00:00
<TabAtkins>
Then why did you state something to that effect?
00:02
<Hixie>
i didn't
00:02
<Hixie>
my position was more subtle than that
00:03
<TabAtkins>
Obviously too subtle. ^_^
00:03
<Hixie>
my point is that if the user clears cookies, the user agent should suggest to the user that if the goal is to prevent user tracking, that clearing the local storage, session storage, web database, flash storage, cache, and other such mechanisms is necessary to ensure the user's privacy
00:04
<Hixie>
how this happens is up to the ua
00:04
<Hixie>
whether it happens by just treating them all the same, or by having a complicated six-page explanation in a modal dialog, is up to the UI designer
00:07
<SamerZ>
what does the "I've read the changes!" button do?
00:07
<SamerZ>
on the tracker
00:07
<TabAtkins>
What you actually *said* was: "we have to make sure that clearing one clears the other".
00:07
<TabAtkins>
You did *not* say "We have to give users the option to clear both at once."
00:08
<Hixie>
clearing both at once automatically would be ideal, imho, yes
00:08
<TabAtkins>
Your words imply that it's a requirement that they are cleared together, rather than an option.
00:08
<Hixie>
but what _i_ want isn't relevant here
00:08
<Hixie>
as it's not what the spec requires
00:08
<TabAtkins>
Indeed. As it is, browsers can *choose* to make LocalStorage completely ephemeral, so we can't rely on it.
00:09
<Hixie>
they can also "choose" to make every click result in the computer exploding, but that doesn't mean we can't use <a href="">.
00:10
<TabAtkins>
Sure, but no one's suggesting that behavior is vitally important for the user's privacy. ("No one can violate your privacy if the computer's blown up!")
00:11
<TabAtkins>
On the other hand, *you* are suggesting precisely that wrt LocalStorage.
00:13
<Hixie>
i also think we should drop style="" and <div>, and that we shouldn't even allow xmlns="" in text/html. So?
00:21
<AryehGregor>
Dropping <div> without much better CSS support would be kind of insane.
00:21
<AryehGregor>
Possibly even with unreasonably good CSS support.
00:21
<Hixie>
like i said, what i want doesn't have much relevance here :-)
00:22
<AryehGregor>
How would you mark up the <h1> on http://aryeh.name/ without <span>? Maybe an <hgroup> with a few headings?
00:23
<AryehGregor>
With HTML 5, you can come pretty close to eliminating <div> and <span>, it's true . . . at least for fairly simple stuff.
00:23
<Hixie>
<span> is fine
00:29
<AryehGregor>
That's an interesting position. Why is <span> fine but <div> not? Do blocks inherently have more restricted semantics or something?
00:30
<Hixie>
<span> is fine if used with lang="" or dir="" or the like.
00:30
<Hixie>
my opinions on this aren't very logical, i should point out
00:31
<Hixie>
that's why the spec doesn't do what i want :-)
00:45
<AryehGregor>
Hixie, since when is speccing the result of invalid markup condoning it?
00:54
<Hixie>
AryehGregor: wasn't the request to allow it? The spec already says what the result is.
00:55
<AryehGregor>
Ah, okay then, never mind.
00:55
<AryehGregor>
I'm not an implementor, so thankfully I don't have to look at the parsing section very closely. :)
00:56
<Hixie>
:-)
00:57
<MikeSmith>
AryehGregor: I'm looking at http://bugzilla.validator.nu/show_bug.cgi?id=640 and may have a fix for it shortly
00:57
<AryehGregor>
MikeSmith, nice, thanks.
04:25
<amstan>
hello everyone
04:26
<amstan>
have you guys saw this? pretty cool and valid post:
04:26
<amstan>
http://www.webmonkey.com/blog/This_Spec_Has_a_Posse%3A__Superfriends__Suggest_Changes_for_HTML5
04:59
<tantekc>
amstan - thanks for the URL
06:01
<MFen>
html5lib for python.. anyone in here connected to that?
