixie>
white-space:normal?
06:48
<Hixie>
if so, yes, white space at the end of the line is collapsed in 'normal'
06:49
<Hixie>
see css2.1 bottom of section 10, iirc
06:51
<zcorpan>
i don't see anything about this in section 10
06:52
<Bhanu>
what is irc?
06:52
<Bhanu>
I saw it first time
06:52
<Bhanu>
can anyone tell me
06:52
<Bhanu>
what is irc?
06:54
<zcorpan>
Hixie: "Each tab (U+0009), carriage return (U+000D), or space (U+0020) character surrounding a linefeed (U+000A) character is removed if 'white-space' is set to 'normal', 'nowrap', or 'pre-line'." (in 16.6.1)?
06:55
<zcorpan>
Bhanu: https://en.wikipedia.org/wiki/IRC
06:56
<zcorpan>
oh, no, it's "If a space (U+0020) at the end of a line has 'white-space' set to 'normal', 'nowrap', or 'pre-line', it is also removed."
06:59
<zcorpan>
hmm. there was a site relying on it *not* being collapsed (and firefox doesn't collapse it). at least in almost standards mode/quirks mode.
07:00
<zcorpan>
wonder if we should try to follow webkit/spec and break the site or specify not to collapse it in almost/quirks
07:03
<Bhanu>
thanks zcorpan
07:05
<zcorpan>
np Bhanu, and welcome
07:05
<zcorpan>
what brought you here?
07:30
<annevk>
was it some kind of special day yesterday? no new email
07:43
<Bhanu>
sorry to delay... zcorpan. I brought to here by reading a book named "HTML5 for web designer"
08:24
<zcorpan>
so about the global scope pollution thing, would it be a reasonable compat compromize to limit it to quirks mode *and almost standards mode* ?
08:26
<Ms2ger>
Currently the table thing is the only limited quirk, right?
08:27
<zcorpan>
the line height calculation quirk, yeah
08:28
<annevk>
still no answer from anyone http://lists.w3.org/Archives/Public/www-dom/2011JulSep/0153.html
08:28
<annevk>
meh
08:29
<hsivonen>
zcorpan: seems like a slippery slope to let Almost Standards and Standards diverge some more
08:29
<annevk>
same: http://lists.w3.org/Archives/Public/www-dom/2011OctDec/0265.html
08:30
<annevk>
guess I'll just file bugs on those too
08:30
<zcorpan>
hsivonen: maybe. but if the choice is that and having global scope pollution in standards mode...
08:30
<Ms2ger>
Meh, I think EnforceRange is silly
08:30
<annevk>
you can say so in the bug
08:31
<zcorpan>
annevk: i think we should allow manipulation of xmlns attributes and let namespace algorithms do what they do (i.e. don't try to prevent people to shoot themselves in the foot)
08:32
<annevk>
do what they do?
08:32
<annevk>
not sure they look at xmlns attributes now
08:33
<zcorpan>
i thought they do, but i never researched the namespace algorithms properly
08:34
<zcorpan>
lookupPrefix i think looks at xmlns attributes
08:34
<zcorpan>
in some impls at least
08:35
<zcorpan>
per spec as well it seems
08:37
<zcorpan>
annevk: note that the html parser can produce namespace declarations that are wrong, so we're not making sure the tree doesn't have wrong namespace declarations by restricting setAttribute*
08:37
<zcorpan>
e.g. <svg xmlns="foo">
08:37
<annevk>
Ms2ger: fyi, just filing these bugs so I don't have to keep track of the email anymore, I don't actually believe we should fix all of them
08:38
<Ms2ger>
Yeah
08:38
<annevk>
inbox <250 teehee
08:38
<Ms2ger>
90
08:39
<annevk>
zcorpan: the spec does not look at the namespace of xmlns?
08:39
<annevk>
hmm nope
08:40
<annevk>
bug?
08:42
<zcorpan>
i think we should do the minimal set of checks that work when the attributes are correct
08:42
<zcorpan>
and let the result be bogus when they are not correct
08:43
<zcorpan>
just like we don't care about innerHTML making sense when modifying the tree in a way that text/html cannot represent
08:46
<annevk>
i guess that's fair
08:47
<annevk>
so, does it match implementations? :)
08:48
<zcorpan>
last time i checked impls aren't anal about keeping namespace declarations correct
08:48
<zcorpan>
but with little interop
08:48
<annevk>
serializing will also be a bitch :/
08:50
<zcorpan>
serializing should rewrite namespace prefixes to match the actual namespaces, iirc. but yeah, that's a bitch
08:50
<zcorpan>
and browsers don't actually do that last i checked :-)
08:51
<annevk>
I think Gecko might
08:51
<annevk>
but that still leaves namespaceless xmlns
08:51
<annevk>
meh
08:51
<Ms2ger>
zcorpan, a spec for that would be appreciated :)
08:52
<zcorpan>
Ms2ger: i'll try to finish this quirk thing first :-P
08:53
<Ms2ger>
I guess that's fair
09:42
<jgraham>
zcorpan: Is there any evidence that almost-standards mode covers a significant fraction of the sites that depend on the quirk?
