00:10
<caden>
do you ant to see it?
00:10
<caden>
s/ant/want/
00:15
<shepazu>
Hixie: do elements in HTML5 have "default actions"?
00:20
<shepazu>
ok, I'll try again... anyone, do elements in HTML5 have "default actions"?
00:20
<shepazu>
or is that nonsense?
00:23
<annevk4>
shepazu, they have activation behavior
00:24
<annevk4>
shepazu, e.g. the activation behavior of <a> is to navigate
00:25
<shepazu>
annevk4: ok... so, is that the terminology of HTML5, HTML4, both...?
00:26
<shepazu>
and if link is activated, what events does it produce?
00:26
<annevk4>
activation behavior is associated with the click event
00:26
<shepazu>
does it produce a click, or only if it was activated by the mouse?
00:26
shepazu
writes a test
00:26
<annevk4>
always click
00:27
<annevk4>
anyway, the idea was for DOM3Events to define activation behavior and various related thingies, but bjoern never got around to it so now it's defined in HTML5
00:30
<shepazu>
well, I'm actively editing DOM3 Events, so I'll define it there if it needs to be
00:31
<annevk4>
cool
00:33
<othermaciej>
shepazu: I think to support "click" and "DOMActivate" the way other browsers do, for elements that have an activation behavior, clicking needs to dispatch a "click" event whose default action is to dispatch a "DOMActivate" event, whose default action in turn is to perform the activation behavior of the element
00:34
<othermaciej>
shepazu: similarly, any activation not driven by the mouse (such as via keyboard or by calling the click() method) needs to dispatch a synthetic "click" event same as above
00:34
<othermaciej>
shepazu: if DOM3EV defines all this, then HTML5 could limit itself to stating which elements have activation behavior, and what that behavior is for each one
00:34
<othermaciej>
(I think)
00:35
<annevk4>
yeah, that is pretty much how HTML5 does it
00:35
<annevk4>
I still think we should nuke DOMActivate though
00:36
<shepazu>
what would change your mind?
00:37
<othermaciej>
DOMActivate adds code complexity, and it's not clear if it serves a useful purpose
00:37
<othermaciej>
otoh now that we've implemented it, it's not a huge burden to maintain support
00:37
<shepazu>
othermaciej: do you think it would be useful for D3E to define it? I guess if it did, I reckon I'd more or less take whatever HTML5 says on it
00:38
<shepazu>
it would be useful for e.g. SVG, I think
00:38
<annevk4>
shepazu, a good reason for having it besides click?
00:39
<othermaciej>
shepazu: I think it makes sense to define the notion of activation in D3E, because I think all specs with activation behavior would want to do it the same
00:40
<shepazu>
ok, I'll put that in an upcoming draft, thanks
00:40
<othermaciej>
annevk4: does Opera support DOMActivate?
00:41
<annevk4>
seems like we don't
00:42
<annevk4>
I filed the original bug for it in 2005 or so and all duplicates were from third-party testers
00:42
<othermaciej>
I'm wondering how likely it is that any Web content depends on it
00:42
<annevk4>
given the details I have it seems highly unlikely
00:43
<annevk4>
apart from some tests here and there
00:43
<shepazu>
could we find out from MAMA?
00:43
<shepazu>
or from Google's research?
00:44
<annevk4>
we can never be really sure, but both Opera and IE not supporting it seems like a good indicator to me
00:46
<othermaciej>
despite what Chaals said, I don't see WCAG specifically requiring DOMActivate
00:47
<othermaciej>
annevk4: it's possible sites could be using it in non-IE code paths and not caring or knowing that they break in Opera, although there doesn't seem much reason to doing it
00:47
<Darxus>
The head element says it contains "One or more elements of metadata content, of which exactly one is a title element."
00:48
<annevk4>
othermaciej, I guess
00:48
<annevk4>
othermaciej, I'd be interested in that code :)
00:48
<othermaciej>
Is there any SVG content that depends on DOMActivate?
00:48
<Darxus>
The meta element says it can be contained in the head element. Nowhere does it says anything about number of times. This annoys me. Although I suppose that means >=0.
00:48
<othermaciej>
(whether Web-published or not?)
