01:24
<zewt>
gar ff9->ff13 lost half of my cookies
06:17
<smaug____>
what on earth is "suppress observers flag"
06:19
<smaug____>
and, fun. inserting multiple script elements using a document fragment isn't specified properly
06:22
<smaug____>
and browsers do it in a different way
06:22
<smaug____>
awasome
06:22
<smaug____>
awesome even
06:23
<smaug____>
Gecko, Presto and WebKit do all different things
06:30
<kennyluck>
awesome
06:53
<annevk>
smaug____: it isn't specified properly or it doesn't work properly?
06:57
<smaug____>
annevk: not spec
06:57
<smaug____>
ed
06:57
<smaug____>
just filing a bug
06:58
<annevk>
the suppress observers flag is needed
06:59
<annevk>
e.g. for records where you want to merge stuff
07:00
<annevk>
although we could perhaps in concept-node-pre-insert use concept-node-replace-all instead
07:01
<annevk>
that would avoid the usage of one instance
07:10
<Ms2ger>
annevk, tests? :)
07:10
<annevk>
hmm?
07:12
<smaug____>
annevk: "and there is" ?
07:12
<smaug____>
ahaa
07:12
<annevk>
right
07:13
<smaug____>
hard to follow...but spec are
07:13
<smaug____>
specs
07:13
<annevk>
feel free to rewrite it
07:14
<annevk>
I couldn't really find a better way
07:14
<smaug____>
I prefer just complaining :)
07:18
<Ms2ger>
Heh
07:20
<annevk>
smaug____: should the scripts execute after the mutationrecord is fired or before?
07:20
<annevk>
smaug____: also, seems like something HTML should define, not DOM
07:21
<smaug____>
s/fired/created/
07:22
<smaug____>
yeah, this is perhaps something for HTML spec
07:22
<smaug____>
annevk: currently the problem is that there is mutation record about the insertion. it contains all the nodes
07:23
<smaug____>
but some script may move those nodes before they are actually inserted to the parent
07:23
<smaug____>
(this is in webkit/presto)
07:24
<smaug____>
and all sorts of fun could happen...spinning the event loop during document fragment insertion
07:24
<annevk>
that's a known bug in WebKit
07:24
<annevk>
DocumentFragment insertion is not atomic
07:24
<annevk>
dunno about Presto actually
07:25
<smaug____>
what spec says it should be atomic?
07:25
<annevk>
well it's not observable from the DOM spec anymore because of the suppress observers flag
07:25
<smaug____>
(I hope I've missed something and that some spec actually *requires* Gecko's behavior )
07:26
<annevk>
but in HTML it should be clarified still I guess for <script>
07:26
<annevk>
or maybe I should add wording to DOM, but I'm not sure how to make that requirement
07:27
<smaug____>
annevk: how does suppress flag help here
07:27
<annevk>
otherwise you would get several removal records for instance
07:28
<annevk>
the DOM should maybe have an "inserted" hook for other specs
07:28
<smaug____>
but there is one record for the df insertion
07:28
<annevk>
yes because of that flag
07:28
<smaug____>
target parent
07:28
<annevk>
target parent?
07:28
<smaug____>
7 is If suppress observers flag is unset, queue a "childList" record with target parent, addedNodes nodes, removedNodes null, nextSibling child, and previousSibling child's previous sibling or parent's last child if child is null.
07:29
<smaug____>
then in 8 script may be executed
07:29
<smaug____>
and it can move some node, which would be in addedNodes
07:30
<smaug____>
if the script moves some node, where should it be after insert algorithm
07:30
<zcorpan_>
http://www.w3.org/mid/CA+c2ei99_Ov8avB-b3qN-ZCyWYJOo7-NMH-gAQQjBGVS_wciLw⊙mgc looks like some swedish auto-correct thing has messed up his email
07:30
<annevk>
smaug____: it's moved location?
07:31
<Ms2ger>
Heh
07:31
<smaug____>
annevk: to elsewhere in the DOM
07:31
<annevk>
I think we should add a step 9 and equivalent steps elsewhere that provide HTML with a hook for "the element(s) have been inserted, do something"
07:32
<annevk>
that would make it explicit when certain actions happen
07:32
<smaug____>
annevk: and step 9 would actually execute the scripts?
07:32
<smaug____>
(that would be spec'ed in HTML)
07:32
<annevk>
yes
07:33
<smaug____>
sounds good to me
07:34
<annevk>
might need that for removal too I guess (e.g. <iframe> removal)
07:45
<zcorpan_>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=17529 seems like fun
07:48
<Von_Davidicus>
Heyo.
07:51
<zcorpan_>
hey
07:56
Von_Davidicus
is dusting off old skills that may actually be of use, unlike others I've been exercising.
07:58
<Von_Davidicus>
Here's a question: Which would actually be faster for rendering and lower on bandwidth? PHP+SQL or XML+XSLT?
07:59
<zcorpan_>
i guess the correct answer to that is "it depends"
08:00
<annevk>
kind of depends on what you end up transmitting over the wire...
08:01
<zcorpan_>
for instance, if the input data is 1GB and the transformed output is 1KB, and you cache the result on the server in the PHP+SQL case...
08:02
<Von_Davidicus>
One thing I liked about XML + XSLT was that the XML files were more "compact", and I'd read somewhere that XSLT files were cached. (That, and the DB I was using at the time was being hacked to fish sticks and fries)
08:03
<hsivonen>
does anyone know what H.264 video and MP3 audio in MP4 container is in terms of RFC 4281?
08:04
<zcorpan_>
XSLT isn't rendered incrementally, so if you care about low-bandwidth users, doing the transformation on the server would be better (assuming the server isn't slow with doing it and/or you cache the result)
08:04
<zcorpan_>
hsivonen: iirc MP3 doesn't have a codecs="" name
08:05
<hsivonen>
zcorpan_: nice trick for avoiding format proliferation
08:05
<zcorpan_>
but i recall seeing audio/mpeg; codecs="mp3" somewhere
08:05
<hsivonen>
zcorpan_: do happen to know if IE and Chrome support MP3 in MP4 container?
08:05
<zcorpan_>
dunno if anyone implements that name, though
08:05
<zcorpan_>
no idea
08:06
<hsivonen>
If they do, I hope Facebook never uses it
08:06
<Von_Davidicus>
I thought that XSLT was kind of like an image or CSS file--your computer stored it for reuse.
08:06
<Von_Davidicus>
Or does "caching" mean something else?
08:07
<zcorpan_>
Von_Davidicus: it can be cached, but that doesn't help with incremental rendering, since the browser needs the full input data before starting the transformation
08:08
<zcorpan_>
Von_Davidicus: it also doens't help with the user's first load, which is a pretty common case
08:09
<zcorpan_>
(but in the first load case, the data file and the xslt file could be downloaded in parallel, so it might not make much of a difference)
08:10
<Von_Davidicus>
Yeah, that would be the XML file. *Looks at XSLT stylesheet.* My biggest is 9.73 kilobytes. I've made stylesheets bigger than that.
