00:00
<Hixie>
precisely to stop this kind of argument :-)
00:00
<annevk2>
Hixie, I use that!
00:00
<Hixie>
no, you don't. As an author you do, but not as a user.
00:00
<Hixie>
and as an author i bet it's never done you one iota of good.
00:00
<tantek>
e.g. <cite id="qs01">tantek</cite> said <q cite="#qs01">which can be connected to a quotation q through its cite attribute</q>
00:00
<Hixie>
tantek: you don't need any markup whatsoever for that. Just write tantek said "which can be connected to a quotation q through its cite attribute
00:00
<Hixie>
"
00:00
<ttepasse>
annevk2, how is that a problem? In a non-CSS-world - but where ist such a thing?
00:01
<annevk2>
Hixie, I have some script that displays it to the user
00:01
<annevk2>
ttepasse, cite { font-style:normal } would not be good if it was actually the title of a work
00:01
<ttepasse>
cite.name?
00:01
<annevk2>
ttepasse, too much effort?
00:02
<Hixie>
annevk2: and i'm sure it made you feel great to know that you're doing the semantically right thing, huh
00:02
<annevk2>
(this is also the problem I have with the role= rather than element advocates; it's just too much typing most of the time to bother)
00:03
ttepasse
is a big fan of tab-completion for those things.
00:03
<annevk2>
Hixie, yeah and if you kill the feature I'll set up my own rogue cite4all group and make a lot of noise!
00:04
<annevk2>
ttepasse, my text editor is akin to notepad / <textarea>
00:05
<Hixie>
so, anyone want to volunteer to write a script that scans the spec to generate these tables? :-) http://www.whatwg.org/specs/web-apps/current-work/#index
00:21
<tantek>
Hixie, your point about "don't need any markup ... for that" again depends on NLP/entity-recognition, which as noted, is non-trivial in one language, let alone internationally.
00:21
<tantek>
Regarding the harm "incorrect typography for names" - I will accept that issue and get back to you with appropriate default style sheet rules.
00:21
<Hixie>
humans are good at NLP
00:21
<Hixie>
very good
00:22
<tantek>
Regarding "show me one example of that confusion actually happening, where <cite> would actually have helped." I will research and provide more examples.
00:23
<tantek>
Hixie, semantic markup is for UAs to present better user interfaces though right? If all we needed was to present things for humans, plain text with appropriate whitespace would be sufficient for many/most use cases.
00:24
<rubys>
Hixie: my experience is otherwise.
00:24
<Hixie>
tantek: i'm not convinced there's any need for better UI for users to work out who is saying what, to be honest.
00:24
<tantek>
Anyway - your request for more examples is certainly very reasonable.
00:24
<tantek>
Hixie, as a sample UI, being able to right-click on a source and get info on it/them.
00:25
<Hixie>
who does that? that's a dream of Semantic Web people that in my experience "normal" users don't share
00:25
<tantek>
Indeed in Mozilla (and perhaps Opera?) I believe it was/is possible to right click on a q/blockquote and get the cite attribute(d) source.
00:25
<Hixie>
rubys: literacy rates might be better in europe :-)
00:26
<tantek>
from there, when that cite attribute links (through a frag id as shown in the example above) to a <cite> element, then that can provide further information
00:26
<Hixie>
i encourage you to consider what real problem users are facing would be addressed by this, rather than what theoretical problems might be addressed by this
00:27
<Hixie>
i've never, ever, ever heard a user say "i wish i could figure out who wrote that quote... maybe if i right-click on it it will highlight the name of the author for me"
00:27
<Hixie>
i have heard them say things like "i wish i could read my web mail while on a plane without network connectivity"
00:28
<tantek>
Hixie, one real world usage is Wikipedia, where quotes/assertions are cited, and hyperlinks are used to create a UI from quote/assertion to source etc.
00:28
<Hixie>
what's wrong with what they do now?