00:49
<othermaciej>
So far it seems like the only hard dependency is XForms, and that doesn't seem like a good enough reason by itself
00:50
<othermaciej>
SVG 1.2 Tiny seems really vague about what elements will trigger a DOMActivate or what causes this to happen
00:50
<Darxus>
It'd be fun for gnuplot to have a terminal type of html5canvas.
00:51
<Darxus>
Actually, has anybody written a SVG / postscript / etc. converter to <canvas> yet?
00:51
<othermaciej>
it's implied that any rendered element can have activation behavior, the example given is a <circle>
00:51
<othermaciej>
Dino wrote a basic SVG-to-<canvas> converter long ago
00:52
<shepazu>
sorry, othermaciej, I was chopping mushrooms
00:54
<shepazu>
othermaciej: I don't know if any SVG content depends on it... I'm at a loss to find out if it does, other than surveys like MAMA
00:55
<annevk4>
figuring out if stuff breaks in experimental impl is likely a good solution
00:55
<annevk4>
again though, also for SVG we have received no bug reports on DOMActivate (apart from a test)
00:57
<shepazu>
othermaciej: well, if a circle had a DOMActivate event listener, or maybe a SMIL child element element... but other than that, I don't think any element in SVG should have default "activation behavior" than <a>
00:57
shepazu
tries to think of what other elements there might be in SVG that have activation behavior...
00:59
<shepazu>
http://www.w3.org/TR/SVGTiny12/elementTable.html
00:59
<shepazu>
maybe <audio> and <video>...
00:59
<annevk4>
elements are activatable based on their event listeners in SVG? ouch
01:00
<annevk4>
oh well, I suppose it does not matter much
01:02
<shepazu>
annevk4: how is it controversial to make an element activateable by putting a "DOMActivate" listener on it? I'm confused...
01:03
<shepazu>
but would the activation behavior of a multimedia element vary on whether the content was already playing or not? what's the activation behavior of HTML multimedia elements?
01:03
<annevk4>
shepazu, it seems activation behavior should come from semantics
01:04
<othermaciej>
shepazu: making an element activatable by putting a "DOMActivate" listener on it would be somewhat weird - an ancestor element with a DOMActivate listener would see the inner element get DOMActivate if and only if the inner element had its own listener
01:05
<shepazu>
othermaciej: ok
01:06
<othermaciej>
shepazu: since DOMActivate bubbles, that would be somewhat illogical I think
01:06
<othermaciej>
also, in general, listeners observe events, but don't observably affect what events get dispatched just by their presence
01:06
<shepazu>
I'm just grasping at straws on what elements in SVG would have inherent activation behavior... links, multimedia elements, maybe SMIL elements that didn't have explicit @begin values?
01:07
<annevk4>
seems like www-svg might be a better place to ask that than #whatwg :)
01:07
<othermaciej>
you could make the case that every SVG element should be activatable, but it's kind of unclear what forms of non-click activation, if any, would work for something like <circle> or <path>
01:07
<othermaciej>
and what the activation behavior would be
01:07
<Darxus>
I just had an HTML5 comprehension breakthrough. The "meta" element is in the "encoding declaration state" when its "http-equiv" attribute has a value of "content-type".
01:08
<othermaciej>
I could buy the argument that an activation-triggered SMIL animation makes you activatable, although that's a complicated rule
01:08
<othermaciej>
I think SVG <a> elements are clearly activatable
01:08
<shepazu>
yes, <a> is understood
01:09
<shepazu>
it's not actually a concept that SVG has had to deal with much, since there are so few truly activateable elements, I guess
01:10
<shepazu>
whereas HTML has quite a few
01:10
<Darxus>
Contexts in which the "meta" element can be used really should be changed from "If the charset attribute is present, or if the element is in the Encoding declaration state: in a head element." to 'If the charset attribute is present, or if the attribute "http-equiv" is set to "content-type".'
01:11
<shepazu>
annevk4: are you seriously telling me that I shouldn't try to resolve open questions about improving SVG by asking opinions in #whatwg
01:12
<annevk4>
I thought this was a question about how SVG works today
01:12
<annevk4>
because of "would"
01:12
<othermaciej>
annevk4: I think it's more a question of how it should work per spec, if D3E centrally defines the concept of activation
01:13
<othermaciej>
but it would also be useful to determine when actual SVG implementations fire DOMActivate events, and if this seems logical
01:13
shepazu
looks closer at activation behavior in HTML5, including where it's used
01:13
<annevk4>
shepazu, also, I included a smiley...