08:11
<hsivonen>
whoa. AMR and H.263 have remarkably reasonable codecs="" names for codecs whose names are defined in the RFC
08:13
<Von_Davidicus>
And besides: this was for a webcomic--the XSLT was kinda dwarfed by the images. :)
08:13
<zcorpan_>
hsivonen: heh, http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1589 in chrome says 'probably'
08:14
<zcorpan_>
Von_Davidicus: when can the browser know to fetch the image? before or after the transform?
08:15
<zcorpan_>
(opera and firefox seem to have type-specific codecs="" instead of global which chrome seems to have)
08:16
<Von_Davidicus>
After, I'd say, since the <img> element is set via XSLT.
08:17
<zcorpan_>
hsivonen: codecs=mp3 doesn't say 'probably' in safari or chrome, so it seems that codec name isn't a supported one
08:18
<zcorpan_>
Von_Davidicus: that's bad for perf
08:20
<Von_Davidicus>
perf?
08:22
<zcorpan_>
performance
08:25
<Von_Davidicus>
Well, it got the file name from the XML file, but that's about it. I dunno why it would be any worse than using PHP to create HTML: the browser still has to see the IMG tag to get the image.
08:35
<zcorpan_>
the browser can see the img earlier in the HTML case
08:36
<Von_Davidicus>
Ah. Okay/.
08:39
<hsivonen>
zcorpan_: codecs in Chrome are per-type somehow: http://hsivonen.iki.fi/test/moz/canplaytype.html
08:39
<hsivonen>
zcorpan_: note audio/webm; codecs="vp8"
08:41
<hsivonen>
zcorpan_: video/webm and audio/webm are not like the other type in Chrome
08:44
<hsivonen>
zcorpan_: for supported non-WebM containers, Chrome says "maybe" for bogus codecs
08:44
<hsivonen>
which is not nice
08:47
<zcorpan_>
indeed
08:47
<zcorpan_>
my copy of safari also does that
08:48
<zcorpan_>
shouldn't audio/webm; codecs="vp8" say 'probably'?
08:48
<zcorpan_>
does in opera/firefox
08:49
<hsivonen>
zcorpan_: apparently they don't like having a video codec in <audio>
08:50
<zcorpan_>
audio/ doesn't mean it'll go in <audio>
08:52
<hsivonen>
oh, right
08:53
<hsivonen>
in IE, audio/mpeg is maybe and audio/mpeg; codecs="mp3" is probably
08:54
<zcorpan_>
interesting
08:55
<hsivonen>
do people deploy H.264/MP3/MP4 to Flash Player?
08:56
<hsivonen>
I wonder if Shumway will end up needing that combo
09:07
<Lachy>
annevk, what do I need to do to update selectors api 2 to use the new style exceptions you mentioned on the list?
09:09
<Ms2ger>
s/SYNTAX_ERR/SyntaxError/g
09:09
<Lachy>
what's the difference?
09:09
<annevk>
I think selectors api 1 should be updated too
09:10
<annevk>
or just dropped in favor of 2 :)
09:10
<annevk>
Lachy: it sets the err.name attribute appropriately
09:11
<odinho>
Lachy: And new ones now don't have any numeric "code".
09:12
<odinho>
Lachy: So it's basically not an numeric enum any more.
09:13
<Lachy>
is e.code retained for backwards compatibility with legacy exceptions?
09:13
<Lachy>
Interface DOMError seems to omit the code property entirely.
09:14
<odinho>
Lachy: Yeah. You have e.code on legacy DOMExceptions. And it is still on new ones, only always set to 0 there.
09:17
<Lachy>
ok, but shouldn't the DOMError interface at least define the code attribute or state that it implements the DOMException interface or something? As it is right now, the domcore spec seems to indicate that there is no code property for a DOMError at all.
09:17
<Ms2ger>
annevk, let poor ArtB have his rec ;)
09:17
<Ms2ger>
Lachy, DOMError doesn't have anything to do with exceptions
09:21
<Lachy>
then I don't understand. When a syntax error occurs in selectors API, there object that gets thrown has both a .name and .code property. If I define that it raises a SyntaxError, then by that definition, it shouldn't have a code property.
09:21
<Lachy>
What am I missing?
09:22
<Ms2ger>
Why not?
09:22
<Ms2ger>
http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#exception-domexception
09:22
<Ms2ger>
exception DOMException { ... unsigned short code;
09:23
<odinho>
Lachy: SyntaxError is DOMException, not DOMError.
09:23
<Lachy>
But S
09:23
<odinho>
Lachy: All the exceptions are named *Error.
09:23
<Lachy>
"SyntaxError" is in the table under the DOMError section.
09:23
<Lachy>
and DOMException doesn't have a name property.
09:24
<Ms2ger>
It does
09:24
<Ms2ger>
See the example in the DOMException section
09:24
<Lachy>
where?
09:24
<Ms2ger>
(It's well hidden)
09:24
<Lachy>
See the IDL.
09:24
<Ms2ger>
See WebIDL
09:25
<zcorpan_>
Lachy: the table is under "Error types", not in "Interface DOMError"
09:25
<zcorpan_>
Lachy: saying "throw a SyntaxError exception" means http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#concept-throw
09:26
<Lachy>
"In JavaScript, this exception will have a name property whose value is "TimeoutError"" - why is that name property completely omitted from the IDL?
09:26
<Ms2ger>
It's not on the prototype
09:26
<annevk>
IDL lists attributes, not properties
09:27
<Ms2ger>
Well, "fields", in the case of exceptions
09:27
<annevk>
ah yeah
09:27
<odinho>
I agree with Lachy that's it's mightly confusing though. "type" means "name" by the way. That was what confused me very much before.
09:28
<odinho>
Lachy: So when they say "set its *type* to /type/" then it means set it's e.name to /type/. Because A type DOMError means a DOMError object whose name attribute is initialized to type.
09:28
<zcorpan_>
Lachy: see step 5 of the second algo in http://dev.w3.org/2006/webapi/WebIDL/#es-throwing-exceptions
09:28
<Lachy>
so then where is it defined that, in javascript, the name property gets set, if it's only mentioned in that one little non-normative example?
09:31
<annevk>
Lachy: step 1 of concept-throw
09:32
<Lachy>
oh, ok. Very confusing.
09:32
<annevk>
if you'd just read it...
09:34
<Lachy>
You could at least put some notes in there referring to WebIDL and mention the name property for JavaScript.
09:35
<odinho>
Lachy: Ms2ger put in this example because I got confused by all the "type" talk --> To throw a "TimeoutError" exception, a user agent would construct a DOMException exception whose type is "TimeoutError" and code exception field value is 23, and actually throw that object as an exception. In JavaScript, this exception will have a name property whose value is "TimeoutError".
09:35
<Lachy>
I did read it. I just didn't make the necessary links to put it all together
09:36
<odinho>
It's easy to read when you know what it's supposed to say :]
09:36
<odinho>
Harder otherwise.