09:43
<jgraham>
(the global scope polluter, I mean)
09:54
<zcorpan>
jgraham: no. though i haven't seen any evidence about standards mode usage, either, apart from the demo
09:55
<zcorpan>
all i know so far is adrian saying "i think this will break many sites in IE"
10:25
<zcorpan>
matjas: why do you have a mapping for zero width space?
10:26
<matjas>
zcorpan: surprisingly, i need it quite frequently. mostly when tweeting e.g. @font-face without turning it into a mention
10:28
<matjas>
zcorpan: it also makes it a bit easier to test js-escapes and css-escapes for white space edge cases (although i still need https://github.com/mathiasbynens/dotfiles/blob/master/.functions#L53-57 for proper testing)
10:28
<zcorpan>
k
11:27
<zcorpan>
hmm. http://simon.html5.org/specs/quirks-mode#the-blocks-ignore-line-height-quirk applies in almost standards mode too
12:26
<smaug____>
why do I have some many copies of DOM4 open..
12:26
<smaug____>
AppTab helps
12:29
<Ms2ger>
Because it's such a nice spec? ;)
12:40
<zcorpan>
foolip: <!-- As of
12:40
<zcorpan>
2005-12, studies showed that around 0.2% of pages used the
12:40
<zcorpan>
<image> element. -->
12:40
<zcorpan>
foolip: changing <image> now seems disruptive even if it's less than 0.2% today
12:41
<foolip>
zcorpan, yeah, I guess :(
12:42
<zcorpan>
foolip: having multiple <img>s seems to have the wrong behavior in legacy browsers
12:42
<foolip>
zcorpan, because they are fetched even or display:none? (are they?)
12:42
<zcorpan>
they are, yeah
12:42
<foolip>
again, :(
12:43
<zcorpan>
and this feature could be designed to work without needing CSS for legacy browsers
12:44
<foolip>
without even CSS? that seems excessively backwards compatible
12:44
<zcorpan>
the proposed markup doesn't need css to show a single image in legacy browsers
12:47
<foolip>
Yes, but I'm not a fan of using <source>
12:47
<zcorpan>
that's reasonable
12:47
<foolip>
Maybe I wouldn't care if I didn't know about the resource selection algorithm
12:47
<zcorpan>
call it something else :-)
12:57
<zcorpan>
seems to me it should be more like
12:57
<zcorpan>
<picture>
12:57
<zcorpan>
<source src=mobile.jog resolution=20dpi>
12:57
<zcorpan>
<source src=hires.jpg resolution=200dpi>
12:57
<zcorpan>
<img src=lowres.jpg resolution=96dpi alt="fallback text">
12:57
<zcorpan>
</picture>
12:58
<zcorpan>
where the browser downloads the image that is "good enough" for the current zoom level, and never downloads a lower resolution image than the ones it already has
12:59
<foolip>
um, so <img> takes part in the resource selection and is not just a fallback?
12:59
<zcorpan>
supporting full MQ here probably leads to authors using it incorrectly, like the spec currently has it wrong (it would download the mobile version when you zoom in)
12:59
<zcorpan>
yeah, that seems nice for DRY
13:00
<zcorpan>
and making sure legacy browsers don't get left in the cold
13:00
<zcorpan>
maybe the <img> has an implied resolution=96dpi and that can't be changed
13:00
<foolip>
yeah, but it'll just look weird in 5 years when legacy browsers are irrelevant
13:00
foolip
-> fika
13:00
<zcorpan>
so?