01:16
<othermaciej>
HTML5 actually defines that any element that is forced to be keyboard focusable by tabIndex is activatable
01:16
<othermaciej>
(with an activation behavior that does nothing, unless it would otherwise have an activation behavior)
01:16
<shepazu>
yeah, just looking at that
01:16
<shepazu>
ah, ok, thanks, that clarifies it
01:17
<othermaciej>
and <a>, <label>, <input>, <area>, <command> and <button> may have some built-in activation behavior, depending on circumstances
01:18
<othermaciej>
oddly, <bb> does not have defined activation behavior
01:18
<shepazu>
in that case, most SVG elements would have similar behavior for those elements, but still no true activation behavior... maybe it would be more accurate to say that the focusable attribute adds activation behavior
01:19
<shepazu>
othermaciej: maybe an oversight?
01:19
<othermaciej>
dunno - I'll send a comment
01:20
<shepazu>
I don't know the <bb> elements... you know of an example that uses it? the spec seems vague
01:22
<othermaciej>
it's not implemented or used by anyone yet
01:22
<Darxus>
Suggested changes to the draft go to the implementors list?
01:22
<othermaciej>
it's just... buttonish
01:23
<annevk4>
Darxus, no, to whatwg⊙wo
01:23
<annevk4>
I'm not really sure <bb> makes a lot of sense and the name is confusing
01:23
<Darxus>
annevk4: Just figured that out, thanks.
01:24
<othermaciej>
it's foolish to use such a heavily abbreviated name for an element with such a (relatively) obscure purpose
01:25
<annevk4>
I suggested (on behalf of someone else) <browserbutton>
02:22
<tantekc>
annevk4 - I just took a look at http://dev.w3.org/html5/markup/bb.html - and assuming that is close to what the <bb> tag currently means in HTML5, I strongly recommend NOT calling that a <browserbutton>
02:23
<tantekc>
the term "browser button" is already fairly widely used (Google, Technorati, others) as a less jargony replacement for what is otherwise known as a "bookmarklet" or "favelet"
02:23
<tantekc>
http://www.google.com/options/buttons.html
02:23
<tantekc>
http://technorati.com/tools/favelets.html
02:36
<csarven>
<bb> sounds pretty close to <input type="button"> http://www.w3.org/TR/html401/interact/forms.html#push-button
02:37
<csarven>
At least my understanding of it
02:57
<tantek>
csarven - in that case, if it's not clear how it's substantially different from <input type="button"> I would be for dropping the feature from the spec.
02:59
<tantek>
(dropping <bb> that is)
02:59
<csarven>
:)
03:06
<csarven>
http://dev.w3.org/html5/markup/input.button.html#input.button still exists, so, probably best that <bb>'s definition is cleared up, otherwise dropped if it had the same intention as <input type="button">
03:08
<csarven>
Is there a backlog sort of a document for what (and why) makes it into draft? Not particularly in the mood to track this down in the mailing list =)
03:10
<tantek>
csarven - I don't think I'm ever in the mood to track something down in a mailing list.
03:12
<csarven>
I don't know how Hixie manages his emails. That requires mad skills and sanity.
03:41
<othermaciej>
<bb> is supposed to be a special button for built-in UA functionality that it's generally believed should be restricted to user actions
03:41
<othermaciej>
so it's somewhat different than <input type="button"> - may have a custom rendering and its behavior is likely different than a normal button
03:41
<othermaciej>
but I think an element for that particular slice of functionality is overkill
03:46
<tantekc>
othermaciej - is there a wiki page where we can vote down elements?
03:47
<othermaciej>
tantekc: I don't believe there is
03:47
<othermaciej>
I think the official way to register your displeasure would be by sending mail to the list
03:51
<tantekc>
I think I'd prefer to unofficially collect more data before making an official statement
04:06
MikeSmith
makes not to remember to tell tantek that another mechanism is to raise an issue at http://www.w3.org/Bugs/Public/enter_bug.cgi?product=HTML%20WG
04:54
<Darxus>
Body elements are a lot easier to define than those in the header. I initially assumed the header would be much easier.
05:07
<Darxus>
Er, the page ended. The list of elements didn't.