00:29
<tantek>
nothing, in the same way that <br><br> was sufficient to indicate paragraphs. but now just as we have the <p> element for semantically marking up paragraphs so that we can avoid "hacking" a presentational paragraph, it makes sense to *keep* the cite attribute / cite element to semantically markup citations and sources
00:30
<Hixie>
<p> gives true styling benefits
00:30
<tantek>
and "heard them say" is not the best way of interpreting user needs. watching/analyzing user behaviors (e.g. in content publishing) is often better than simply listening to what they say they think they need.
00:31
<Hixie>
if your proposed variant of <cite> solves real problems, this would change my reaction, but currently i'm not aware of any real problems that would be solved by making <cite> cover citing names of people
00:31
<Hixie>
tantek: totally agreed re: user needs
00:31
<Hixie>
tantek: i've never seen a usability study where anything like citations on quoted has come up
00:31
<tantek>
hence I mentioned content publishing behavior
00:32
<Hixie>
(and i've seen a lot of them, i'm on google's usability lab access list so that i can keep an eye on this very kind of thing)
00:32
<tantek>
user studies may be sufficient, but are not necessarily necessary to make such a case
00:38
<tantek>
Hixie, while I'm not sure I can necessarily meet your threshold for what you call a "real" problem, I'm fairly certain I can provide real world use cases with examples, and I'll at least do that.
00:38
<Hixie>
cool, that would certainly help
00:40
<tantek>
I think with sufficient real world use cases (with examples), at some point the advantage of keeping functionality which is allowed by (the admittedly ambiguously broad) wording of the existing HTML 4.01 spec is greater than the minor advantage of simplification/reduction that you gain by removing/dropping that functionality.
00:40
<tantek>
I understand that is a judgment call, and that reasonable people can come to different conclusions based on similar facts on such a point.
00:40
<tantek>
I hope by providing more data I can narrow any such gaps in conclusions at least a bit.
00:43
<tantek>
And for that matter, let me just say that I do commend your general effort at simplification/reduction, lest that is not clear by the arguments I've made above.
00:43
<Hixie>
fwiw, what html4 says has no bearing on what html5 says -- i only base the spec on legacy content and legacy UAs, not legacy specs.
00:44
<Hixie>
and don't worry, i don't take any offence at all from our discussions, i know your heart is in the same place as mine :-)
00:45
<tantek>
understood. I'm looking at the legacy content part of that, which, often reflects what html4.01 says, especially as interpreted by the broader web design / semantic HTML community.
00:45
<Hixie>
the broader web design / semantic HTML community is sadly a tiny minority of the wider web :-(
00:45
<Hixie>
from what i've seen, <cite> is almost always used to mean "italics"
00:45
<Hixie>
i'm willing to ignore that on the basis that that was never a valid use
00:46
<Hixie>
never even remotely a valid use
00:47
<Hixie>
i agree that some people mark up names using <cite>, as well as marking up titles, and some people (me, circa 2004) go out of their way to only mark up citations with <cite>
00:47
<Hixie>
so there is definitely use to back up all of these options
00:47
<Hixie>
this leaves the question of what is the most useful use we can put the element to
00:48
<Hixie>
and my conclusion so far is that the most useful use for <cite> is as an element to allow typographic control over titles, since those are often made italics
00:48
<Hixie>
and generally, names and titles aren't typeset the same way, so i am reluctant to make the element apply to both
00:48
<Hixie>
especially since we have so many ways of marking up names already (e.g. hCard, microdata vCard, RDFa vCard, etc)
00:49
nathanhammond_
patiently waits for the conversation about cite to come to a conclusion (2 hours now)
00:49
<Hixie>
hehe
00:50
<Hixie>
it's been much longer than 2 hours :-)
00:50
<Hixie>
we were talking abotu this in like 2005 already
00:50
<nathanhammond_>
okay, so just two hours today. :)
00:50
<Hixie>
:-)
00:51
<nathanhammond_>
at some point I've got a few comments and questions about some of the history management stuff in the spec
00:52
<nathanhammond_>
better to document it all and send it your way? or just converse here?