06:05
<MFen>
how would i serialize a generic HTMLParser().parse() document instance?
06:05
<MFen>
HTMLSerializer apparently only works on this treebuilder thing which i don't want
06:06
<MFen>
and can i not just use it to parse xml? it seems to wrap an html document around everything
06:15
<MFen>
never mind. all it does is reproduce the broken api of some other thing
06:15
<MFen>
do i want no namespace support, no unicode support, or random_broken_stuff? i have the power to choose.
07:08
<othermaciej>
hi everyone
08:09
<hsivonen>
cursed Unicode normalizations! apparently I wont be able to nuke files on a Linux server over NFS from Mac OS X when the file names differ in NFC and NFD!
08:13
<hsivonen>
I wonder whose bright idea it was to make Mac OS X use something other than NFC for file names
08:13
<Dashiva>
Think different
08:15
<hsivonen>
Apple is violating the first rule of Unicode normalization
08:15
<hsivonen>
(always use NFC when you write to disk or to a socket)
08:18
<jgraham>
Something something, fight club reference, something
08:26
<Dashiva>
hsivonen: When do you use NFD?
08:28
<hsivonen>
Dashiva: when Mac OS X stringifies a file name, it converts it to an approximation of NFD
08:29
<hsivonen>
Dashiva: IIRC, it's NFD except NFC for Hangul
08:29
<hsivonen>
Dashiva: Apple just made stuff up there. it's not exactly NFD!
08:29
<Dashiva>
No, I mean, when should you use NFD?
08:30
<hsivonen>
Dashiva: in RAM when doing diacritic-insensitive searches
08:30
<Dashiva>
Oh, okay
08:46
<hsivonen>
AryehGregor: thanks for fixing the Wikipedia article on XHTML
09:35
<zcorpan_>
AryehGregor: send feedback about = in unquoted attributes
09:36
<zcorpan_>
AryehGregor: it's only disallowed to catch this authoring error: <p class= title=hello>
09:37
<zcorpan_>
AryehGregor: there are no compat issues with it
09:38
<zcorpan_>
AryehGregor: also see http://code.google.com/p/html5lib/issues/detail?id=92
09:39
<zcorpan_>
AryehGregor: and http://code.google.com/p/html5lib/issues/detail?id=93
10:08
zcorpan_
witnesses another real-world case of <article><header></header><section>...</section><footer></footer></article>
10:08
<zcorpan_>
and headings in the page <footer>
10:13
<zcorpan_>
actually maybe we should drop <section> and rely on implied sections
10:13
<zcorpan_>
it would still do the right thing when you put content in <article> or <blockquote>
10:14
<zcorpan_>
people are clearly overusing sectioning elements in general and clutter the outline
10:15
<Hixie>
i think to some extent it's overcompensation
10:15
<Lachy>
maybe we make <section> without a <header>, <hgroup> or <h1> to <h6> invalid?
10:15
<Hixie>
we should just get out there with a bunch of articles talking about <article> and using a <div> for the main content
10:15
<Lachy>
Hixie, I'm in the process of writing an article about that
10:15
<zcorpan_>
Hixie: i really don't think evang will fix this problem
10:16
<zcorpan_>
Hixie: it's a problem with the spec; people are using the new elements the way they think they should be used, and the result is bad
10:16
<Hixie>
lachy: making a <section> without header invalid would break all kinds of use cases
10:16
<Hixie>
Lachy: cool, i look forward to it
10:17
<Lachy>
I really don't get how people are reaching incorrect conclusions about how they're meant to be used, since the structure was based on how people frequently used <div><h3>...</h3><p>...</div>
10:17
<Hixie>
zcorpan_: evang is what introduced this problem, i don't know why it wouldn't solve it
10:17
<Lachy>
what are the use cases for section without it's own heading? I've been struggling to think of any
10:17
<zcorpan_>
Hixie: what evang introduced this problem?
10:17
<Hixie>
open any random blog, there's all kinds of bits of the page that are sections without headings
10:18
<Hixie>
zcorpan_: "html5 has <article>!"