05:07
<Darxus>
I am thankful for recursion.
07:32
<MikeSmith>
Hixie: I have a big favor to ask
07:45
<Philip`>
MikeSmith: http://dev.w3.org/html5/spec/404.html has wrong charset
07:47
<Philip`>
(Also, multipage pages don't seem to run the broken-link-fixing script, so if e.g. a block moves from one page to another then links to it won't work)
07:55
<MikeSmith>
Philip`: I (re)disabled the broken-link-fixing script temporarily
07:56
<MikeSmith>
because, genius that I am, the makefile change I added that enabled it also had the effect of generating a sort of endless cvs commit loop
07:56
<MikeSmith>
which by the time I stopped it had robotically committed about 80 updates to the draft
07:57
<MikeSmith>
to do it proper, I will need for Hixie to add <body onload="fixBrokenLink()"> to his W3C Overview.html file
07:59
<Philip`>
Oh, sounds excellent
07:59
<Philip`>
It's lucky you were using CVS, rather than a fast modern VCS that would have done millions of commits
08:00
<MikeSmith>
yeah
08:01
<MikeSmith>
thank god for the inefficiency of cvs
08:07
<MikeSmith>
Philip`: OK, I set UTF-8 as the default charset for the directory
09:25
<Philip`>
Hmm, a few days ago I attempted to use lang="" properly, for about the first time ever
09:26
<Philip`>
and now I discover that although my first use was correct, my second use was an accidentally copied <html lang="tr"> onto an English page
09:26
Philip`
apologises for polluting the world of metadata
09:30
<Lachy>
Philip`, nevermind, at least you have found and corrected your mistake, and order has been restored to the web.
09:33
<MikeSmith>
"the Web is an unkind mistress"
09:34
<Philip`>
I also completely forgot to add alt text to some images that really should have them (some plain icons, some icons in links) and the validator didn't remind me at all
10:13
<takkaria>
MikeSmith: it's the XML people who want to make it an unkind mistress, right now it doesn't really bother with the mistressing part AFAICT
10:15
<MikeSmith>
it's a fickle mistress, then
10:18
<takkaria>
I can live with that
10:33
Philip`
thinks something is really broken in his monitor, before realising Opera 10 now has a funky background image on error pages
10:48
<Rik|work>
arg, http://hsivonen.iki.fi/ is down :(
11:36
<takkaria>
hmm
11:36
<takkaria>
HTTP headers with q values were a really stupid idea
11:37
<takkaria>
should just have put things in a comma-separated list in order of preference
11:38
<Philip`>
Is the idea of "preference" not really stupid?
11:39
<Philip`>
I can't imagine how a browser developer would decide whether they prefer JPEG or PNG, or whether they prefer HTML or XHTML, etc
11:39
<Philip`>
since it depends on the content, and that depends on the server and the author and not on the browser
11:39
<takkaria>
I was thinking about Accept-Language myself
11:40
<Philip`>
Oh, okay
11:40
<takkaria>
but yeah, asking them to prefer one format over the other is a bit bizarre
11:40
<Philip`>
In that case it makes more sense, because users are likely to have preferences, and they use their browser to convey those preferences to the server
11:41
<Philip`>
(for languages)
11:41
<takkaria>
yeah, but the q-value bit is still nutty
11:41
<annevk4>
on top of that content negotiation does not really work
11:49
<gsnedders|work>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/185 really does seem suboptimal in WebKit
12:06
<olliej>
gsnedders|work: in the nightly?