00:53
<tantek>
Hixie - indeed: http://tantek.com/presentations/2005/09/elements-of-xhtml/#slide32
00:55
<Hixie>
nathanhammond_: i'm going offline in 2 minutes, but i'll be back in a few hours. If it's comments on problems in the spec, best to send the feedback to the list.
00:55
<Hixie>
nathanhammond_: if it's just questions about how it works, I can reply when I get back, and IRC is fine. Also, feel free to interrupt ongoing conversations if there is one in progress when you come here. :-)
00:56
<nathanhammond_>
arrr, it be related to problems
00:56
<nathanhammond_>
so it'll go to the list
00:56
<nathanhammond_>
but I'll probably document my concerns on a blog so that I can just hit the highlights on the list
00:56
<nathanhammond_>
and link for the full conversation
00:57
<nathanhammond_>
thank you sir.
00:58
<Hixie>
please send anything you want me to read to the list, so i can reply to it easily
00:58
<Hixie>
it's hard to reply to blogs in pine :-)
00:58
<Hixie>
ok gotta go
00:58
<Hixie>
bbl
00:58
<nathanhammond_>
true
01:15
<othermaciej>
Hixie: I think the "installed apps" thread has not yet led to either a good statement of use cases, or a mechanism that actually has reasonable security characteristics
01:16
<othermaciej>
Hixie: but it might be "good" in the sense that there is likely no spec action required
01:38
<othermaciej>
annevk: I wonder how the XHTML2 folks plan to continue outside the W3C given the W3C's trademark on "XHTML" and standard document copyright
01:38
<othermaciej>
annevk: I guess they'll have to rename it and rewrite the spec from scratch...
03:45
<nathanhammond>
Hixie: you've got your email
03:46
<nathanhammond>
I'll stick around here in case I can help clarify some of what I said.
07:24
<jm2k>
If you start a Web Worker on a page but then navigate to another page or domain (in same tab), does the web worker continue to compute?
07:30
<jm2k>
eh, no one here :(
07:37
<inimino>
jm2k: no, see http://www.whatwg.org/specs/web-workers/current-work/#the-worker%27s-lifetime
11:58
<zcorpan>
Hixie: how does the spec handle document.documentElement.onclick=function(){alert(1)};document.documentElement.onclick=null
11:59
<zcorpan>
Hixie: the second assignment should cause the first event listener to be removed
12:00
<zcorpan>
Hixie: but in the spec i only see that a new event listener is to be registered
12:24
<MikeSmith>
zcorpan: about http://bugzilla.validator.nu/show_bug.cgi?id=624
12:25
<MikeSmith>
from looking through the v.nu code briefly, I think maybe that error message is not generated by v.nu code, but instead by URI-checking code in third-party library code
12:25
<annevk>
validator.nu should have a URL implementation :)
12:26
<zcorpan>
yeah i guess v.nu needs to know the URL encoding thing
12:26
<gsnedders|work>
I guess all URLs have a valid URL representation
12:26
<MikeSmith>
it may be that Henri's build patches some of the library code already, I don't know
12:27
<gsnedders|work>
As you can just use a URI
12:28
<zcorpan>
MikeSmith: i guess the simplest path is to update the microsyntaxes wiki page
12:30
<MikeSmith>
zcorpan: yeah
12:30
<zcorpan>
done
12:40
gsnedders|work
discovers zcorpan is better at spelling than him
12:40
<gsnedders|work>
(in English, this is)
12:50
<MikeSmith>
zcorpan: given that the error message has the term "whitespace", I'm wondering if the microsyntax page wording might better be, "Whitespace (spaces) should be escaped as %20."
12:51
<MikeSmith>
or something
12:56
<zcorpan>
MikeSmith: it's a wiki :)
12:57
<MikeSmith>
zcorpan: well, I was asking for your opinion about whether you think it'd be more clear, or less clear?
