06:19
<Aleoss>
Can someone tell me how to fix/resolve the Tidy errors in my document? http://pastebin.org/67228
06:26
<Vito`>
Aleoss: looks like Tidy doesn't like the HTML5 doctype, so it's not parsing the HTML entities
06:26
<Vito`>
because your page validates through html5.validator.nu
06:27
<Aleoss>
Vito: So is there a way I can make the entities validate through Tidy?
06:27
<Aleoss>
Vito: Or is that out of my control?
06:29
<Vito`>
I only see people complaining on it in a google search
06:29
<Vito`>
I'd guess out of your control until Tidy becomes HTML5-aware
06:30
<Vito`>
You could put in an HTML4 doctype to make sure your entities validate, and then ignore all the errors about the HTML5 stuff
06:32
<Aleoss>
Vito: I'm designing my page for HTML5 using HTML5 standards. I'm not going to test it through a doctype that I don't intend on using. That would leave room for REAL errors.
06:32
<Aleoss>
Vito: Thanks for confirming my conserns in regards to Tidy.
06:36
<Aleoss>
Truly is a pity..
06:37
<Vito`>
Why are you using Tidy in particular? Isn't the other validator enough?
06:37
<Vito`>
(I've never personally used Tidy, so I don't know its benefits for < HTML5)
06:40
<Aleoss>
Vito: I like to use Tidy (Serial) + Total Validator (Serial with WAI-AAA WCAG) + W3 HTML Validator.
06:42
<Aleoss>
Vito: And if I'm really bored, I'll validate with cynthiasays.com as well.
06:44
<Aleoss>
Vito: Call it over-kill if you wish, but I call it getting the opinions of various sources to ensure 100% accessibility and validity.
08:13
<a1>
Hello
08:13
<a1>
Is this the place to ask questions about HTML5?
08:14
<Vito`>
that's what the topic in #html says ;)
08:15
<a1>
My question: is it possible to add <audio> tag from javascript, make it play file from some javascript scalar; it contains actual OGG data and not link to it?
08:16
<a1>
var audioElement = document.createElement('audio'); audioElement.setAttribute('src','data:audio/ogg;'+data);audioElement.play();
08:16
<a1>
this doesn't work
08:17
<a1>
Converting to base64 with javascript seems to be pretty slow
08:18
<Vito`>
how interesting
08:18
<Vito`>
does it work with a data URL without using javascript to inject the whole thing?
08:18
<Vito`>
as a regular <audio> tag in the page?
08:19
<Vito`>
the notes on MDC around data URLs say you shouldn't expect long data URLs to work everywhere, and I imagine an OGG file would be huge
08:20
<a1>
It works in firefox; 2MB ogg file
08:20
<a1>
It doesn't have to be data:
08:21
<a1>
I just want to play file I have stored in javascript variable.
08:22
<a1>
I actually only need this to work on firefox, as I'm writing a greasemonkey script
08:32
<Vito`>
a1: are you actually adding the element to the page? appending it to the body or some such? and have you tried delaying the play to make sure it has time to attach to the DOM?
08:33
<Vito`>
I haven't tried this myself but if it works with a data URL as part of the HTML, I'd expect it to also work when it's injected
08:33
<a1>
I'll try
08:53
<roc>
a1: that shoudl work
08:53
<roc>
the element doesn't need to be in the DOM
08:54
<roc>
if you base64 encode the data, you need to add 'base64' to your data: URL
08:54
<a1>
I think one problem is that 'data:audio/ogg;'+data should be 'data:audio/ogg;'+encodeURI(data)
08:55
<erlehmann>
a1, roc, cn you repost the offending data URI ?
08:55
<erlehmann>
just joined
08:55
<a1>
no, it's nothing
08:55
<a1>
problem in me, not data
08:56
<zcorpan>
al, it should be a comma, not a semicolon
08:56
<zcorpan>
s/al/a1/
08:57
<erlehmann>
… anyone visiting Berlin for christmas and attending 26C3 ? (Won't ask again.)
09:03
<GPHemsley>
Does HTML5 define named entities? Or point to a reference of them somewhere else?
09:04
<zcorpan>
it defines them
09:04
<GPHemsley>
Oh, found it: Section 9.6
09:05
<GPHemsley>
Called them "Named character references"
09:05
<zcorpan>
yeah, they were called entities before
09:05
<zcorpan>
but people said it clashed with xml's terminology
09:06
<GPHemsley>
ah
09:07
<GPHemsley>
Hixie: It'd be nice if that table also had a column that showed what the symbols looked like (e.g. by using the &#x....; form and letting the browser render them)
09:09
<zcorpan>
GPHemsley: there are nicer interfaces than a big table if you want to find and use a character...