12:08
<gsnedders|work>
olliej: In Chromium linux nightly at least
12:08
<olliej>
gsnedders|work: meh
12:09
<gsnedders|work>
olliej: I can't test Safari here :P
12:09
<olliej>
gsnedders|work: poor excuse
12:09
<olliej>
gsnedders|work: there's epiphany-webkit :P
12:09
<gsnedders|work>
olliej: But that's not in Ubuntu's repos :P
12:09
<olliej>
gsnedders|work: which uses real webkit :D
12:09
<olliej>
gsnedders|work: :p
12:10
<gsnedders|work>
olliej: Yeah, I know. I'm just trying to think of what does use WebKit+JSC and is in the repos :P
12:18
<zcorpan>
hsivonen: http://simon.html5.org/test/html/parsing/innerhtml.html hangs with the html5 parser
12:21
<annevk4>
zcorpan, your progress bar is tag abuse
12:21
<MikeSmith>
gsnedders|work: you can test WebKit nightlies on Linux using Arora
12:21
<gsnedders|work>
annevk4: It's not zcorpan'S test
12:22
<MikeSmith>
gsnedders|work: Qt-based .. it's pretty easy to build
12:22
<gsnedders|work>
MikeSmith: But that means I can't just reply upon package management to update it ever day ;P
12:23
<MikeSmith>
gsnedders|work: you update it by running "make" and "make install", as God intended
12:25
<gsnedders|work>
MikeSmith: I guess I could use a cron to update it
12:27
<MikeSmith>
gsnedders|work: http://code.google.com/p/arora/wiki/source#Building_with_Trunk
12:46
<Philip`>
gsnedders|work: You could switch to Gentoo, and then use www-client/arora
13:11
<zcorpan>
has anyone pointed out that DOMFocusIn bubbles while focus doesn't?
13:12
<othermaciej>
zcorpan: I believe it has been mentioned several times
13:12
<othermaciej>
I don't think that makes for a usefully distinct use case
13:13
<annevk4>
there's hardly any use case attached to that though, unless you have event handlers
13:13
<othermaciej>
non-bubbling is the sensible behavior for foucs
13:14
<othermaciej>
unlike clicks or mouse moves, a child of element X being focused should generally *not* be treated the same as if element X was focused
13:15
<othermaciej>
(and yes, nested focus is possible and may not even be entirely useless)
13:15
<othermaciej>
(consider an area that you can keyboard scroll when it has focus, but which contains form controls which can themselves have focus)
13:15
<annevk4>
(or the <video> element)
13:16
<othermaciej>
I think the only plausible argument for DOMFocusIn is if content indeed depends on it
13:16
<othermaciej>
and I think there's no argument whatsoever for standardizing focusin/focusout
13:17
<annevk4>
yeah
13:17
<annevk4>
I'm guessing I'm not going to be happy with the solution; similarly namespace support is still included despite nobody really asking for it
13:24
<gsnedders|work>
Philip`: Ubuntu has Arora packages, just not with WebKit trunk
13:36
<gsnedders|work>
Is there anything apart from @import that takes a URL but not as <uri> in CSS
13:37
<gsnedders|work>
?
13:40
<othermaciej>
annevk4: I thought we dropped namespaced event support long ago
13:40
<othermaciej>
annevk4: is it really still in there?
13:41
<othermaciej>
I wish the D3E editor's draft was easily googlable
13:43
<annevk4>
it is still there http://dev.w3.org/2006/webapi/DOM-Level-3-Events/html/DOM3-Events.html
13:44
<annevk4>
and for some reason the editor seems to assume my comments are in bad faith :/
13:44
<annevk4>
wtf
13:50
<zcorpan>
annevk4: do we support the namespaced stuff in opera?
13:51
<annevk4>
jl made an experimental impl once, I don't think it ever landed
13:51
<annevk4>
you'd have to check with bratell to be sure though
14:54
<gsnedders|work>
http://html5.digitalbazaar.com/a-new-way-forward/
14:56
<takkaria>
uh?
14:56
<takkaria>
"The single greatest complaint heard from the standards community concerning the development of HTML5 is that it has not allowed for the scientific process.
14:58
annevk4
wonders what "If your company depends on the Internet and can afford to fund just a small fraction of the work above (minimum $8K grant), then please contact him at msporny⊙dc" is about
15:06
<takkaria>
Manu seems to have missed the point as to why people in the whatwg list can't join the HTMLWG list
15:06
<takkaria>
*not all people
15:07
<takkaria>
and the idea that merging mailing lists will suddenly make people happy is a bizarre one
15:10
<takkaria>
and again with the "HTML5 has no distributed extensibility" meme
15:10
<takkaria>
sheesh
15:13
gsnedders|work
reminds himself that replying to that is not a productive use of his time
15:14
<takkaria>
a fair bit of that makes some interesting proposals
15:14
<gsnedders|work>
Mostly ones that have been made before, though
15:14
<takkaria>
but things like "Better, more precise accessibility language for HTML5." do kind of make me want to scream
15:15
<takkaria>
and saying that the scientific process is being disregarded whilst also saying experts don't get to dictate what happend next
15:15
<takkaria>
but yeah, responding to this isn't a good use of time
15:16
takkaria
goes back to work
15:18
<gsnedders|work>
I'll reply to it later, probably
15:18
gsnedders|work
creates a lynx -dump copy of that file to reply to
15:22
<Lachy>
"HTML5 is currently a walled garden that must be opened to the greater standards community in order to ensure a stable framework for the future of the Web." -- http://html5.digitalbazaar.com/a-new-way-forward/
15:23
<Lachy>
It's ironic that HTML5 has the most open development process compared with any previous HTML spec.