09:36
<hsivonen>
where do I find a list that says what codecs each gst-plugins-* package contains?
09:36
<Lachy>
I mean, clarify that in step 1 about type, to make the link between type and name easier. That example says nothing to clarify where the name property gets set.
09:37
<annevk>
maybe Web IDL should rename type to name
09:38
<zcorpan_>
wfm
09:38
<hsivonen>
http://gstreamer.freedesktop.org/documentation/plugins.html
09:47
<Lachy>
selectors api 2 has now been updated for this.
09:50
<annevk>
Lachy: in Selectors 2 some instances of Element in the IDL should be Element?
09:50
<Lachy>
?
09:51
<annevk>
Lachy: also making '(Element or sequence<Node>)' nullable if it's already optional seems unnecessary
09:51
<annevk>
Lachy: as in "Element?" because they can be null
09:53
<Lachy>
it has to be nullable to handle the case where authors use libraries that just pass whatever value some variable is set to, rather than first checking if its null and omitting it.
09:54
<Lachy>
i.e. document.find(selector, ref) where the ref variable happens to be set to null.
09:55
<annevk>
yeah I think there's a bug on allowing optional arguments to be null already
09:55
<Lachy>
So arey ou saying I should change this "Element find(DOMString selectors, optional (Element or sequence<Node>)? refNodes);" to "Element find(DOMString selectors, optional (Element or sequence<Node>) refNodes);"
09:56
<annevk>
I guess I'm mostly saying to make it Element? find...
09:56
<annevk>
maybe leave that other stuff for now
09:57
<Lachy>
oh, for when the function returns null? I probably should.
09:57
Von_Davidicus
ponders.
09:57
<annevk>
jgraham: does lkpg have spare computers?
09:58
<Von_Davidicus>
Whether 'tis better to simply go the SQL+PHP = HTML route or SQL+PHP = XML+XSLT+XSL+DTD (since I have those files lying about...)
09:58
<annevk>
jgraham: mah never mind, guess I can carry this T60 thing around one last time :)
09:58
<jgraham>
annevk: I have a G4 Mac with broken network under my desk :p
09:58
<annevk>
ooh :)
09:59
<odinho>
annevk: Not happy with your laptop any more?
09:59
<jgraham>
I'm sure we can find you a computer if you need, but it is probably easier if you bring the laptop
10:00
<jgraham>
When do you arrive?
10:00
<Lachy>
done
10:01
<annevk>
odinho: MacBook broke down Friday/Saturday
10:01
<odinho>
annevk: Typical.
10:02
<annevk>
odinho: so now I'm using my somewhat ancient T60
10:02
<annevk>
jgraham: tomorrow afternoon, so I guess I actually only work Wed-Fri
10:02
<odinho>
Well, my main computer at home is still an ancient X60... Though I was cursing *A LOT* yesterday.
10:02
<jgraham>
Isn't Friday a puiblic holiday here?
10:03
<annevk>
teehee less work
10:11
<wodmal>
any opera devs around? annevk?
10:12
<annevk>
?
10:12
<wodmal>
annevk: where can i find some docs on the opera:crossnetworkwarning feature and what it gets triggered by?
10:12
Ms2ger
sees four talking in the last 15 minutes
10:20
<jgraham>
Ms2ger: Technically QA or whatever it is that annevk does. "IETF botherer" perhaps.
10:20
<jgraham>
wodmal: I am not aware of any documentation
10:20
<Ms2ger>
jgraham, what would that make you? :)
10:21
<jgraham>
Maybe I meant s/or/and/
10:21
<shwetank>
wodmal if you want, i (or some other opera employee here) can connect you with someone who does know about it though on email
10:24
<hsivonen>
zcorpan: In case Opera feels forced to support MP3, too, you might be interested in https://bugzilla.mozilla.org/show_bug.cgi?id=760140#c7
10:25
<doublec>
Or use the javascript decoders that seem to be popping up (mp3 and flac)
10:26
<hsivonen>
doublec: I'm kinda worried that the same bug will show up on Android and B2G if we don't get unit tests for this stuff
10:27
<Sigbjorn>
I was told there were questions regarding Opera's cross network policy?
10:27
<zcorpan>
hsivonen: thanks
10:28
<Ms2ger>
Why don't we rename this channel #opera? :)
10:28
<jgraham>
wodmal: Sigbjorn should be able to help you
10:28
<jgraham>
Ms2ger: Might invite discussion of Wagner
10:28
<Ms2ger>
"What, why can't I apply logic to Wagner?"
10:28
<zcorpan>
Ms2ger: what are you doing here anyway?
10:29
<Ms2ger>
Complaining about Opera bugs
10:29
<zcorpan>
ah
10:32
<doublec>
hsivonen: same bug as what?
10:33
<hsivonen>
doublec: same as https://bugzilla.mozilla.org/show_bug.cgi?id=760140 i.e. canPlayType getting out of sync with playback capability with external codecs
10:34
<doublec>
hsivonen: at the moment we restrict on android and b2g to specific mime typea
10:34
<hsivonen>
doublec: gst did, too, but canPlayType still got out of sync
10:34
<doublec>
hsivonen: although they do actually work to play other types if you serve video/mp4 for example since libstagefright ignores it and sniffs
10:34
<hsivonen>
doublec: :-(
10:35
<hsivonen>
doublec: can't we sniff for file headers and reject stuff we don't want to expose to the Web?
10:35
<doublec>
hsivonen: I'm sure we can do something. I'll raise a bug for it.
10:35
<hsivonen>
doublec: thanks
10:36
<hsivonen>
even Apple restricts codecs for security reasons
10:36
<jgraham>
So… if I have an empty script element (in a document) and append a document fragment to it containing some text nodes representing script followed by a second script that itself contains script content, which order should the two scripts execute in?
10:36
<hsivonen>
(but even then, they support too many codecs, IMO)
10:36
<jgraham>
s/script content/text content representing script/
10:37
<hsivonen>
I'd prefer Apple not to support MPEG-4 Visual Simple Profile, video/quicktime and video/3gpp
10:37
<hsivonen>
those are pure format proliferation
10:40
<hsivonen>
I wonder if MS likes it that the FSF considers bundling the VC++ runtime redistributable GPL-incompatible
10:40
<jgraham>
In fact is the order of inserting multiple scripts simultaneously well defined at all?
10:41
<hsivonen>
jgraham: unclear. I meant to reply to sicking, but I got distracted by this canPlayType review
10:41
<annevk>
jgraham: I'm "Developer" if we're talking intranet technically
10:42
<annevk>
also, good news, shuffling around with the RAM repaired my MacBook
10:42
<jgraham>
hsivonen: I didn't have any context in mind that might involve sicking… is that the mutation observers thread?
10:42
<annevk>
or not
10:42
<hsivonen>
jgraham: no, the document.parse thread
10:44
<jgraham>
Oh right, the part-Swedish mail
10:45
<jgraham>
hsivonen: So you can already do that by constructing an explicit DocumentFragment containing multiple scripts, right?