12:58
<zcorpan>
MikeSmith: maybe make it more general and give a specific example... Special characters should be URL-escaped, e.g. %20 for whitespace
12:59
<MikeSmith>
yeah
13:00
<MikeSmith>
maybe we could ask dude what he thinks would have been a helpful error message for his case
13:00
<MikeSmith>
they guy who tweeted about it, I mean
13:00
<MikeSmith>
*the
13:03
<zcorpan>
also see http://www.tomleadbetter.co.uk/blog/entry/html5/
13:06
<MikeSmith>
the _URL-escaped_ part could be a hyperlink
13:27
<MikeSmith>
anyway, I think in general the way to get better error messages from v.nu is not to mess too much with the code (e.g., jing), but to supplement with refinements to the microsyntax stuff and other stuff in the wiki, and to put more into the code for the assertions-checking phase
13:40
<Philip`>
http://h1debate.com/
13:40
Philip`
thinks it kind of misses the idea of a debate, vs an opinion poll
13:41
<MikeSmith>
what the hell
13:41
<MikeSmith>
I can7t tell what the debate is
13:42
<MikeSmith>
where's the question?
13:42
<Philip`>
See the column headings
13:42
<annevk>
Hixie, I thought of more IANA business
13:42
<annevk>
re: http://krijnhoetmer.nl/irc-logs/whatwg/20090701#l-557
13:43
<annevk>
the form submission formats
13:44
<annevk>
application/x-www-form-urlencoded and multipart/form-data
13:44
<annevk>
(it was brought up in the HTTP WG meeting)
13:50
<annevk>
application/x-www-form-urlencoded seems to be nowhere defined
13:50
<annevk>
the other is http://tools.ietf.org/html/rfc2388
13:55
<annevk>
oh duh, HTML5 references rfc2388
13:55
<annevk>
but application/x-www-form-urlencoded would still need to be defined
14:12
<annevk>
sebmarkbage, interesting idea
14:15
<sebmarkbage>
annevk: yea. Downloading might be specified too. So that no browser downloads the whole archive several times. It should be loaded once for atleast the life time of the document.
14:17
gsnedders
fails to find the idea in the logs
14:18
<gsnedders>
Oh, sebmarkbage = Sebastian Markbåge?
14:18
<sebmarkbage>
yes
14:18
<annevk>
life time might be tricky to define
14:18
<gsnedders>
sebmarkbage: I think HTTP pipelining over comes the majority of issues
14:19
<annevk>
and you probably want a custom protocol to make it work so you can have fragment identifiers for referenced resources as well
14:19
<annevk>
gsnedders, yeah
14:19
<annevk>
gsnedders, though you'd need a new version of that that isn't broken by server implementations :/
14:20
<gsnedders>
annevk: We do use it sometimes, though, albeit with apparently crazy heuristics to decide whether to enable it
14:21
<gsnedders>
annevk: What servers actually break with it, nowadays?
14:21
<annevk>
i know
14:21
<annevk>
i don't know
14:22
<gsnedders>
Apparently broken with IIS4 and 5
14:26
<Philip`>
I think I remember people saying proxies were a problem for pipelining, more than servers
14:26
<gsnedders>
http://groups.google.com/group/comp.infosystems.www.browsers.ms-windows/msg/6766229fd0b610ca?hl=en
14:27
<sebmarkbage>
yea, it's troublesome for high performance servers too.
14:27
<Philip`>
which means you can't just ask the server whether it's non-buggy
14:27
<sebmarkbage>
But there's an additional gain by packaging it all together - You can gzip the content as a single unit.
14:29
<gsnedders>
https://bugzilla.mozilla.org/show_bug.cgi?id=264354
15:37
<annevk>
Why was < in an attribute name made a parse error? It would be non-conforming anyway...