15:23
<takkaria>
no, see, 1000 people on a mailing list isn't open enough
15:23
<takkaria>
we need to open it to the experts. :)
15:27
<Lachy>
"In a CTC process, as used at the W3C, consensus should be reached before an editor changes a document." - we would never get anything done with that process
15:28
<Lachy>
we'd get more criticism from people complaining about how the spec has stalled if we adopted a process like that
15:29
<takkaria>
he doesn't suggest we move to a CTC model
15:30
<takkaria>
just that we magic up more committers
15:30
<gsnedders|work>
Lachy: Well, we can't publish until we have consensus, and I think in terms of time taken there's no difference between getting consensus on changes first and getting consensus after
15:35
<annevk4>
the HTML5 editor model is not that weird actually
15:35
<annevk4>
the CSS WG and WebApps WG are also using it
15:35
<annevk4>
those groups do make group decisions a bit more often than the HTML WG
15:35
<Lachy>
"it is difficult to tell which parts of the specification are at which level of maturity without deep knowledge of the history of the document" - Now he's just blatently ignoring the status markers we've had for over a year
15:36
<annevk4>
he might be reading the W3C edition
15:36
<Lachy>
annevk4, those groups are also smaller and don't have such large and diverse communities forming around them
15:37
<annevk4>
I'm not sure I'd say that about CSS, but the community there has arguably less of a say
15:38
gsnedders|work
wonders why HTML 5 and CSS 3 doesn't
15:47
<Lachy>
"The issue is that there are currently no proposals for distributed extensibility in HTML5 " - now that's just a blatant lie. He knows full well that the microdata section is in the spec for that. He may not like it as it is, but lying about it not being there doesn't help
15:49
<MikeSmith>
annevk4: I guess we should try to get the annotations added to the W3C version also
15:52
<Darxus>
It confused me for a few days that the status markers shown in the demo video weren't in the W3C version, because I wasn't looking at the WHATWG version.
16:00
<zcorpan>
demo video?
17:16
<vs-hs>
whatwg?
17:16
<vs-hs>
thelinx: :D
17:17
<Lachy>
vs-hs, what?
17:17
<vs-hs>
Lachy: No idea
17:47
<Darxus>
The spec has not been written primarilfy for validator implimentors.
17:49
<gsnedders|work>
The spec hasn't been written primarily for anyone
17:51
<Darxus>
The "Contexts in which this element may be used" and "Content model" have very different wording cosidering that they directly correspond to each other.
17:52
<Darxus>
Maybe change "Contexts in which this element may be used" to "Content models in which this element may be used"
17:54
<Darxus>
I should just write a parser that converts http://www.whatwg.org/specs/web-apps/current-work/ to schema :P
18:01
<Darxus>
<define name="phrasing_content"></define name="phrasing_content"> is invalid xml? :(
18:01
<Philip`>
Only start tags have attributes
18:02
<gsnedders|work>
Darxus: It's invalid in SGML and HTML too.
18:03
<Darxus>
Lame.
18:03
<Darxus>
I just prefer it over
18:03
<Darxus>
<define name="phrasing_conent">
18:04
<Darxus>
</define> <!-- end of phrasing_conent -->
18:04
<takkaria>
most people use indentation. :)
18:04
<Darxus>
Yeah, but this is for a case where the <define> is several pages long, so I'm not really sure which one I'm adding to without something like the above.
18:05
<Darxus>
Well, phrasing_contenet isn't that big, but flow_content is.
18:05
<Philip`>
<phrasing_content>...</phrasing_content>
18:05
<Darxus>
Philip`: I'd rather more closely follow relax ng.