10:19
<Lachy>
so it seems that people aren't realising that both <article> is just a special class of <section>, and struggling to realise how to use both together properly
10:19
<Lachy>
s/that both/that/
10:19
<zcorpan_>
maybe it should be <section type=article>
10:19
<Lachy>
that might work
10:20
<Lachy>
but makes styling in IE even harder without attribute selectors
10:20
<Hixie>
how hard can this be, really
10:20
<Hixie>
maybe i should rename <article> to <blogpost>
10:20
<Lachy>
Hixie, I didn't think it was hard
10:20
<Lachy>
I don't think renaming it would help
10:20
<Lachy>
since it's a structural problem, not a semantic problem
10:20
<zcorpan_>
Hixie: in this case, it was used for blog posts correctly
10:20
<zcorpan_>
Hixie: it's just <section> inside that's incorrectly used
10:21
<Hixie>
oh, then why <section type="">?
10:21
<zcorpan_>
Hixie: i.e. he wanted a <main> element
10:21
<Hixie>
i never thought we'd manage to wean people off <div> so fast
10:21
<Hixie>
i'm getting whiplash
10:22
<zcorpan_>
because <section type=article><h2/><section><p> *looks* wrong whereas <article><h2/><section><p> doesn't
10:22
<zcorpan_>
people don't realize <article> is a type of <section>
10:22
<zcorpan_>
<section type=article> would make it clearer
10:22
<Hixie>
that isn't the confusion
10:22
<Hixie>
if it was they'd be doing <article><section><h2/><p>...
10:23
<Hixie>
but they're putting the header in the article
10:23
<Hixie>
and then just putting the contents in a section
10:23
<Hixie>
it _has_ to be for styling, surely
10:23
<Hixie>
no?
10:24
<zcorpan_>
no
10:24
<zcorpan_>
it's not styled in any way
10:24
<zcorpan_>
no selector in the style sheet is based on the section
10:24
<zcorpan_>
it's just that they think it's the right thing to do
10:25
<annevk2>
funny, I think we should drop <article>
10:25
<Hixie>
i wish i understood how authors were coming to this conclusion
10:25
<Hixie>
is there some documentation somewhere that they're all following?
10:25
<jgraham>
I wonder what the "many use cases" are for untitled sections. Like how to they translate to UA features?
10:25
<Hixie>
is this problem really widespread?
10:25
<Hixie>
someone should study a random sample of sites from html5gallery
10:26
<zcorpan_>
Hixie: i've seen it several times with people trying to use html5
10:26
<Hixie>
anecdotal evidence is often misleading
10:26
<Hixie>
i think we need hard data on this
10:26
<jgraham>
BTW I already tried the evangalism thing. Whilst the WHATWG blog might not be the best pulpit you might have thought it would reach early adopters
10:26
<jgraham>
AT least indirectly
10:27
<Lachy>
html5doctor and ALA seem to be among the best places to reach early adopters
10:27
<zcorpan_>
html5doctor did the exact same mistake
10:27
<Lachy>
I'm planning to get my sectioning article published on either or both of those places
10:28
<Lachy>
zcorpan_, yeah, I know. I suspect that might be where develoeprs are learning it from
10:28
<zcorpan_>
this guy sais he sees the structure of <article> the same as <table>
10:28
<Lachy>
which guy?
10:29
<zcorpan_>
a friend
10:29
<jgraham>
I think part of the fix has to be making UAs robust to failure. That means ignoring untitled sections when calculating nesting depth
10:30
<Lachy>
the question is, do they want the structure to be like a table, or is that just how he perceives it was designed?
10:30
<zcorpan_>
he says that since he framed the content in a <header> part and a <footer> part, it felt natural to frame the content part in some element too
10:30
<zcorpan_>
author of http://www.frojd.se/
10:30
<Hixie>
sigh maybe we should just give people a meaningless <content> element
10:31
<annevk2>
s/article/main/ plus the appropriate changes to its semantics
10:31
<jgraham>
Hixie: I think that is a good idea.