10:45
<hsivonen>
jgraham: yes
10:48
<jgraham>
(I agree this is something of a corner case currently, but much less so in the document.parse world)
11:18
<wodmal>
Ah - hi Sigbjorn... :)
11:18
<Sigbjorn>
o/
11:19
<wodmal>
What exactly is it that triggers opera:crossnetworkwarning?
11:19
<Sigbjorn>
automatic attempts to access an innermore network
11:20
<Sigbjorn>
e.g. an internet page attempting to iframe an intranet page
11:20
<Sigbjorn>
"automatic", as opposed to "manual click"
11:21
<Sigbjorn>
The rationale being that a lot of intranet (or localhost) pages are protected by a same-network-only protection, and we don't want to let the browser be an unwitting proxy for attacks on such pages
11:21
<wodmal>
Sigbjorn: why would it have been triggered by gmail, then?
11:21
Philip`
annoyingly gets the crossnetworkwarning whenever connecting via BT FON (free-to-customers wireless thing in the UK), plus certificate warnings, when it does the redirect to the "please enter username and password" page after you first connect
11:22
<Philip`>
(Firefox doesn't seem to give any warnings at all)
11:22
<Sigbjorn>
wodmal: no idea, I guess gmail, or an email in gmail, had an iframe/image/resource pointing to an intranet resource
11:23
<wodmal>
Sigbjorn: meaning 192.168.x.x or 10.x.x.x?
11:23
<Sigbjorn>
Philip`: some wifi proxies do mess with intranet/internet settings - there is an override on the page itself, and a global switch to turn the feature off
11:23
<Sigbjorn>
Philip`: Opera is the only browser with this protection
11:24
<Sigbjorn>
wodmal: yes, there are 4 IPv4 local ranges, including localhost, iirc
11:25
<Sigbjorn>
Philip`: the reason being that you are trying to access an internet page, e.g. gmail.com, and that page automatically redirects you to an intranet page - that is indistinguishable from an attack
11:25
<Sigbjorn>
that page might have been a hidden iframe inside another page, attempting to abuse a local server of yours
11:25
<wodmal>
or 127.0.0.1....? <---this would make more sense to me as something that might have for w/e reason been referenced by gmail. Although I do have this weird wifi-proxy sort-of-thing box which i don't really know too much about that my friend is developing... so that could explain it. afaik(new) though, it was fully transparent.
11:26
<Sigbjorn>
wodmal: there are three networks, localhost, intranet, internet - access from inside out is allowed, but not the other way around
11:26
<wodmal>
Sigbjorn: but why is the english-prose prompt on that warning page something to the effect of "this site is trying to access data which is stored __on your computer__" then?
11:26
<wodmal>
Sigbjorn: ah, that makes sense.
11:26
<Sigbjorn>
"your computer" means "localhost"
11:28
<Sigbjorn>
Philip`: the reason for the certificate warnings are that you are attempting to load a secure page, and your ISP is hijacking the connection, serving you another page under that domain instead (presumably just a redirect page, so you will never actually see it)
11:28
<Sigbjorn>
though instead of it being a redirect, it might as well have been a full fledged page, pretending to be your email provider/bank/etc
11:31
<wodmal>
Sigbjorn: what's the "attempting to access data on your comp" part all about?
11:32
<Philip`>
Sigbjorn: Out of interest, is there a way the ISP could implement its wifi proxy stuff so that it's not indistinguishable from real attacks and doesn't trigger warnings?
11:32
<Sigbjorn>
E.g. after proxy rewriting, gmail will now have a request for an image from 127.0.0.1/img/404.png
11:32
<Sigbjorn>
that would qualify as "Gmail is trying to access data from your comp"
11:33
<Sigbjorn>
Of course, much worse, was if gmail loaded an iframe as the result of a POST
11:33
<wodmal>
Sigbjorn: what type of "proxy rewriting"?
11:34
<Sigbjorn>
wodmal: you tell me, you said you had a wifi-proxy sort-of-thing
11:34
<Sigbjorn>
Philip`: for http, no problem, just redirect. For the internet->intranet warning, put their signup page on an internet page. For https, not possible, by the very design of https
11:35
Philip`
likes the FON thing since you can sign up for a cheap internet connection with a 10GB/month usage limit, and then connect to your own wireless router over the FON network and get lower bandwidth (clamped to ~3Mbit/s) but it doesn't count towards your usage limit so you can download as much as you want
11:35
<Sigbjorn>
if you explicitly say you want https, your browser will complain unless the returned value is signed by the proper certificate - which you ISP cannot do
11:36
<Sigbjorn>
moving discussion to #opera btw
12:00
<annevk>
back on my MacBook, with 4GiB RAM less
12:00
<annevk>
the logic board is somewhat broken most likely, and because I dropped it a few times with visible damage, the way expensive Apple Care does not care
12:11
<zewt>
oh god please tell me you aren't one of those people who unironically types "gib"
12:18
<Ms2ger>
GiB*
12:18
<zewt>
words are not case-sensitive
12:20
<jgraham>
Not only is that untrue, it's irrelevant since GiB isn't a word.
12:20
<jgraham>
(I'm pretty sure there are cases where distinguishing proper nouns makes a difference)
12:22
<Philip`>
zewt: Case is quite important if you think you're getting a gigabyte of network traffic but it's actually a gigabit
12:22
<zewt>
not in any serious way; the worst is people who try to pretend "kb" vs. "kB" is a serious way to distinguish units
12:23
<Philip`>
(Even worse if it's actually 10^9 bits)
12:23
<zewt>
(if it needs disambiguation, you say kbyte or kbit; if you assume "kb" is in bits, then you're going to be wrong more often than not)
12:24
<jgraham>
Using case to distinguish units is rather common; think mJ vs MJ
12:25
<AryehGregor>
I was trying to think of an example that would come up in everyday life. I guess mJ vs. MJ is the most plausible.
12:26
<Philip`>
Think of data measured in LOC (lines of code) vs LoC (Libraries of Congress)
12:26
<zewt>
(the real objection is to a committee thinking they can retroactively modify the language, and tell us "gb doesn't mean 2^30 anymore, use this ludicrous-sounding word we invented instead"; sorry, no)
12:27
<zewt>
so I cringe at the rare times somebody actually does that, heh
12:27
<Philip`>
(Yeah, who wants standards committees that try to fix ambiguities)
12:27
<AryehGregor>
Haven't hard disk manufacturers been successfully sued for advertising capacities in powers of ten instead of two?
12:28
<zewt>
never heard of that
12:28
<AryehGregor>
http://www.crn.com/news/channel-programs/189602434/western-digital-settles-hard-drive-capacity-lawsuit.htm;jsessionid=H2bWDbyY+895tVjM4e54vQ**.ecappj03
12:28
<AryehGregor>
That says it was settled, mind.
12:28
<zewt>
i'd take it as a small victory against misleading advertising if so :)
12:28
<Philip`>
They always seem to add footnotes "* 1 GB = 1,000,000,000 bytes" on their packaging and marketing materials, I think
12:28
<annevk>
zewt: what's up with your deity?