18:06
<gsnedders|work>
Darxus: Why not use RelaxNG?
18:06
<Darxus>
Is xml 1.0 sgml conformant? To make </define name=""> valid xml, would I need to also update the sgml standard?
18:06
<gsnedders|work>
Darxus: And why not http://syntax.whattf.org/?
18:06
<Darxus>
gsnedders|work: Because it is not capable of specifying HTML5 to the HTML5 validator conformance requirements.
18:06
<gsnedders|work>
Darxus: XML 1.0 is a subset of SGML.
18:07
<Darxus>
gsnedders|work: That sounds like a yes.
18:07
<gsnedders|work>
Darxus: Sure, but do we need another schema language?
18:07
<Darxus>
gsnedders|work: If there isn't one that can sufficiently specify HTML5, I think so.
18:07
<gsnedders|work>
Darxus: Is RelaxNG with Schematron not enough?
18:07
<Darxus>
gsnedders|work: Yes.
18:07
<gsnedders|work>
Darxus: In what way?
18:08
<Darxus>
gsnedders|work: While Henry's thesis index is loading... one example was validating the layout of a table element.
18:09
<Darxus>
HTML5 requires that a table have, for example, no cells overlapping. No existing schema language can specify that. Henry implimented the requriement in java.
18:09
<gsnedders|work>
*Henri
18:09
<gsnedders|work>
Schematron should be able to check that though, I think. hsivonen's point is that you can't do it in any schema language, but Schematron isn't; a schema language.
18:10
<Philip`>
s/;//
18:10
gsnedders|work
doesn't entirely like this keyboard
18:11
<Darxus>
gsnedders|work: So why, when he was using both relax ng, and schematron, did he impliment some rules in java?
18:11
<gsnedders|work>
Darxus: Performance.
18:11
<Darxus>
And... schematron isn't a schema language?
18:12
<gsnedders|work>
I wouldn't call rule-based validation a schema
18:12
<gsnedders|work>
But yes, I guess technically it is
18:13
<gsnedders|work>
Ah, some of the data types can't be expressed in Relax NG apparently
18:13
<Darxus>
I believe schematron is not capable of sufficiently specifying HTML5. If I am incorrect, I would love to know.
18:15
<takkaria>
Darxus: you're inventing a schema language so you can write a validator that takes that schema language and uses it?
18:15
<gsnedders|work>
According to the intro of hsivonen's thesis, it is possible to check everything apart from some data types using RelaxNG+Schematron
18:15
<Darxus>
Schematron.com's header totally needs to be an svg.
18:15
<Darxus>
takkaria: That's my plan.
18:16
<Darxus>
I make no claims at being qualified to impliment this plan, but it entertains me.
18:16
<gsnedders|work>
Darxus: What do you gain by using a custom schema language to writing it all in code yourself?
18:16
<Darxus>
gsnedders|work: The ability to then write schemas for other things in that language.
18:17
<Darxus>
http://www.oreillynet.com/xml/blog/2007/01/diagram_comparing_schema_langu.html - ven-ish diagram of schema language capabilities.
18:17
<Philip`>
Hopefully nothing else in the world is as crazy as HTML5 and needing the same kind of schema language features
18:17
<gsnedders|work>
Darxus: What do you gain by being able to express data type restrictions in a schema?
18:17
<Darxus>
It seems like there should be a schema language that contains all of them plus the ability to validate HTML5.
18:17
<gsnedders|work>
Philip`: Atom Syndication Format can't be expressed in any either, AFAIK
18:17
<gsnedders|work>
Philip`: Actually, that's untrue.
18:18
<gsnedders|work>
Philip`: It isn't expressed fully by the included RelaxNG schema.
18:18
<Darxus>
gsnedders|work: Also, hopefully it would make it easier to update an HTML5 validator for HTML5.1.
18:18
<gsnedders|work>
Philip`: But you can express it fully if you combine that with Schematron
18:18
<Philip`>
gsnedders|work: I assume it doesn't need the ability to e.g. express table constraints, though
18:18
<gsnedders|work>
Philip`: But that still doesn't limit atom:updated, atom:published, etc. to being valid dates
18:19
<gsnedders|work>
Oooo! Tests have finished running!
18:19
gsnedders|work
goes back to work
18:20
<inimino>
Darxus: what happens if HTML5.1 adds new conformance criteria that are inexpressible in the new language?
