02:29
<erlehmann>
there is no redirect? http://www.whatwg.org/specs/web-apps/current-work/multipage/converting-html-to-other-formats.html#atom
02:31
<erlehmann>
i see http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2010-June/068847.html
02:36
<erlehmann>
http://edward.oconnor.cx/2013/03/herodotus-on-standards-work
02:56
<Hixie_>
erlehmann: i think that section is long gone
02:57
<erlehmann>
Hixie_ yeah, i see. can't you add a note about that?
02:57
<erlehmann>
it really took me some time to find out that it has not moved around or something
02:57
<erlehmann>
because of software changes
02:57
<erlehmann>
and was just scrapped
02:57
<Hixie_>
sure. file a bug. not sure where i'd put it exactly, or what it should say, but i can try
02:59
<erlehmann>
Hixie_ where can i file bug?
02:59
<erlehmann>
i'll try to figure it out
02:59
<Hixie_>
whatwg.org/newbug
03:00
<erlehmann>
thx!
03:00
<erlehmann>
www.w3.org
03:00
<erlehmann>
hahaha
03:00
<erlehmann>
in the spirit of slashdot.org
05:17
<zcorpan>
hsivonen: MikeSmith: https://github.com/svenkreiss/html5validator
05:32
<karlcow>
wondering who is in charge at Apple for Mixed Content Blocking issues https://bugzilla.mozilla.org/show_bug.cgi?id=881485
06:41
<Ms2ger>
"This is not enough quorum to discuss anything meaningful so we ended the call after 15 minutes."
06:42
<Ms2ger>
One wonders how it takes 15 minutes to figure that out
07:33
<ato>
Ms2ger: Maybe only non-meaningful things were discussed.
07:56
<jgraham>
othermaciej: http://w3c-test.org/url/
08:22
<annevk>
ah, was just about to answer that :-)
09:28
<annevk>
hsivonen: are you going to a do a blogpost on DNSSEC one day?
09:31
<foolip>
annevk: do you know which spec defines how Web browsers detect the encoding of text/plain input, i.e. the heuristic that looks for a UTF-8 BOM, valid UTF-8 sequences and that kind of thing?
09:32
<jgraham>
HTML, I imagine
09:33
<annevk>
foolip: currently we don't have sniffing for text/plain
09:33
<foolip>
http://www.whatwg.org/specs/web-apps/current-work/#encoding-sniffing-algorithm seems relevant
09:33
<annevk>
foolip: if user agents employ sniffing for text/plain, that should probably be defined in HTML
09:34
<foolip>
I said text/plain because I'm interested in a good heuristic that isn't HTML-specific
09:34
<foolip>
for a non-Web, non-work thing
09:34
<annevk>
the current rules are only checking BOM for text/plain
09:34
<foolip>
jgraham: detecting the encoding of files loaded in Critic, in fact
09:34
<jgraham>
From a web-browser point of view text/plain is very like HTML
09:34
<annevk>
no hueristics
09:35
<jgraham>
foolip: Ah. Isn't the canonical solution there to use chardet, or something?
09:36
<foolip>
jgraham: dunno, molsson asked me and all I know is Web stuff :)
09:36
<jgraham>
Heh
09:36
<Ms2ger>
def detect_charset(content):
09:36
<Ms2ger>
return "utf-8"
09:37
<foolip>
Ms2ger: that is a serious option, yes :)
09:39
<jgraham>
try: input.decode("utf-8"); except UnicodeDecodeError: review.add_issue("Your source is in the wrong encoding. Please use UTF8.")
09:39
<foolip>
jgraham: yes, or input.decode("utf-8", errors="replace") :)
09:41
<Ms2ger>
Nah, what jgraham said
09:41
<hsivonen>
annevk: maybe. more likely if my domain experiences "DNSSEC suicide" for the third time...
09:41
<hsivonen>
annevk: DNSSEC is probably the fourth post in my queue of blog post ideas
09:42
<annevk>
hsivonen: ah shit, I just changed the Preface
09:43
<hsivonen>
(the three ahead of it are: "You should work on Firefox OS and self-hostable services if you want Software Freedom for phones", "You only need 3 or 4 ciphersuites" and "Character Encoding Menu in 2014")
09:45
<annevk>
those are all great, please take a week to write them :)
09:46
<hsivonen>
annevk: very productive to try to use a religious snowclone to end a bikeshed about wordsmithing non-normative text not to be offensive
09:47
Ms2ger
appears to be missing context on that one
09:47
<hsivonen>
Ms2ger: https://www.w3.org/Bugs/Public/show_bug.cgi?id=26693#c4
09:49
<Ms2ger>
Ta
09:51
<hsivonen>
annevk: the new text is good. gives less rationale than my suggestion, but the less you say, the less text there is for people to complain about
09:53
<annevk>
I'll see about integrating some of it, the extra rationale is actually nice
09:55
<hsivonen>
now that I'm disillusioned about DNSSEC, maybe I should ask my ISP to support IPv6
09:56
<annevk>
Did you see http://tack.io/ btw?