12:28
<Philip`>
At least that's less misleading than measuring hard disk capacity in numbers of MP3
12:28
<zewt>
haha
12:29
<zewt>
what about advertising bandwidth in number of modems
12:29
<annevk>
heh
12:29
<jgraham>
Isn't bandwidth measured in copyright infringments / second?
12:30
<zewt>
if you're sending haikus or something you could inflate that number pretty effectively
12:30
<annevk>
zewt: kb might be bytes, but kb/s typically isn't
12:30
<AryehGregor>
That's not a useful type of measurement, because you can commit copyright infringement by copying like five notes in some cases.
12:30
AryehGregor
usually sees "kbps"
12:31
<Philip`>
To pick a completely arbitrary example, http://www.wdc.com/en/products/products.aspx?id=260#tab3 still has a table with "8 TB / songs: up to 2,000,000" etc
12:32
<Philip`>
Surely someone in the company must have realised that it's no longer a useful metric
12:32
<AryehGregor>
Is 4 KB a reasonable amount for a song?
12:32
<zewt>
annevk: it says kbytes to me--"kbit" or occasionally kbps for bits
12:32
<Philip`>
AryehGregor: No, but they're saying it's 4MB each
12:32
<zewt>
Philip`: what about MIDI?
12:32
<AryehGregor>
Oh, I see.
12:32
<Philip`>
AryehGregor: (4 minutes at 128kb/s)
12:33
<annevk>
anyway, given this long discussion I'll make sure to use GiB going forward for maximum #whatwg productivity
12:33
<zewt>
it's 7:30am here, waking up takes priority over productivity :)
12:34
<annevk>
I love me some gibibytes
12:34
<zewt>
i'm still waiting to see somebody say that out loud with a straight face
12:34
<Philip`>
zewt: MIDI is too hard to pirate, since you can't rip it from CDs, so it's not a use case for ginormous hard disks
12:35
<Philip`>
(Even when they're reporting capacities of 40,000 songs, presumably they don't really expect their customers to have bought every one of them)
12:35
<Ms2ger>
Hmm
12:36
<zcorpan>
annevk: now you need to get "gibibytes" in your next talk and make sure you have a straight face
12:36
<zewt>
zcorpan: in that situation it'll see whether the viewers keep a straight face
12:36
<annevk>
it will be my new life goal
12:37
<Philip`>
Pronounce it "guy buy bytes" for extra fun
12:37
<zewt>
they should have gone with "gigglybytes"
12:37
<zewt>
i'd say that
12:38
<annevk>
I like how Wikipedia says gigabyte is only used in some contexts to mean gibibyte
12:38
<zewt>
(straight face not guaranteed)
12:39
Philip`
reads the WD disk page more and sees "As used for storage capacity, one megabyte (MB) = one million bytes [...] As used for buffer or cache, one megabyte (MB) = 1,048,576 bytes", so they're not even being consistent within a single product
12:41
<zewt>
next thing, display manufacturers will market their displays in the size the pixels would be if you laid them in a straight line
12:41
<zewt>
try our new one mile tv
12:42
<zcorpan>
zewt: that'd be a pretty sucky tv
12:42
<zcorpan>
or small
12:43
<zcorpan>
or both
12:45
<zewt>
my TV is 60" wide with 1080 rows of pixels = 64800 inches = 1.02 miles, unless my early morning math is broken
12:47
<zcorpan>
ah, you're right. seems my math was broken, probably somewhere in my pointless conversion to ISO units
12:48
<Philip`>
Even counting pixels is hard now that people are doing weird subpixel tricks (like on PenTile displays) :-(
12:48
<zcorpan>
except, isn't your TV 60" in diameter?
12:49
<hasather>
zcorpan: zewt: most likely diagonally
12:49
<zcorpan>
hasather: uh, that's what i meant, yeah
12:49
<zewt>
no, across
12:49
<zewt>
(sold as a '65"' tv)
12:50
<hasather>
I see
12:50
zcorpan
hasn't seen 65" in ads
12:50
<Philip`>
I wish we could switch from square pixels to hexagonal pixels
12:51
<zewt>
(actually that's including the bezel, which is about 4")
12:54
<annevk>
that's a big tv
12:54
<jgraham>
Bigger (in width) that quite a few people (in height)
12:56
<zcorpan>
jgraham: maybe that's how they'll market TVs next
12:57
<jgraham>
"If this falls on you, not even your feet will stick out"?
12:57
<zewt>
jgraham: infants, maybe
13:57
<AryehGregor>
Shepherd needs: 1) E-mail notification. 2) A way to comment and respond on multiple files at once.
14:05
<annevk>
Ms2ger: yo yo
14:05
<annevk>
Ms2ger: so something coming by about that XHR constructor dictionary
14:05
<annevk>
s/so/saw/ doh
14:06
<Ms2ger>
Ah, sicking filed it, good
14:06
<annevk>
Ms2ger: the problem is that you restrict what we can do with it in web context the moment you support that
14:07
<Ms2ger>
We don't expose anything to content atm
14:07
<annevk>
kk
14:07
<annevk>
not sure it's going to be a problem either way
14:08
<Ms2ger>
Also, it's prefixed
14:09
<annevk>
the constructor?
14:09
<Ms2ger>
No, the properties
14:10
<annevk>
yeah, but that does not necessarily help
14:10
<annevk>
if we e.g. change parts of the model based on that argument being present
14:11
<Ms2ger>
What I'm trying to say is that you don't need to worry about us :)
14:11
<annevk>
i'll say kk again then
14:13
<annevk>
sweet, document.documentURI is readonly in WebKit too now
14:14
Ms2ger
goes back to fixing xhr.responseType
14:57
gsnedders
fixes Ms2ger
15:00
<Ms2ger>
Do I need to reboot?
15:01
<jgraham>
Ms2ger: I think you misspelt "I'm sorry I can't let you do that, Dave"
15:44
<hsivonen>
I think I found a serious bug in IE10. How do I report it so that 1) I don't have to click through legalese that would be unwise to click through for someone who develops a competing products and 2) someone actually pays attention?
15:47
<jgraham>
Make friends with someone on the IE team?
15:47
<Ms2ger>
Push them into a lake?
15:47
<Philip`>
I like it when web pages say "SQL injection or Scripting Attack Found! Loading Halted!" when you perfectly innocently fill in a form that might lead to you giving lots of money to the people running the site
15:47
<Ms2ger>
Or is that annevk's privilege...
15:48
Philip`
wonders if it provides any real value in stopping actual attacks, to balance out how it drives potential customers away
15:48
<annevk>
Ms2ger: it's Opera's
15:48
<annevk>
Ms2ger: because Opera did it first
15:48
<Ms2ger>
Well played
15:49
<Ms2ger>
Hey, you
15:49
<Ms2ger>
CSSOM sucks
16:02
<hsivonen>
The IE bug I found is so bad that ACPI shutdown doesn't respond. Task Manager window opened and then doesn't respond.