10:31
<Hixie>
or heck, maybe the simplest solution is just to special-case a heading-less section in the outline as a direct child of another section
10:31
<Hixie>
that actually might be the better solution
10:31
<Hixie>
since if we introduce another element, it's bound to get misused even more
10:32
<zcorpan_>
makes sense
10:32
<jgraham>
I think there is a balance in the design process between things that are designed to transmit semantics and things that make authoring easy
10:32
<jgraham>
<content> would make authoring easier
10:33
jgraham
doesn't like the idea os special cases
10:33
<Hixie>
well it wouldn't make it any easier than an ignored <section>, right?
10:33
<Hixie>
how do we get people to use <content> instead of <section>?
10:33
<annevk2>
content { ... } section { ... }
10:34
<zcorpan_>
we say <content> is the elemetn to use between <header> and <footer>
10:34
<jgraham>
Hixie: It would because people seem to have a mental model of their content that splits it into distinct logical areas; "header", "content", "sidebar", "footer"
10:34
<Hixie>
i think we need hard data on how much of an issue this is, if we're going to add another element. we definitely don't want to be doing something quite as drastic if it's not a real problem.
10:34
<Hixie>
jgraham: pity we can't reuse <body>
10:35
<Hixie>
<sbody> (section body)
10:35
<annevk2>
Hixie, I think another look at some URLs of your 2005 webstats study could be sufficient
10:35
<Lachy>
Hixie, if we use the name <content>, then is the clash with the <content> element in XBL2 problematic as was suggested on the list?
10:35
<Hixie>
annevk2: i think we should take a random sample of html5gallery pages
10:35
<Hixie>
annevk2: and see how many do this wrong
10:36
<annevk2>
that also seems like a good idea, but checking what class=content is used for given http://code.google.com/webstats/2005-12/classes.html would also give a clue
10:36
<Hixie>
Lachy: yeah <content> is probably suboptimal.
10:36
<Hixie>
i kinda like <sbody>
10:36
<annevk2>
or class=main for that matter
10:36
<Hixie>
annevk2: i'm more interested in how big a problem it is, rather than how many sites used <div>s in this way
10:37
<zcorpan_>
i think <main> is a better name than <sbody> (unless header and footer are renamed to <shead> and <sfoot>)
10:37
<jgraham>
<sbody> isn't going to work. It will _really_ help if it is a simple english word that means "content" or "main"
10:37
<Hixie>
<main> doesn't work since there could be one per section
10:37
<jgraham>
I don't think that matters
10:37
<zcorpan_>
i think <content> is a better name than <sbody> too
10:38
<annevk2>
I'm not sure allowing <header> / <footer> anywhere else than <body> is a good idea either
10:38
<jgraham>
annevk2: I don't see why
10:38
<jgraham>
Why can't a blog post have a header?
10:38
<jgraham>
I agree that the <footer> content model should be changed to be the same as header
10:38
<Lachy>
annevk2, it is, because allowing sections to have additional introductory content in addition to the heading is useful
10:39
<jgraham>
In practice they are almost interchangable in the type of content they can contain
10:39
<Hixie>
<footer>'s content model has been the same as <header>'s for at least 30 minutes now
10:39
<zcorpan_>
jgraham: yes; frojd.se also has a "fat footer"
10:39
<Hixie>
keep up :-P
10:39
<zcorpan_>
oh
10:39
<Lachy>
cool
10:39
<Dashiva>
So RDFa doesn't actually use namespaces... yet they chose to use namespace-like attributes?
10:39
<jgraham>
Hixie: :p
10:40
<Dashiva>
Why would you do that
10:40
<annevk2>
to mislead the XML gods
10:40
<annevk2>
apparently it worked
10:40
<annevk2>
jgraham, the header of a section is way different from the header of a page
10:40
<annevk2>
jgraham, same for footer
10:41
<jgraham>
annevk2: Why?