09:10
<GPHemsley>
zcorpan: True. But my mind always goes straight to source when wondering about something.
09:12
<GPHemsley>
Hixie: BTW, the definition of "Named character references" in section 9.1.4 is in conflict with the table in section 9.6, specifically with regard to the point about having to end in a semicolon. The table has duplicate entries for NCRs that do not end with a semicolon (presumably, for backwards compatibility).
09:14
<zcorpan>
GPHemsley: 9.1.4 excludes those without semicolon
09:15
<GPHemsley>
zcorpan: Is that not the only place that references the NCRs defined in 9.6?
09:15
<GPHemsley>
IOW, wouldn't that make the NCRs without a semicolon in 9.6 orphans?
09:17
<zcorpan>
the parser also references 9.6
09:17
<GPHemsley>
which section is that?
09:17
<zcorpan>
9.2
09:19
<GPHemsley>
hmm... can't find where; do you have a subsection?
09:19
<zcorpan>
9.2.4.70 Tokenizing character references
09:20
<GPHemsley>
ah, that's why... 9.2.4 is a separate page :P
09:22
<GPHemsley>
zcorpan: So, are you referring to the "Anything else"."Otherwise, ..." part?
09:25
<zcorpan>
i see 2 references to 9.6 in "anything else"
09:26
<GPHemsley>
right... I'm asking specifically about the NCRs that do not end in a semicolon
09:26
<zcorpan>
what about them?
09:27
<Dashiva>
"Otherwise, if the next character is a U+003B SEMICOLON, consume that too. If it isn't, there is a parse error."
09:27
<Dashiva>
That part?
09:27
<GPHemsley>
zcorpan: well, you said they weren't orphans and/or that they were excluded from 9.1.4
09:28
<GPHemsley>
Dashiva: No, "Otherwise, return a character token for the character ..."
09:28
<GPHemsley>
zcorpan: So I was asking where they were included.
09:28
<zcorpan>
if the input is foo&aumlbar then the parser will treat it as fooäbar
09:28
<zcorpan>
with a parse error
09:29
<zcorpan>
"Consume the maximum number of characters possible, with the consumed characters matching one of the identifiers in the first column of the named character references table (in a case-sensitive manner)."
09:29
<GPHemsley>
right, that's what I was asking about
09:31
<GPHemsley>
I see, so the ones without the semicolon are still processed, but they also raise a parse error?
09:31
<zcorpan>
yes
09:31
<GPHemsley>
If, however, you use an NCR that does not have a semicolon-less counterpart, then it's just a parse error and nothing is shown, or everything is shown?
09:32
<zcorpan>
then it's a parse error but the entity is not replaced
09:32
<zcorpan>
because it's not an entity
09:32
<GPHemsley>
so, it's displayed as literal?
09:32
<zcorpan>
yes
09:32
<GPHemsley>
gotcha, OK
09:32
<GPHemsley>
thanks
09:35
<a1>
I managed to get it to work
09:35
<a1>
The problem was that data I was getting from xmlhttprequest was interpreted as utf-8 string
09:36
<a1>
and was base64 encoded incorrectly
09:36
<a1>
Thanks
09:39
<zcorpan>
a1 wants xhr2's responseBody
09:41
<zcorpan>
or possibly <audio src>
09:50
<GPHemsley>
Does HTML5 make any requirements for the style of the bullet for <li> with regard to embedded lists?
09:51
<gsnedders>
GPHemsley: No, HTML 5 has no rendering requirements
09:54
<GPHemsley>
gsnedders: Hmm... that's not entirely true: http://www.whatwg.org/specs/web-apps/current-work/multipage/rendering.html#punctuation-and-decorations
09:55
<GPHemsley>
But for some reason it specifies differing/alternating styles for embedded uls, but not embedded ols
09:56
<gsnedders>
That's non-normative
09:56
<gsnedders>
So it doesn't require anything :P
09:56
<GPHemsley>
:P
09:57
<GPHemsley>
looks like I'll have to specify ol[type]
09:57
<zcorpan>
type is invalid
09:57
<GPHemsley>
ugh, was just looking that up
10:10
<a1>
It's me again with same question.
10:10
<a1>
I have a javascript scalar contatining OGG stream
10:11
<a1>
it starts with "OggS\0\2" and contains some music I'd like to be able to play with it
10:11
<a1>
with <audio> tag
10:12
<a1>
It's possible to play it by setting src attribute of audio to 'data:audio/ogg;base64,'+encode64(a)
10:12
<a1>
but encoding huge sound file with javascript is very slow
10:12
<a1>
my browser freezes
10:12
<a1>
So I'm asking
10:12
<a1>
is there a way to set data of <audio> tag, and not source of that data with javascript?