16:02
<hsivonen>
Surely the scheduler in Windows 8 shouldn't let IE eat up resources this crazily
16:02
<hsivonen>
now over a minute late, ACPI shutdown was honored
16:03
<hsivonen>
I ended up DMing an IE team member on Twitter
17:29
<annevk>
Ms2ger: well yeah...
17:29
<annevk>
Ms2ger: want to fix it?
17:29
<Ms2ger>
Me, fix something?
17:29
<annevk>
Ms2ger: don't think anyone else has yet, at least last time I checked CSSOM View no edits were made
17:29
<annevk>
Ms2ger: good point
17:29
<Ms2ger>
Oh, CSSOM
17:30
<Ms2ger>
I might end up doing it, I guess
17:30
<Ms2ger>
Otoh, CSSWG
17:30
<annevk>
but you're a rogue entity
17:31
<Ms2ger>
I'll have to think if I should feel offended by that
17:46
<jwalden>
gsnedders: fwiw, I got pointed at the __proto__ disaster that is https://mail.mozilla.org/pipermail/es-discuss/2012-May/022834.html and will likely send mail about which of data/accessor/[[Get]]-modification I'd prefer when implementing, soon
17:46
<jwalden>
trying to shortcut on whether there's been any followup on that since then before throwing in my two cents, tho
17:49
<annevk>
Ms2ger: it's a feature, but please do set up some kind of kickstarter project for extra RAM so you can reply faster and work on the CSSOM
18:00
<gsnedders>
jwalden: Which would you rather?
18:00
Ms2ger
eats strawberries instead
18:00
<gsnedders>
(I don't like the current accessor proposal as that means having context for each object)
18:00
<jwalden>
gsnedders: well, nothing; but assuming arguendo something must be in place, I think pretty clearly an accessor
18:01
<jwalden>
magic properties impose far more cost on anything at all that looks up properties
18:01
<jwalden>
accessor shoves all the complexity into one narrow method
18:01
<jwalden>
that can be as awful as needed, but at least it contains the contagion
18:02
<jwalden>
whereas now, __proto__ pervades many areas
18:02
<gsnedders>
jwalden: ^^that.
18:02
<gsnedders>
A thousand times that.
18:02
gsnedders
is basically sitting on an accessor impl waiting for this whole, endless debate to be hashed out
18:04
<gsnedders>
If we're so worried about the setter function, just poision Object.getPropertyDescriptor.
18:04
<Ms2ger>
gsnedders, ship it
18:04
<Ms2ger>
You're at Opera, you're supposed to do it first
18:04
<gsnedders>
There, we still have a clean impl, and the security risk of having the function is gone.
18:05
<gsnedders>
Ms2ger: Okay, the only reason why it wasn't shipped in 12 was because I had to back it out finding a horrible bug I'd introduced :P
18:06
<gsnedders>
jwalden: You'll have to address the SES concerns, though, given that's where all the arguments over __proto__ go around… :\
18:06
<jwalden>
gsnedders: was making the accessor property configurable, ergo deletable, not enough?
18:07
jwalden
doesn't have time to read es-discuss regularly enough to follow all this stuff :-\
18:07
Ms2ger
wonders what SES is
18:07
<jwalden>
which unfortunately tends to put me in damage-control mode more often than not in reaction
18:07
<Ms2ger>
A Scandinavian airline?
18:07
<jwalden>
secure ES
18:07
<jwalden>
think caja sandboxlike stuff
18:07
<Ms2ger>
"secure"? I guess it must be good, then
18:07
<jwalden>
I don't know what __proto__ specifically entails for the idea in its latest iteration
18:08
<jwalden>
well, the idea is you close off bad functionality first thing, then you have something that's contained in what it can do
18:08
<jwalden>
it's crazy
18:08
<jwalden>
but it actually is kind of workable
18:08
<gsnedders>
jwalden: var f = document.createElement("iframe"); var s = Object.getPropertyDescriptor(f.contextWindow.Object.prototype.__proto__).set;
18:09
<gsnedders>
jwalden: If that function can mutate __proto__ in the local global, then deleting it gained nothing.
18:09
<gsnedders>
I don't like the proposal to make the function check what context it's dealing with, as that means having context for all objects.
18:10
<Ms2ger>
Shrug, we have that ;)
18:10
<jwalden>
gsnedders: wait, isn't allowing untrusted code to create a new window that's not secured kind of wrong already?
18:10
<jwalden>
gsnedders: I don't think you need context to make that work, actually; you just need a way to track from an object back to the global object where it was created
18:10
<gsnedders>
jwalden: I believe the theory is that it can't mutate anything in the current window, and hence safe
18:11
<gsnedders>
jwalden: Well, yes, but we don't have that :)
18:11
<jwalden>
gsnedders: how do you make stuff like new otherWindow.Array() return the right thing?
18:12
<gsnedders>
jwalden: In terms of what its prototype is?
18:13
<gsnedders>
jwalden: Functions store that; most objects don't.
18:13
<jwalden>
oh, hm
18:13
<jwalden>
I guess we unnecessarily had the find-associated-global thing from the start, didn't we
18:13
<gsnedders>
I'd rather just poision .set, on the whole.
18:14
<jwalden>
that seems probably doable/reasonable
18:14
<gsnedders>
It's not that elegant, but it doesn't leak any new capabilities (which is brendan's big concern), and it results in __proto__ being an accessor pair and one special case.
18:15
<gsnedders>
Then need to decide to object literal behaviour.
18:15
<jwalden>
object literals at least are syntactically distinguishable, thankfully
18:15
<gsnedders>
Compared with the number of special-cases it is now…
18:24
<Ms2ger>
annevk, congratulations on http://www.w3.org/TR/2012/REC-css3-mediaqueries-20120619/ ;)
18:25
<annevk>
heh
18:28
<annevk>
is there a wiki page describing IETF failures already?
18:28
<annevk>
I'd like to nominate http://lists.w3.org/Archives/Public/www-archive/2012Jun/0044.html
18:28
<annevk>
I'll make one now
18:30
<annevk>
http://wiki.whatwg.org/wiki/IETF
18:31
<rniwa>
annevk: it'
18:31
<rniwa>
s
18:32
<rniwa>
funny how some people don't know how bad IETF is
18:35
<MikeSmith>
rniwa: IETF is just people
18:35
<rniwa>
MikeSmith: sure.
18:36
<rniwa>
MikeSmith: there are a lot of people with good intentions just like W3C
18:36
<MikeSmith>
yeah
18:36
<annevk>
added some more
18:36
<rniwa>
MikeSmith: the problem is that there are handful of people who LOVE beaurocracy and adding policies
18:36
<Ms2ger>
There are a lot of people with good intentions at the W3C?
18:36
<Ms2ger>
</troll>
18:36
<MikeSmith>
granted, a number of them have their heads way up their asses as a far as browser technologies goe
18:36
<annevk>
the problem is rules
18:36
<annevk>
once you have too much of those
18:36
<annevk>
changing anything is a burden
18:36
<rniwa>
annevk: yup.