09:56
<annevk>
Although https://lists.riseup.net/www/arc/tack/2014-01/msg00001.html suggests it might be dead
09:57
<hsivonen>
annevk: I'm aware of TACK
10:00
<jgraham>
Wow, GitHub finally got side-by-side diffs
10:01
<annevk>
hsivonen: you think it might work for replacing CAs?
10:33
<othermaciej>
jgraham: thanks!
10:36
<hsivonen>
annevk: it seems to me that TACK still needs CAs for the first connection
10:52
<hsivonen>
annevk: if you are looking for encoding permathreads to debate, Thunderbird just improved its encoding UI but shows both windows-1252 and iso-8859-1 as distinct items: https://hg.mozilla.org/comm-central/rev/f2a455fd8e80
11:08
<JakeA>
Any indexeddb experts awake? If I have an index with two keys, say ['name', 'birthdate'], how do I open a cursor for a specific name but any birthday (but in birthdate order)
11:08
<JakeA>
?
11:13
<hsivonen>
it's been too long since I've paid attention to html5lib tests
11:15
<hsivonen>
please remind my what happened to the SVG filterRes attribute
11:15
<hsivonen>
html5lib explicitly tests for it *not* getting camelCased
11:20
<hsivonen>
oh. awesome. the html5lib test format has changed
11:20
<hsivonen>
that's why things are failing
11:21
<jgraham>
hsivonen: Blame Hixie
11:21
<hsivonen>
jgraham: for test format or filterRes?
11:22
<jgraham>
Test format
11:22
<jgraham>
But maybe both! Who knows :)
11:22
hsivonen
blames Hixie
11:29
hsivonen
is quite unamused by the #script-on / #script-off stuff
11:42
hsivonen
finds https://www.w3.org/Bugs/Public/show_bug.cgi?id=24900
11:51
<hsivonen>
oh. great. there are three other attributes that got un-camelCased
12:06
<caitp->
how about element names?
12:47
<Domenic>
I forget ... did anyone propose measuring DOMRectList.prototype.item usage to see if we can remove it in favor of Array?
12:50
<Ms2ger>
You did just now
12:51
<Domenic>
I don't want to re-propose it if someone already found a bunch of popular GItHub libraries using it...
12:54
<foolip>
Domenic: do you mean measuring ClientRectList.item in Blink?
13:04
<Domenic>
foolip: yeah
13:06
<foolip>
Domenic: I can make it so if the relevant spec editors want it. zcorpan?
13:07
<Ms2ger>
Make it sp
13:07
<Ms2ger>
so
13:08
<foolip>
Domenic: or are you already a Chromium/Blink committer?
13:08
<zcorpan>
foolip: want what? nuke ClientRectList?
13:08
<Domenic>
foolip: I am definitely not a committer, although I really should start committing some stuff sometimes soon :)
13:09
<jgraham>
It's a little known fact that in his high school yearbook foolip was voted "Person most likely to go on to measure web platform feature usage"
13:09
<foolip>
zcorpan: whatever it is one does to make it Array-like I guess.
13:10
<Ms2ger>
zcorpan, yeah, nuke
13:10
<Ms2ger>
If possible
13:10
<Domenic>
zcorpan: the proposal on the table is to use-counter ClientRectList.prototype.item and if it's not used then DOMRectList can die in favor of plain old Array.
13:10
<zcorpan>
foolip: http://dev.w3.org/fxtf/geometry/#DOMRectList already has [NoInterfaceObject, ArrayClass]
13:11
<foolip>
interesting
13:11
<zcorpan>
it should make it array-like as i understand it but item() is still there
13:11
<foolip>
so is ClientRect and ClientRectList a WebKit-only thing?