10:16
<zcorpan>
no
10:16
<zcorpan>
do you need to base64 it though?
10:16
<a1>
well, it's binary data
10:17
<zcorpan>
sure
10:17
<a1>
'data:audio/ogg,'+a doesn't work
10:17
<zcorpan>
does encodeURIComponent(a) work?
10:17
<a1>
'data:audio/ogg,'+encodeURIComponent(a)
10:17
<a1>
this?
10:17
<a1>
No..
10:18
<zcorpan>
hmm, wonder why not
10:18
<jgraham>
zcorpan: encodeURIComponent does something quite different
10:18
<jgraham>
Doesn't it?
10:19
<jgraham>
Oh wait
10:19
<a1>
Output of encodeURIComponent(a) seems to be okay...
10:19
<a1>
"OggS%00%02%00%00%00%00%00%00%00%00'%EF%9F%B7%EF%9E%83Z%00%
10:19
jgraham
was answering a different question
10:20
<jgraham>
a1: Have you tried using a2b or whatever it's called, in browsers that support it?
10:20
<zcorpan>
atob
10:22
jgraham
wonders why that isn't a part of html5
10:22
<zcorpan>
file a bug? :)
10:26
<a1>
It's data I get from XMLHttpRequest, and characters that >127 need to be and-ed with 0xff
10:26
<a1>
calling btoa on that results in exception
10:29
<zcorpan>
why do you get the data with xhr rather than with <audio src>?
10:30
<a1>
Basically, it's a jpg image catenated with ogg audio file. I can't just specify that as source of <audio>
10:30
<a1>
I get it with xhr, throw away jpeg data, and feed what's left to <audio>
10:31
<jgraham>
OK new question: Why do you have a JPEG image concated with an OGG audio file?
10:31
<a1>
some widely used discussion board allows uploading of images
10:32
<a1>
I want to make to make it possible to allow users to upload and listen to sounds
10:32
<a1>
with greasemonkey
10:32
gsnedders
notes you'll hit issues with some browsers with null bytes in the data, and with stuff that isn't UTF-8
10:33
jgraham
wonders if you could use some clever hack with <img> <canvas> and getImageData
10:34
<a1>
getImageData?
10:35
<a1>
That returns pixels, doesn't it?
10:36
<zcorpan>
yes
11:16
<zcorpan>
can we update http://www.w3.org/QA/2002/04/valid-dtd-list.html ?
11:18
<hsivonen>
zcorpan: maybe dom could if you can convince him.
11:19
<zcorpan>
and the referenced http://www.w3.org/QA/Tips/Doctype has factual errors
11:30
<zcorpan>
does firefox still look at content-style-type?
11:33
<karlcow>
zcorpan: for the qa tip it is easy to modify, there is a process
11:33
<zcorpan>
what's the process?
11:33
<karlcow>
for the dtd list it's easy to do.
11:34
<karlcow>
zcorpan: http://www.w3.org/QA/Tips/
11:35
<karlcow>
if you want I can send some suggestions, I see indeed some factual errors
11:36
<karlcow>
Tools which process HTML documents, such as Web browsers, need to know which DTD an (X)HTML document
11:36
<karlcow>
s/such as Web browsers//
11:43
<zcorpan>
sent an email to dom
11:47
<karlcow>
zcorpan: hehe http://lists.w3.org/Archives/Public/public-evangelist/2009Dec/0000
11:48
<zcorpan>
karlcow: thanks. my email was about the other document
11:48
<karlcow>
excellent
11:52
<zcorpan>
karlcow: i'd also suggest replacing the first further reading link with http://hsivonen.iki.fi/doctype/
11:55
<karlcow>
http://www.alistapart.com/stories/doctype/ plus this one has went the way of the dodo. the one of hsivonen should be added too maybe a bit technical for certain people but good to have indeed.
12:00
<gsnedders>
karlcow: http://www.alistapart.com/articles/doctype/ I guess is it
14:01
<hsivonen>
the third Definition below says only the 5 XML ones
14:02
<hsivonen>
MikeSmith: so I suggest making a copy of NamedCharacters that only has 5 entries
14:03
<MikeSmith>
that'd certainly make things easier
14:04
<MikeSmith>
but are you sure that's what the third Definition is actually saying?