18:37
<rniwa>
any organization is bounded by the number of rules and policies they have.
18:37
<MikeSmith>
rniwa: not just policy, they have this idea of architectural purity which doesn't match implementation realities
18:37
<Ms2ger>
MikeSmith, sounds like the W3C ;)
18:38
<rniwa>
MikeSmith: that too is policies and beaurocracy to some extent.
18:38
<annevk>
WHATWG only requires people to not be an ass
18:38
<MikeSmith>
Ms2ger: I don't think anybody ever accused the DOM of architectural purity :)
18:38
<annevk>
everything else goes
18:38
<annevk>
MikeSmith: architecture astronouts
18:38
<MikeSmith>
yup
18:39
<annevk>
MikeSmith: and too much Java club
18:40
<MikeSmith>
we are by the way about to get some more love and attention about the web+ scheme
18:40
<annevk>
from the IETF?
18:40
<MikeSmith>
from the usual suspects
18:40
<annevk>
ask them if skype: or spotify: is on a registry yet
18:40
<annevk>
or magnet:
18:40
<MikeSmith>
heh
18:40
<MikeSmith>
well
18:41
<rniwa>
what we need to do in near future is kick out all those people who just like to create rules
18:41
<rniwa>
from both W3C and IETF
18:41
<MikeSmith>
we always merit special attentions
18:41
<rniwa>
then both of those organizations will become much more productive.
18:41
<MikeSmith>
rniwa: please god yes
18:41
<annevk>
MikeSmith: sure sure, we break the rules :)
18:41
<MikeSmith>
rniwa: if you can find a way to make that happen, you will be my hero forever
18:42
<MikeSmith>
rniwa: they don't necessarily want to create new re
18:42
<rniwa>
of course, people are kicked out immediately if they argued that we should create a policy/rule on how to kick people out.
18:42
<MikeSmith>
*new rules
18:42
<MikeSmith>
he
18:43
<rniwa>
MikeSmith: i guess. they have done enough damanges
18:43
<MikeSmith>
rniwa: it's more like, they want to protect the entirely arbitrary set of rules they already created
18:43
<rniwa>
they just want to maintain their collateral damages they have already caused :\
18:43
<rniwa>
MikeSmith: yeah.
18:43
<rniwa>
they're on the defense mode now.
18:43
<rniwa>
because new rules are created as a result of existing rules at this point.
18:43
<annevk>
W3C too
18:43
<MikeSmith>
arbitrary rules wrapped in the disguise of sound "architecture"
18:43
<rniwa>
MikeSmith: yup...
18:44
<annevk>
"oh you want change?" "did you talk to the AC yet?"
18:44
<annevk>
ooh
18:44
<annevk>
that's actually a nice meme
18:44
<rniwa>
i sometimes look at longdesc discussion on html WG
18:44
<Ms2ger>
Why?
18:44
<Ms2ger>
Do you hate yourself?
18:44
<rniwa>
and realize i don't understand most of stuff they're talking about
18:44
<rniwa>
like 90% of their discussion is about chair doing X while, requesting to reopen Y for reason Z
18:45
<rniwa>
s/ while//
18:45
<rniwa>
and all that non-sensical beaurocratic cr*p no technical person would care about.
18:45
<MikeSmith>
in the late 1960s, the counter-culture movement had this saying, "Don't trust anybody under 30". An analogy is, "Don't trust anybody who's never worked in a browser product-development team."
18:46
<rniwa>
MikeSmith: that might be not a sufficient criteria
18:46
<Ms2ger>
MikeSmith, I was thinking that includes annevk, but it doesn't
18:46
<MikeSmith>
it's a start
18:46
<rniwa>
MikeSmith: there appears to be a lot of beaurocratic people form browser vendors as wel.
18:46
<Ms2ger>
He got a patch into Gecko at some point :)
18:47
<MikeSmith>
heh
18:47
<gsnedders>
I write PPC machine code, if you believe our VCS. I don't.
18:47
<rniwa>
anyways, we need to make it happen.
18:48
<rniwa>
beaurocacy is destroying our future :(
18:48
<MikeSmith>
rniwa: you are already making it happen. the other stuff is noise
18:48
<MikeSmith>
it's an annoyance but it's not blocking you from getting work done in the end
18:49
<Ms2ger>
Unless you let it
18:49
<MikeSmith>
but it would be nice to sideline it more
18:49
<MikeSmith>
as far as W3C, I think in the webapps WG, the noise is minimal, relatively
18:49
<rniwa>
MikeSmith: not sure. i think a lot of beaurocratic stuff is scarying people away
18:50
<rniwa>
MikeSmith: yeah, web apps is working relatively well.
18:50
<MikeSmith>
OK, point taken, there
18:51
<annevk>
WebApps meetings are quite bad, but then HTML is so much worse...
18:52
<MikeSmith>
yeah
18:52
<MikeSmith>
it's all relative
18:52
<MikeSmith>
so one lesson I guess is, for API work at least, do it in places where the loonies and IETF architectural astronauts have the least amount of process influence
18:52
<Ms2ger>
"meetings"
18:52
<annevk>
http://w3cmemes.tumblr.com/post/25376892373/changing-the-w3c
18:52
<MikeSmith>
heh
19:16
<tantek>
annevk - is it convince the Advisory Committee or Advisory Board?
19:17
<tantek>
rules are just code that run on people. perhaps they belong on github.
19:18
<Ms2ger>
bitbucket
19:22
<annevk>
tantek: AC makes the decisions on WG creation, copyright, what the W3C works on in general, etc.
19:23
<tantek>
AC votes - right. And results are either secret or member-only?
19:23
<annevk>
tantek: I believe what the AB does is subject to AC approval too, but if the AB advises something the AC would probably accept it
19:23
<annevk>
tantek: pretty much everything the AC does is Member-only
19:35
<annevk>
hober: w3cmemes is not tweeting?
19:46
<kennyluck>
Do people here seriously think changing rules is useful? I don't see how chaining rules at the W3C would give as an active editor of CSSOM.
19:47
<Ms2ger>
You realize why we don't have one right now, right? :)
19:47
<kennyluck>
Ms2ger, I have my theory.
19:57
<tantek>
kennyluck, changing rules can be useful just like changing code can be useful. both can also lead to bloat / featuritis.
19:57
<tantek>
also, CSSOM is both difficult and relatively thankless to edit.
19:58
<annevk>
CSSOM would have been easier for me to maintain if there were less rules
19:59
<tantek>
annevk - indeed. though in this case, the cost of uncivil behavior appeared to be worse than the cost of rules.
19:59
<kennyluck>
annevk, I don't think it's at the maintenance stage. It's just incomplete.
20:01
<annevk>
it's both really
20:01
<annevk>
the model is in my head mostly and in XXX comments throughout the spec
20:01
<annevk>
just had to be written out at some point
20:02
<annevk>
but e.g. getComputedStyle, Media Queries, and style sheet switching was defined in detail
20:02
<tantek>
I'd like to see Chaal's comments in private email about "not pissing off productive editors" made public somewhere.