13:01
<zcorpan>
in five years bandwidth won't be a problem so you can just use <img> again :-)
13:01
<zcorpan>
also, lots of stuff in html looks weird because they were designed to be backwards compatible
13:24
<zcorpan>
media="" on <source> might well be a misfeature that we should drop
13:27
<zcorpan>
ok i commented on the picture spec
13:51
<zcorpan>
Hixie: btw, seems i was wrong about coordinate vs coordinates
13:52
<jgraham>
zcorpan: One might wonder if your <img adaptive> proposal couldn't be spelt <img>
13:52
<jgraham>
I guess that will break something for some reason
13:53
<zcorpan>
jgraham: i suspect making Range requests for all <img>s would break sites
13:53
<zcorpan>
but i don't have data on that
14:05
<zcorpan>
or pages might have progressive JPEGs but expect them to be fully downloaded so they can paint them on a canvas or so
14:07
<zcorpan>
also, an attribute can be feature-checked so you could choose to use a 96dpi image for legacy browsers and a 200dpi image for supporting browsers
14:09
<zcorpan>
which maybe could be spelled as <img src=96dpi.jpg adaptive=200dpi.jpg> (empty string for adaptive would use src instead)
14:15
<jgraham>
Interesting idea
14:15
<jgraham>
The one I like most so far, I think
14:19
<zcorpan>
ok updated my comment with that
15:24
<Wilto>
My ears were ringing. Hi, guys!
15:26
<Wilto>
jgraham: Unfortunately, we were informed pretty early on by a number of vendors that they weren’t willing to bypass image prefetching based on the existence of an attribute.
15:27
<Wilto>
jgraham: <img src="…" adaptive="…"> would still leave us downloading the original src first, then swapping it for the second src.
15:28
<Wilto>
The Chrome folks did all but post “LOL” somewhere. Hence our working around <img> altogether—for that reason, and to avoid introducing potential issues in older browsers (granted, the `adaptive` attr wouldn’t, but in general).
15:29
<Wilto>
We did a lot of that early brainstorming here: https://etherpad.mozilla.org/responsive-assets
15:29
<Wilto>
And a lot of this ground has been covered in the CG, even though it’s a bit distributed across posts and comments. We did our best to keep all conversation public and easily-parsed.
15:33
<Wilto>
It's been a disjointed "process," but there isn’t a hell of a lot of precedent for this "from Community Group to standard" chain of events. If theres anything I can do to help out with the history, please don’t hesitate to pester me. I turned into the spokeperson somewhere along the line.
15:35
<Wilto>
And I want to stress that I'm after _a_ solution, not "my" solution. I’m not married to <picture>, but all roads seem to lead that way the longer the discussion goes on. Completely open mind, here.
15:36
<Wilto>
(End rant.)
15:58
<jgraham>
10:27 < Wilto> jgraham: <img src="…" adaptive="…"> would still leave us downloading the original src first
15:58
<jgraham>
why?
15:59
<jgraham>
Wilto: ^
15:59
<jgraham>
AIUI in zcorpan's proposal you download exactly one image, ever
16:00
<jgraham>
It's just that you use http range requests to only download enough data for the resolution you have
16:01
<jgraham>
ofc it doesn't solve the use case of "display entirely different images based on screen size"
17:33
<volkmar>
is there an api out there with something like success()/succeded()/postSuccess() methods? I'm trying to name a method with a bit of consistency...
17:38
<AryehGregor>
volkmar, IDB comes to mind.
17:38
<AryehGregor>
Resource loads tend to use onload/onerror.
17:39
<volkmar>
actually, I was looking for something doing the call, not handling the event
17:40
<volkmar>
but I guess I will just go with something and hope someone will tell me if I'm wrong
20:14
<Ms2ger>
Oh hey, it's jgraham in my bugzilla
21:10
<jgraham>
Ms2ger: Be afraid, very afraid
21:11
<Ms2ger>
So, when are you guys implementing event ctors?
21:15
<jgraham>
Ms2ger: Wrong tense
21:15
<Ms2ger>
Oh?
21:16
<jgraham>
Unless you mean more than "new Event('foo')"
21:16
<Ms2ger>
Yep
21:16
<Ms2ger>
StorageEvent
21:18
<jgraham>
Ah. Probably when jl (or sof or bratell or a number of other people) has five minutes and the inclination then.
23:15
<Hixie>
hmm, i suppose if path is null and we're using the current default path i should throw if a transform is provided to addHitRegion...
23:18
<Hixie>
any dom core editors around?
23:18
<Hixie>
actually i guess MouseEvent isn't in DOM Core
23:18
<Hixie>
what spec defines the MouseEvent constructor these days?
23:23
<paul_irish>
http://html5labs.com/dom4events/#constructors-mouseevent maybe?