18:20
<Philip`>
gsnedders|work: You need to write more tests, so you can spend more time on IRC waiting for them to finish
18:20
<Darxus>
inimino: Then I extend the new language.
18:30
<Darxus>
A machine readable version of the HTML5 spec would be much more difficult if it were not for recursion.
19:37
<ap>
Hixie: is the idea of using Upgrade header for WebSocket based on something web servers can do today? We're adding test infrastructure for WebSocket now, and I wonder if we can make it work inside Apache easily
19:40
<gsnedders|work>
ap: I think I heard someone talking about an Apache module for it
19:42
<ap>
gsnedders|work: googling suggests more like "venting about the lack of thereof", http://www.pmsenthilkumar.blogspot.com/2009/07/apaches-html5-websocket-conundrum.html :)
19:43
jwalden
needs to get around to writing a websocket server
19:43
<gsnedders|work>
ap: I mean about actually writing one themselves :)
19:44
<ap>
jwalden: we have an initial version at https://bugs.webkit.org/show_bug.cgi?id=27490 - not integrated with Apache in any way, though
19:45
<jwalden>
ap: I think the initial plan is for me to hack our web server enough to be able to implement the parsing, although it would also be lax and accept HTTP-style constructs that aren't also websocket
19:45
<jwalden>
eventually want something super-strict to throw out ill-formed inputs, tho
20:18
<Darxus>
Hah, HTML5 can't be valid XML because it doesn't allow associating a DTD. Which is required for valid XML.
20:21
<ttepass->
Darxus, <!DOCTYPE html> is a valid doctype declaration per XML, isn't it?
20:23
<Darxus>
ttepass-: It is a valid doctype, but XML is not valid if it does not specify (and validate with) a DTD.
20:23
<takkaria>
it's still wellformed, though, and you can still validate it. :)
20:24
<ttepass->
Ah: ?Note that it is possible to construct a well-formed document containing a doctypedecl that neither points to an external subset nor contains an internal subset.?
20:24
<Darxus>
"The XML Recommendation says that an XML document is 'valid' if it has an associated document type declaration and if the document complies with the constraints expressed in it."
20:25
<Darxus>
takkaria: Yes you can validate it against another schema but that still, apparently, makes it invalid xml.
20:25
<takkaria>
it doesn't make it invalid
20:25
<Darxus>
Quote from http://perl-xml.sourceforge.net/faq/ section 6.
20:25
<takkaria>
it just means that it's neither valid nor invalid
20:25
<Darxus>
I'm looking for the wording in the XML Recommendation.
20:26
<ttepass->
Darxus, section 2.8:
20:26
<ttepass->
[Definition: An XML document is valid if it has an associated document type declaration and if the document complies with the constraints expressed in it.]
20:27
<Darxus>
"An XML document is valid if it has an associated document type declaration and if the document complies with the constraints expressed in it" - http://www.w3.org/TR/REC-xml/ section 2.8
20:27
<Darxus>
Yeah...
20:27
<Darxus>
I interpret that as meaning that it is invalid without specifying a DTD.
20:29
<takkaria>
I think you're misinterpreting
20:30
<takkaria>
it would be silly to make documents without a DTD invalid, since having a DTD is optional
20:32
<Philip`>
Is there any reason to prefer <html dir=rtl> over/under <style>html { direction: rtl }</style> ?
20:32
<Darxus>
Wow, validator.w3.org won't do just xml?
20:35
<annevk2>
Philip`, style sheets are optional
20:36
<ttepass->
Non-CSS-user agents one would think. Like a search engine which want's to display the first n sentences and needs directionality for display.
20:37
<cying>
dear whatwg
20:37
<cying>
when a video element is appended to the DOM, should it trigger a load event?
20:39
<gsnedders>
cying: A load event on what?
20:39
<cying>
gsnedders: on the video element
20:39
<gsnedders>
cying: That'll fire once the video is loaded, no?
20:39
<cying>
http://satine.org/research/webkit/tests/video-events.html
20:40
<Rik|work>
cying: how does it behave on firefox ?
20:40
<cying>
Rik|work: ah i should try that
21:26
<Darxus>
Parse::Simple is upsetting me: "Entity 'lt' not defined".