20:04
<kennyluck>
Yeah. I am not trying to piss of productive editors. Changing rules has an effect on making the environment perhaps more comfortable to the editors, but it seems pretty irrelevant in terms of "getting more people".
20:04
<Ms2ger>
I'm not sure I'd agree with that
20:05
<kennyluck>
Ms2ger, mind explaining a bit more?
20:05
<Ms2ger>
Sure, but not tonight, I'm about to head off :)
20:07
<annevk>
if people are excited about the W3C, more people will come
20:08
<annevk>
if your existing editors are deserting for more exciting fields (WHATWG / WHATCG), more people will go there
20:10
<jgraham>
Hixie: What's the point of step 6 in the "browsing context is /navigated/ to a new resource" algorithm?
20:10
<kennyluck>
annevk, I just don't think changing rules at the W3C would actually increase the sum of efforts spent at WHATWG + W3C in a significant way.
20:12
<annevk>
I guess you're not editing anything
20:13
<annevk>
the thing is, there's a lot of bullshit that comes on top of the editing
20:15
<kennyluck>
But at least I don't think the CSSOM editors as *listed* for now are not actively editing the spec because of this.
20:16
<annevk>
I don't really see how CSSOM is that relevant into whether or not W3C has too much bureacrazy
20:16
<annevk>
bureaucrazy*
20:18
<annevk>
all the people that are just forwarding email around and shepherding groups of people could do actual work instead
20:18
<kennyluck>
annevk, I guess it's not. I am just trying to understand what problems getting rid of W3C bureaucracy solve.
20:18
<kennyluck>
A
20:20
<annevk>
technical details get addressed rather than postponed "to the next version"
20:20
<annevk>
no need to have endless discussions over rules
20:20
<kennyluck>
annevk, I do think it's useful as to provide positive environment to the editors, but I am more interested in understanding whether it has an effect, positive or even negative, in terms of "getting new people".
20:21
<annevk>
no fricking tarpit that is the HTML WG
20:21
<annevk>
kennyluck: you know, if "getting new people" is your only goal, you're going to lose me
20:22
<annevk>
you lost Hixie, you lost a whole bunch of contributors from browser vendors
20:22
<annevk>
but you know, keep thinking of "getting new people" while those with the experience plant some seeds elsewhere
20:27
<kennyluck>
Didn't Hixie say "W3C is irrelevant"?
20:30
<annevk>
could well be
20:31
<annevk>
I think it does some relevant things still, but I care less and less for it
20:34
<kennyluck>
I don't really think WHATWG and W3C in terms of two parties, and that's why I don't changing rules at the W3C is too useful. Those that don't like bureaucracy would just do the WHATWG way, like Hixie.
20:40
<jgraham>
Some things like CSS are rather entrenched at W3C; HTML they abandoned (arguably twice now) so it was rather easy to get people to move elsewhere. With things that they still work on it is harder to get a critical mass of existing contributers to move elsewhere whilst still having the effect of excluding new contributers
20:43
<jarek>
HTML5 spec says that element's id attribute must be unique in the whole document
20:43
<jarek>
but the browsers don't really care
20:44
<jarek>
so... why not allow duplicate ids in the spec?
20:44
<kennyluck>
So this is sort of CSS in the WHATWG way → http://lists.w3.org/Archives/Public/www-style/2012Jun/0422 . Though, apparently I can't make François and folks to think more like a browser vendor, and I don't think chugging rules at the W3C would help here, either.
20:44
<kennyluck>
s/chugging/changing/
20:48
<Philip`>
jarek: Duplicate IDs cause confusion (when using foo.html#fragment, getElementById, etc), so it's helpful to authors if validators complain about duplicate IDs, so the spec says they're invalid
20:49
<jarek>
Philip`: I would expect using foo.html#fragment and getElementById() to return first ID in the DOM tree
20:52
<Philip`>
jarek: The behaviour is obvious if you think about it, but if e.g. someone later modifies your HTML document to swap a sidebar from the left to the right and it results in a different element being the first in the DOM tree with that ID, then it won't be obvious that the resulting problems were caused by the use of duplicate IDs
20:54
<jarek>
some CMSs are discouraging the usage of IDs altogether because they can't guarantee that there will be no naming clashes with third party modules
20:59
tantek
reads http://lists.w3.org/Archives/Public/www-style/2012Jun/0422 and http://fremycompany.com/TR/2012/ED-css-custom/
20:59
<tantek>
I like the meta-approach - just going ahead and drafting *something* and especially checking it into github (rather than the harder / more painful to use systems like hg etc.)
21:02
kennyluck
is a bit proud that he suggested this :p
21:12
<Philip`>
jarek: That sounds like it would be equally problematic if duplicate IDs were permitted
21:12
<jarek>
Philip`: yeah, but using classes everywhere also feels "wrong"
21:12
<Philip`>
jarek: Presumably the real solution would be to have some kind of scoped identifier system, so each module can uniquely identify elements within itself
21:13
<Philip`>
jarek: or emulate that by adding some unique prefix to each of a module's IDs (in the HTML and CSS and JS)
21:13
<jarek>
Philip`: I guess the solution is to use something like <div data-id="identifierUniqueInComponentScope">
21:14
<Philip`>
jarek: Just using class doesn't sound like a solution to what I presume the problem is (i.e. wanting some kind of unique identifier that won't clash with other modules)
21:14
<jarek>
but targeting such IDs from CSS is not as convenient as regular IDs
21:15
<jarek>
or <div data-componentName-id="blah">
21:17
<tantek>
rather than data-id="identifierUniqueInComponentScope" why not class="identifierUniqueInComponentScope" ?
21:27
<jarek>
tantek: semantically classes are meant to categorize elements, e.g. '.button', '.toggleButton', '.slider'
21:29
<jarek>
technically there is no difference and classes are actually easier to work with (there is element.classList API)
21:34
<tantek>
jarek - a category can have just one thing
21:34
<tantek>
these scoped identifiers are categories of a sort
21:35
<tantek>
since there might be more than one on a page
21:55
<othermaciej>
tantek: are Chaal's comments in private email about "not pissing off productive editors" that you alluded to earlier available somewhere in Memberspace
21:55
<othermaciej>
?
22:01
<tantek>
othermaciej I'm looking...
22:02
<tantek>
othermaciej - in "team-webapps" apparently - which I don't think is available to memberspace. :/
22:49
<tantek>
huh, openmetadata.org launches … wait for it ….
22:50
<tantek>
…. without any of their pages using their own metadata "standards".
22:50
<tantek>
(though due to their use of WordPress, they get a bit of hAtom microformats support)
22:57
<tantek>
June appears to be the time of the year when this happens (new metadata site launches that doesn't actually use/publish their own metadata). this year openmetadata.org. last year schema.org. in 2009 it was commontag.org. oh, Google Base is an exception - launched 2005-11-15 http://googleblog.blogspot.com/2005/11/first-base.html