13:11
<Ms2ger>
foolip, they got renamed
13:11
<zcorpan>
no it was Client* in the spec earlier
13:11
<Ms2ger>
It would be nice to go arraylike -> array
13:12
<foolip>
best outcome is adding DOMRect and then failing to remove ClientRect :)
13:12
<zcorpan>
Ms2ger: sure
13:12
<Domenic>
foolip: oh god now i am afraid that will happen :(
13:13
<foolip>
:)
13:13
<jgraham>
s/best/most likely/
13:14
<Ms2ger>
Gecko's already killed ClientRect
13:15
<foolip>
if you want to talk to someone with an actual clue about where Blink is going with these interfaces, jinho.bang⊙sc and Rik Cabanier seem to be the relevant people
13:15
<zcorpan>
i think i tried to research usage of ClientRect/ClientRectList and, iirc, there was some item() usage but people didn't rely on the name of the interface object
13:15
foolip
wishes "Gecko doesn't have X" would automatically count as justification for removing X
13:16
<Ms2ger>
Make it so
13:18
<foolip>
Here are a few silly things in that category: Event.srcElement, Event.cancelBubble, Window.offscreenBuffering, Document.charset, Document.defaultCharset
13:18
<foolip>
offscreenBuffering is my favorite, it's always true and enjoys ~0.2% usage
13:19
<Ms2ger>
Nice
13:19
<Ms2ger>
Sadly, browser sniffing makes that argument invalid :/
13:19
<foolip>
Indeed
13:20
<Domenic>
we should all copy IE11 Mobile's user agent string
13:20
<Domenic>
Or, perhaps more realistically, all desktop browsers should have the same UA string.
13:21
<zcorpan>
Domenic: it is indeed a race to the bottom, except there is no bottom
13:21
<Ms2ger>
Except that the first browser to copy another UA string disappears from statistics
13:21
<Domenic>
Ms2ger: oh, I didn't think of that :(
13:22
<Ms2ger>
And there's an argument to be made that statistics are a legitimate use case
13:22
<Domenic>
yeah
13:23
<jgraham>
Presumably statistics are the intended use case
13:23
<jgraham>
Otherwise wtf were they thinking?
13:23
<foolip>
when the UA string can't be trusted you get something equally horrible: https://github.com/zynga/core/blob/cd2a5e639a603d4d327fea57fa3bb6c3b554d5d0/src/core/detect/Engine.js
13:24
<foolip>
sorry Gecko, you don't get to remove XULControllers, whatever that is
13:24
<foolip>
although this one did have a fallback that may work, I don't know
13:25
<jgraham>
That sounds like a thing we tried to remove and couldn't because of UA detection
13:25
<foolip>
I found it because of WebKitPoint, a recent removal I'm hoping won't get reverted because of nonsense like this
13:25
<Domenic>
IE mobile' strategy is a nice try though, I must say. I love how on my phone simultaneously a bunch of sites looked a lot better, and a bunch of sites started offering "install app from the Google Play Store" buttons.
13:30
<Ms2ger>
foolip, yeah, we tried to remove it, there's a shim in its place
14:29
<annevk>
hsivonen: baby steps, seems like
14:47
<zcorpan>
wait... standard markdown has processing instructions and CDATA blocks?
14:50
<jgraham>
?!
14:53
<jgraham>
Oh I guess "PIs" are for PHP
14:54
<zcorpan>
yeah. and CDATA for "XHTML" probably
14:54
<jgraham>
Because obviously using markdown to generate PHP to generate HTML is sensible
14:55
<jgraham>
Pretty sure that it should be a requirement that if you use either of those features the markdown processor will uninstall itself from your system, or refuse to grant you further access (if remotely hosted)
14:56
<caitp>
hah
15:20
<zcorpan>
Hixie_: looking at web-apps-tracker, maybe we should drop google-gears and add blink? or is "opera" code for blink now? :-)
15:25
<Ms2ger>
Heh, google gears
15:25
<ondras>
q
15:25
<ondras>
oops typo.
15:26
<zcorpan>
annevk: i don't follow how web-apps-tracker works :-(
16:08
<Hixie_>
hsivonen: i believe i tried to get your feedback on all those issues, fwiw :-)
16:11
<annevk>
zcorpan: what do you need guidance on?
17:18
Sample
thinks if anyone is trying to grab a filename from a file inputs value attribute their code -should- break/be inherently considered broken
17:33
<jgraham>
Sample: That's a fine thing to think, but if your bank is doing it, browsers change their behaviour, and you lose access to your account in a way that costs money, it feels like you aren't going to go "oh my bank had inherently broken code, so this is quite a reasonable outcome"
17:44
<Sample>
jgraham: yeah, unfortunately as well they had followed a specification at some point when they wrote that =/ sad
17:44
<Sample>
this must be the story of your guys' life
18:15
<caitp>
annevk, what if we had a way to do a better job of url resolution? like, a base content attribute for any elent, which if specified, would use a base tag with an appropriate id
18:16
<caitp>
(and by default, would use the first <base> tag in the document, or the usual url resolution procedure)
18:17
<caitp>
or maybe even automatically figuring out which base to use based on the element type or namespace or something
18:20
<caitp>
i'm just throwing ideas around, but it really sucks that you can only have one base that applies to all types of content :(
18:28
<annevk>
there was xml:base, but it was pretty terrible perf-wise and not really implemented well
18:28
<annevk>
in any event, you should start with use cases
18:31
<caitp>
well, the use cases are many, but there are some big ones --- SPA frameworks tend to give a second meaning to <base> tags, and unfortunately angular now requires it (the alternative was just too problematic for browsers that didn't support history), so people now have issues because they need to use one <base> for angular, and then have to specify all of their other content using fully qualified URIs
18:32
<caitp>
but I mean, I've had other use cases for it myself, like just generally needing to look for assets on a different domain while generally using the current domain and default base for everything else
18:32
<caitp>
fully qualified URIs kind of suck in case domain registration or other deployment details change
21:02
<mathiasbynens>
why is `<body id=b onload=b.innerHTML=1>` not equivalent to `<body onload=innerHTML=1>`? i’d expect both to set `document.body.innerHTML` to `1`
21:02
<mathiasbynens>
does the latter assign to `document.innerHTML` instead?