15:45
<gsnedders>
Philip`: http://logand.com/sw/wps/index.html
15:45
<gsnedders>
Philip`: We were joking about doing that a few days ago
15:45
<gsnedders>
mookid: ^^
16:24
<webben>
annevk or annevk2: re http://lists.w3.org/Archives/Public/public-html/2009Jul/0872.html and "if we add APIs to <canvas> that make it do the same as SVG"
16:24
<webben>
it seems that some canvas UI developers don't regard being able to manipulate an accessibility API directly as anything like the same as having to build a DOM
16:25
<webben>
The guys behind Cappuchino took a close look at ARIA, and largely rejected it on the basis that it requires DOM manipulation.
16:25
<annevk>
right
16:26
<webben>
If there's a role for bolt-ons of this sort, it's likely to back widget systems like Cappuchino.
16:26
<annevk>
turning <canvas> into a DOM-based model would not make them happy
16:26
<webben>
I thought Rich was talking about an accessibility API model rather than a DOM model.
16:26
<webben>
which seems to match what the Cappuchino guys were asking for
16:27
<webben>
http://rossboucher.com/2009/03/01/limitations-of-the-wai-aria/
16:27
<webben>
is the post I'm thinking of.
16:27
<annevk>
thanks
16:27
<annevk>
Rich was talking about an object model as well as an accessibility API as far as I can tell
16:27
<webben>
okay
16:28
<annevk>
but yeah, maybe we do need an API
16:28
<webben>
I'm not sure of the practical differences between a DOM and accessibility API manipulation (I've not done any desktop GUI programming).
16:28
<annevk>
I just don't think we should rush into it
16:28
<annevk>
I don't want another ARIA
16:30
<Dashiva>
Darxus: Irony, idioms, implications, etc remain in any language
16:30
<annevk>
webben, btw, "As far as Cappuccino implementation is concerned, there is enough in ARIA that we can significantly enhance the accessibility of Cappuccino, if not 100%."
16:30
<webben>
annevk: I think that reflects the fact that Cappuchino isn't entirely Canvas.
16:31
<webben>
there's a lot of divitis and spanitis "controls" in there too.
16:31
<webben>
so there's already DOM hooks for some things
16:31
<annevk>
as I said in my latest email, entirely <canvas> is not what <canvas> is meant for
16:31
<webben>
Sure.
16:31
<annevk>
it's an indication of other deeper flaws that need fixing
16:32
<annevk>
and his example of SVG is bogus afaict, since SVG does have a DOM
16:32
<annevk>
so that leaves Bespin
16:32
<webben>
yep
16:32
<webben>
i doubt Bespin will be the last tho.
16:34
<annevk>
I'll counter with premature optimization
16:36
<annevk>
webben, you have a point though and I hope you keeping dropping by to make them ;) (don't want to be entirely dismissive)
16:45
<webben>
annevk: will do :)
16:48
<annevk>
ffs, Pirate Bay is to be banned in the Netherlands
16:54
<Dashiva>
annevk: So what? It's going to die as soon as the sale goes through anyhow.
16:54
<Lachy>
annevk, not to worry, if it gets sold, TPB will be dead in a month anyway
16:56
<annevk>
it's the principle
16:56
<annevk>
also, the main guy in charge of the sale apparently quit and there's some rumor over lack of sufficient money
16:56
<annevk>
in any case, I don't like it that this is possible at all
16:56
<annevk>
I don't mind the consequences so much
16:57
<Lachy>
yeah, from what I've read about it, your legal system has been abused by BREIN
16:58
<zcorpan>
annevk: Element.children is in the IDL in web dom core at least
17:03
<annevk>
gsnedders, ^^
17:03
<annevk>
Lachy, BRAIN sucks donkey balls
17:03
<annevk>
really
17:03
<annevk>
BREIN*
17:18
annevk
wonders if Leif realizes that </ no longer needs to be escaped because we removed SGML cruft, not because we kept it around...
17:20
<annevk>
then again, he compares JavaScript and PHP as if they are the same and does not see the logic in treating one as a first-class Web citizen and the other as yet another server-side templating language...