23:23
<Hixie>
ah, cool, thanks
23:24
<jwalden>
anyone have a recent copy of IE10 to test something for me? I need to know what |Object.create(new DataView(new ArrayBuffer(8))).setUint8(2, 5)| in http://www.squarefree.com/shell/shell.html produces there
23:24
<Hixie>
heycam: can i do "partial dictionary Foo { foo myNewMember }" ?
23:25
<heycam>
you cannot, sorry
23:25
<heycam>
did you ask that recently, or was that someone else?
23:25
<heycam>
one problem with allowing it is that we need to have a defined order for dictionary members, so that the properties can be got in a fixed order
23:26
<heycam>
with partials, you don't have any ordering between the different fragments
23:26
<jwalden>
specifying property enumeration order? :-(
23:26
<Hixie>
heycam: hmm
23:27
<heycam>
jwalden, nah, just order of fetching properties from an object that's being used as a dictionary
23:27
<jwalden>
heycam: oh, you mean like for Object.defineProperty(..., ..., { enumerable: true, configurable: true }) what order enumerable/configurable are gotten?
23:27
<jwalden>
that ordering I am happy to have spec'd :-)
23:27
<heycam>
jwalden, yeah
23:27
<jwalden>
good, good
23:28
<Hixie>
heycam: i think it's going to become almost necessary given events
23:28
<heycam>
hmm really?
23:28
<Hixie>
heycam: and their constructors
23:28
<heycam>
why's that?
23:28
<Hixie>
brb
23:29
<jwalden>
event constructors taking dictionaries means you have to say what order the properties are retrieved, so that getters can have effects in the proper order and all, I think
23:30
<jwalden>
and I understand the event ctors to go crazy with dictionaries, because passing all of 'em as arguments is cracktastic
23:30
<heycam>
yeah, it's a much nicer pattern to use
23:30
<heycam>
still not sure why partial dictionaries are required though
23:32
<Hixie>
heycam: sorry, back
23:32
<Hixie>
heycam: say i want to add a member to MouseEvent
23:32
<heycam>
and you don't want to ask annevk to do it? :)
23:32
<Hixie>
heycam: how do i do that, assuming i'm not a mainstream spec (i.e. who can convince microsoft ot add it)
23:33
<heycam>
fork the spec? :)
23:34
<Hixie>
that seems like a bit of a heavyweight for one extra member :-P
23:34
<heycam>
how are you going to use this extra member?
23:34
<heycam>
I mean, what function is going to look at it?
23:35
<Hixie>
i would want to add a member to the dictionary so that it can be passed to the constructor so that it can set a new member i would also add to the MouseEvent interface
23:35
<Hixie>
same way as all the other members on event init dictionaries
23:35
<heycam>
I see, so you wouldn't need to redefine the constructor, because it's already written in terms of whatever members exist on the dictionary
23:36
<Hixie>
right
23:37
<heycam>
if we have two partial dictionary definitions, how do we decide how they are ordered?
23:37
<heycam>
currently dictionary members are defined to be got in the order they appear in the IDL
23:37
<heycam>
with ones on ancestors got before ones on descendants
23:37
<Hixie>
change the definition to be alphabetical :-)
23:38
<heycam>
yeah that's one way! :)
23:39
<paul_irish>
jwalden: browserstack has ie10pp2 to play with
23:40
<jwalden>
paul_irish: ooh, it does? good to know for next time (I got an answer in #developers a second ago)
23:40
<jwalden>
although, I think pp2 is before they went to only-available-with-a-new-OS system, and I actually have that preview installed
23:41
<paul_irish>
correctomundo
23:41
jwalden
is not interested in blowing away his working OS solely to test IE10 :-)
23:41
<heycam>
Hixie, I guess we could even be consistent with current ordering (in case there is a reliance on order already, which I guess is unlikely) by doing the alphabetical thing only for the partial-included dictionary members
23:42
<Hixie>
heycam: we could do that, yeah. though generally i'm a bit suspicious of making partial dicts or interfaces having detectable results since the idea is that 'partial' is just an editorial aid
23:43
<heycam>
Hixie, yeah I guess so
23:44
<heycam>
the alphabetical order just seems a bit funny to me
23:45
<WeirdAl>
[partial] Dictionary
23:45
<Hixie>
could order it by the sum of the unicode codepoints of the characters in the name, it's even more silly :-P
23:45
<WeirdAl>
extra attributes ftw :p
23:45
<heycam>
heh
23:45
<jwalden>
normatively order it depending on the phase of the moon
23:46
<heycam>
mint a uuid for each partial dictionary, order based on that
23:47
<Hixie>
that goes back to the editorial details having normative effects