21:07
<Hixie_>
mathiasbynens: "this" in both cases is the global object (window)
21:07
<Hixie_>
mathiasbynens: so in the latter case you're doing window.innerHTML = 1
21:09
<mathiasbynens>
but for inline event attributes like that there are three ‘scopes’, right? the element itself, the parent form (if any), and the global object
21:11
<mathiasbynens>
for example, <p onmouseover=innerHTML=1>. does set the innerHTML of the `<p>` to `1` on hover
21:11
<Hixie_>
mathiasbynens: <body onload> (along with several other <body> event handlers) are special cases
21:11
<Hixie_>
mathiasbynens: in that they're not body's event handlers, they're window's event handlers
21:11
<Hixie_>
mathiasbynens: but since there's no <window> element, <body> acts as a convenience proxy for them
21:12
<Hixie_>
setting body.onload or <body onload> is equivalent to setting window.onload
21:12
<mathiasbynens>
ah, that explains it – thanks Hixie_!
21:12
<Hixie_>
note that this means you can't catch an event 'load' that bubbles through a <body> element with an onload handler
21:13
<Hixie_>
which leads to funny results when your markup is <html onload="t(event)"> <body onload="t(event)"> <div onload="t(event)"> and you fire a bubbling 'load' event at the div
21:13
<mathiasbynens>
hah
21:44
<Hixie_>
TabAtkins: is there a specific spec i should reference for display: newline? I'm update the <br> default stylings per our last conversation on the matter back in july
21:45
<Hixie_>
i guess i don't actually ref specs in the rendering section's style sheets actually
21:46
<Hixie_>
TabAtkins: hm, looks like 'display-box' has gone away
21:46
<Hixie_>
TabAtkins: what property should i use for br's newline behaviour?
21:49
<TabAtkins>
Hixie_: display-box changed name to box-suppress (but it's not what you want for newline)
21:49
<TabAtkins>
Hixie_: If you're going to assume it exists, it's a display-outside value.
21:50
<Hixie_>
k
21:50
<Hixie_>
do you want me to reassign the bug to you once i've done my side?
21:50
<TabAtkins>
Sure.
21:51
<Hixie_>
k, will do. Thanks. If things change, feel free to reassign to me again.
22:24
<Hixie_>
does nobody yet implement hit regions on canvas?
22:25
<Hixie_>
i see bugs suggesting people have implemented it...
22:26
Hixie_
wonders what he's doing wrong
22:26
<Hixie_>
cabanier: didn't you implement hit regions in mozilla at least? i can't get it to work
22:26
<Hixie_>
i found a bug where you checked code in, and that was months ago
22:27
<Hixie_>
more than a release cycle, at least
22:27
<cabanier>
Hixie_: you need to turn it on in about:config
22:27
<Hixie_>
aah
22:28
<Hixie_>
is it still buggy?
22:28
<cabanier>
canvas.hitregions
22:28
<cabanier>
Hixie_: yes.
22:28
<Hixie_>
ah, ok
22:28
<Hixie_>
cabanier: do you know anything about clearHitRegions() ?
22:28
<cabanier>
Hixie_: I need to finish it but keep getting pulled away.
22:29
<cabanier>
Hixie_: I didn't implement that one yet
22:29
<Hixie_>
cabanier: like, why it was proposed? i'm trying to work out if it's something i should spec. i can't work out what it's for and nobody seems to implement it.
22:29
<cabanier>
ah
22:30
<cabanier>
Hixie_: the hit regions are not removed by drawing over them so there needs to be a way to remove them from the canvas
22:30
<Hixie_>
oh this is for people who just draw opaque stuff and never clearRect() ?
22:31
<cabanier>
yes
22:31
<Hixie_>
and it just clears the entire list?
22:31
<cabanier>
Hixie_: a clearRect on the entire canvas should have the same effect
22:31
<cabanier>
yes
22:31
<Hixie_>
fair enough
22:32
<Hixie_>
seems a bit blunt but i guess why not if people don't need to call clearRect()
22:32
<cabanier>
yeah