17:28
<Dashiva>
I have problems taking the proposal seriously myself
17:34
<Darxus>
Dashiva: Yes but Lajlan has stuff to indicate what is an irony, idiom, etc..
17:35
<Darxus>
Lojban, sorry. I keep getting it confused with the spelling of its predecessor.
17:36
<Darxus>
"pe'a" "marks a construct as figurative (non-literal/metaphorical) speech/text
17:36
<Darxus>
"
17:37
<Philip`>
Hooray for "begs the question" vs "raises the question"
17:38
<Philip`>
particularly when someone actually uses the former correctly (as far as I can tell), and then gets told off for doing so
17:47
<Dashiva>
Darxus: That's same level
17:47
<Dashiva>
But you can still use real irony, without the marker
18:02
<annevk>
oh crap
18:02
<annevk>
dates discussion
19:51
<gsnedders_>
hmm… -0.01% idle CPU time.
19:58
<sebmarkbage>
annevk: About packages. It may be better to supply it as an alternative download for compatibility. <img src="file.jpg" psrc="archive.tgz#file.jpg" /> { background-image: url(file.jpg); background-image: purl(archive.tgz#file.jpg); } I don't think it needs it's own protocol/scheme.
20:04
<annevk>
well, dunno if they're really needed (see HTTP pipelining which would be much neater) and that syntax does not solve the issue I raised
20:11
<sebmarkbage>
annevk: HTTP pipelining needs major fixes to make that viable. Timeout issues for servers, broken servers, broken proxies... And you don't get the added benefit of compressing the traffic as a unit.
20:12
<annevk>
is that a major benefit?
20:12
<sebmarkbage>
What issue did you raise? Fragment identifiers? The multipart, tar, etc. already has identifiers.
20:12
<annevk>
yes, fragment identifiers
20:14
<sebmarkbage>
On text data, at least, it's a major benefit to compress multiple small parts as a single unit. Hence .tar.gz and not .gz.tar
20:19
<sebmarkbage>
annevk: As I see it, there's nothing in the URI specs that would prevent one from using the fragment to identify a part of a multipart document. Of course it needs to be speced. But there's no need to define an additional protocol or URI scheme.
20:24
<sebmarkbage>
rfc2854 for text/html, rfc5147 for text/plain, rfcXXXX for multipart/mixed
20:42
<annevk>
sebmarkbage, the issue is how I'd use a fragment identifier for the document then
20:48
<sebmarkbage>
annevk: Well that would depend on the format chosen. application/tar obviously has filenames. multipart/form-data has a name attribute in Content-Disposition. multipart/related has Content-ID...
20:54
<Philip`>
annevk: Are you able to SSH to Dreamhost? (I get "ssh_exchange_identification: Connection closed by remote host" whenever I try on hobgoblin)
20:57
<annevk>
me too
20:58
<Philip`>
Okay
21:00
<annevk>
sebmarkbage, then it doesn't work
21:01
<annevk>
Philip`, not sure why it doesn't work
21:02
<annevk>
Philip`, if it still doesn't tomorrow I'll ask DreamHost
21:02
<annevk>
(unless you're in a hurry then I can do it in a few minutes)
21:02
<sebmarkbage>
annevk, if one format is chosen as a standard for something like psrc, I don't see why. All the parts are already there. All you'd have to do is specify how the fragment part is suppose to be handled for mime type: multipart/x
21:05
<Philip`>
annevk: No hurry :-)
21:07
<annevk>
sebmarkbage, say for an SVG file I need to use a fragment identifier to get a particular effect; as in /example.svg#view... (forgot the syntax); how do I express that if example.svg is packaged?
21:12
<sebmarkbage>
annevk, rfc3986 (generic uri) allows for the possibility of multiple #, so: /archive.tgz#example.svg#view...
21:13
<sebmarkbage>
if the particular format chosen allows for # in the identifier, I guess it'd be encoded