14:04
<MikeSmith>
I can't see that it's stating that restriction
14:04
<hsivonen>
zcorpan: ^ spec not clear enough :-)
14:05
<MikeSmith>
hsivonen: seems like there are two parts to what is says there
14:05
<MikeSmith>
"Each CharRef in PseudoAttValue is replaced with the character it represents according to XML. [XML]"
14:05
<MikeSmith>
which isn't stating any restriction
14:06
<MikeSmith>
and "Each PredefEntityRef in PseudoAttValue is replaced with with U+0026 (&) if it is "&amp;", U+003C (<) if it is "&lt;", U+003E (>) if it is "&gt;", U+0022 (") if it is "&quot;" and U+0027 (') if it is "&apos;"."
14:06
<hsivonen>
MikeSmith: CharRef is a numeric reference, isn't it?
14:06
<hsivonen>
and PredefEntityRef is a named ref
14:06
MikeSmith
looks at XML spec now
14:07
<MikeSmith>
hsivonen: yeah, OK, CharRef is numeric
16:08
<foolip>
is there even a validator for XHTML5+RDFa that includes the microdata attributes?
16:09
<erlehmann>
dunno lol :3
16:10
<foolip>
I doubt we'll see one a validator.nu
16:10
<foolip>
at
16:12
<erlehmann>
Hmmm, if both is not possible at once, I've no idea how to continue. The Creative Commons person says RDFa is the way to go, Hixie wants Microdata … should I just make it an option in the interface?
16:12
<erlehmann>
like: ( ) RDFa (x) Microdata
16:13
<erlehmann>
Pro: Not my problem anymore. Con: Users will likely not understand anything at all.
16:14
<foolip>
I doubt any consumers care about validity and since the attributes don't clash they won't interfere with each other
16:14
<foolip>
but it seems a bit odd to support both in the same document
16:14
<foolip>
is there any benefit to supporting RDFa, like Google giving preferential treatment or something?
16:15
<erlehmann>
Anyway, nice Trilemma … Choose two: Microdata, RDFa, Valid Markup. :D
16:15
<foolip>
hehe
16:16
<erlehmann>
Well, there are already RDFa consumers out there, I assume.
16:16
<foolip>
I know which one I'd sacrifice, but then I'm kind of biased.
16:17
<erlehmann>
Hehe.
16:18
<foolip>
why not also stuff some microformats in there
16:18
<foolip>
at least that won't affect validity :)
16:21
<erlehmann>
=_=
16:22
<jgraham>
Maybe you could add the data as RDF/XML hidden inside commnts and as JSON blocks inside <script type="application/json"> while you're at it :)
16:22
<erlehmann>
Got it. I'll stay with Microformats for now and fix the styling. Is there some wiki page with a list of Microformat consumers ?
16:23
<foolip>
right, wasn't CC doing RDF/XML in comments?
16:24
<erlehmann>
well, since i link to the CC license deed … you can have microformats in the generated code and RDF in CC territory ;)
16:24
<jgraham>
foolip: Not sure. Trackback use(es|d) it I believe
16:24
<Philip`>
Anyone who wants to extract licensing information from the web will have to support whatever markup CC currently tells people to use
16:25
<Philip`>
so you might as well just use that, rather than adding complexity by requiring consumers support more types of markup
16:25
<MikeSmith>
Philip`: OK - thanks (about EBNF question)
16:25
jgraham
wonders how similar the set of all people who want to extract licesnisng information is to the empty set
16:25
<jgraham>
Do Google or Yahoo or Microsoft or someone read this data?
16:26
<Philip`>
jgraham: Well, I guess they're not disjoint sets so there is some similarity
16:26
<jgraham>
(and do people then make use of it when they read it?)
16:26
<Philip`>
*dissimilarity
16:26
<jgraham>
Philip`: :p
16:26
<erlehmann>
Dunno, Google or Yahoo at least have this data on their pages.
16:26
<erlehmann>
Damn, now I have to learn Git branching.
16:26
<Philip`>
I'm fairly sure some search engines can use it
16:26
<foolip>
I know I often search for a specific license on google images or flickr, perhaps some of that information came from scraping
16:26
<Philip`>
http://search.yahoo.com/web/advanced?ei=UTF-8
16:27
<Philip`>
"Search only for Creative Commons licensed content"
16:27
<foolip>
I don't remember ever finding anything with google though, perhaps the feature doesn't exist
16:28
<jgraham>
foolip: I use it on flickr but that doesn't require scraping since flickr have access to their own db
16:29
<foolip>
I used to use CC:s search tool, but that's just a frontend for other things
16:29
Philip`
wonders how common it is for sites to scrape themselves
16:29
<TabAtkins>
That would be kinda weird.
16:29
<Philip`>
I suppose lots have generic search engines that basically do that
16:30
<TabAtkins>
But yeah, generic search engines would be the common use case.
16:30
<TabAtkins>
I suppose I do that? I use Google Custom Search on my company's site.
16:30
<jgraham>
TabAtkins: So you just confessed to weirdness?
16:30
<TabAtkins>
Yes.
16:31
<Philip`>
I was thinking more of ones that are self-hosted, like Ultraseek or (to some extent) the Google Search Appliance
16:31
<Philip`>
where they could technically get access to the private databases, but for practical purposes they probably just read data via the site's normal web interface
16:31
<foolip>
Philip`: does "scraping" with JavaScript count?
16:33
<foolip>
in that case any site using the Microdata DOM API (none yet I assume) would be scraping themselves
16:34
Philip`
doesn't have a sufficiently coherent point to be able to work out whether JS counts or not
16:34
<karlushi>
jgraham, I think you can imagine tools like tumblr, posterous or any blogging tools included in aggregator.
16:35
<karlushi>
parsing the content for the license and adding the credits in the reblogged page.
16:36
<karlushi>
another cool feature would be an authoring creating a blockquote with the right credits, authors, and/or license information
16:36
<karlushi>
s/authoring/authoring tool/
16:57
<erlehmann>
foolip, Microdata DOM isn't implemented yet, is it?
16:59
<foolip>
erlehmann: not in any browser, no
17:00
<erlehmann>
well, one could fake it with javascript pretty easily.
17:01
<foolip>
yes, which is what I've been playing with
20:46
<a1>
Hello.
20:46
<a1>
Could you please have a look at this:
20:46
<a1>
http://no-info.no-ip.info:6224/sound/test.html
20:47
<a1>
It's an audio tag with data: URI as its src
20:47
<a1>
I'm using firefox, and play/pause/seek functions are not working untl the whole sound has finished playing
20:47
<a1>
Is this how it should be working?
20:50
<annevk>
nope
20:51
<a1>
Bug in firefox or in my web page?
20:52
<annevk>
if your code is correct it's a bug in Firefox
20:52
<annevk>
haven't looked at the code since my connection is crappy
20:52
<a1>
<html><head></head><body><audio src="data:audio/ogg;base64,ALL-DATA-HERE" controls /></body></html>
20:54
<annevk>
looks good to me
20:54
<annevk>
well
20:55
<annevk>
<audio> needs an explicit closing tag
20:55
<annevk>
in text/html documents
20:57
<a1>
Still same behavior
21:04
<annevk>
i'd file a bug then
21:19
<jgraham>
annevk: I don't really know why you think the question about contexts is rhetorical. It seems you might mean "obviously not" but clearly you can write portable code by just using the 2D context or by doing feature testing or...
21:19
<jgraham>
So if by "that is rhetorical, right?" you meant "yes" it wasn't very clear
21:36
<roc>
a1: works for me
21:36
<roc>
oh, I see, we don't get the correct duration
21:37
<roc>
the first time through
21:37
<a1>
and can
21:37
<a1>
't seek
21:37
<roc>
right, because we only support seeking using HTTP
21:38
<a1>
'We' firefox or 'We' html?
21:38
<a1>
I filed a firefox bug.
21:38
<roc>
We firefox
21:41
<roc>
the problem is that Ogg doesn't include the duration at the start of the file, so we have to seek to the end of the file to get it, and that requires a transport like HTTP that supports seeking. We'll need to add some code so that when the download finishes, if we've got everything in the media cache we look at the data at the end to get the duration
21:42
<a1>
But the whole thing is loaded when player appears.
21:42
<a1>
"controls" goes after "src=data:..."
21:42
<a1>
so shouldn't whole steam already be in memory?
21:42
<Vito`>
oh neat
21:43
<Vito`>
glad you sorted it out a1
21:43
<a1>
Yeah, thanks
21:43
<a1>
apparently base64 was relatvely fast and some regexp I used to throw away data from picture was making firefox freeze
22:06
<annevk>
jgraham, okay...
22:06
<annevk>
jgraham, I was just going from if things work differently it's bad for interop
23:14
<abarth>
i'm really tired of the CORS/UM discussion on webapps
23:15
<abarth>
it doesn't seem to be going anywhere
23:16
<othermaciej>
abarth: did my suggestion that people only post if they have new information to add not work?
23:16
<abarth>
tyler and ian are still talking about x-frame-options
23:18
<abarth>
(i should stop complaining and just ignore the threads)