10:42
<Lachy>
annevk2, they will obviously contain different content, but conceptually and structurally, they are analogous
10:42
<Hixie>
ok so plan of attack for this section issue is to look at html5galley, and if it shows more than, say, 50% of a random sample of blog-like pages are doing this wrong, then we introduce a new element that is allowed only as a direct child of sectioning content, and only if the sectioning content only contains <header>, <footer>, this element, and/or heading content. (assuming that makes sense given the errors we see)
10:42
<zcorpan_>
annevk2: maybe the content models of <header> and <footer> should be sensitive to whether they're in <body> or in a nested section?
10:42
<Hixie>
seem reasonable?
10:42
<Hixie>
i think y'all are way too eager to change things, btw. leave <header> and <footer> alone already. they're fine. :-)
10:42
<Hixie>
(now)
10:42
<annevk2>
jgraham, for a page it gives the title of the weblog, tagline, navigation etc. whereas for <section> you would not have something equivalent I think
10:43
<annevk2>
jgraham, maybe for <article> but even there it is quite different imo
10:43
<zcorpan_>
Hixie: we should allow the new element in <body> too
10:43
<Lachy>
zcorpan_, no, making the content models contextually sensitive would be confusing for authors
10:43
<zcorpan_>
Lachy: yeah, probably
10:43
<annevk2>
Hixie, to be frank I'm not sure we studied this problem in as much detail as e.g. the parsing algorithm
10:44
<Lachy>
and it also wouldn't make sense. What exactly would you want to restrict from one or the other?
10:44
<Hixie>
zcorpan_: yes, and <body>.
10:44
<annevk2>
Hixie, which is why I think not everyone is quite so sure about it being ok
10:44
<zcorpan_>
Lachy: i would allow nested sections in body <footer> but not other <footer>s
10:44
<Hixie>
zcorpan_: <article> elements should be able to have colophons
10:45
<zcorpan_>
yeah
10:45
<Hixie>
ok well i'm going to bed now.
10:46
<Hixie>
hopefully the data fairies will do the study on the html5gallery data so i don't have to tomorrow :-P
10:46
<Hixie>
nn
10:47
zcorpan_
is not a data fairy today
10:48
<Lachy>
I will take a look later, no-one else does
10:53
<annevk2>
out of 8, I found 2 sites that used <section> in this way
10:53
<annevk2>
I also found one site that used <header> instead of <hgroup>
10:53
<annevk2>
and one site that used <article> instead of <section>
10:54
<Lachy>
we should drop hgroup anyway. It's unnecessary
10:54
<annevk2>
not sure I agree, but anyway, there are 142 sites and I don't have time right now
10:55
<annevk2>
if anyone starts I suggest a table in a wiki so it's easy for others to help out
10:57
<Lachy>
I've documented more of the polyglot checking issues. http://wiki.whatwg.org/wiki/Validator.nu_Useful_Warning_Requests#Polyglot_Document_Checking
10:57
<Lachy>
If anyone feels motivated to add more, feel free
10:59
<annevk2>
polyglot = boring
10:59
<annevk2>
(except when debated)
11:00
<Lachy>
annevk2, my reason for documenting all of this is so that when the issue is being discussed with people like Tantek and Zeldman who want the validator to check for such issues, we can point them there and get a concrete list of issues they want to have checked
11:03
<jgraham>
Sigh. The RDFa people are really trying to have their namespace cake and eat it too
11:05
<annevk2>
Mark Birbeck still does not understand either a) that we think ignoring the namespace of an attribute is wrong or b) that xmlns:<prefix> is in a namespace in XML
11:05
<annevk2>
And given that it has only been explained a 100 times before I guess that is perfectly reasonable...
11:43
<Lachy>
Hixie, http://www.w3.org/Bugs/Public/show_bug.cgi?id=7483 can now be resolved
12:05
<hsivonen>
Lachy: Making a section with foo invalid would be pain to WYSIWYG editors
12:12
<hsivonen>
I wonder if the RDFa TF was clear with the TAG that RDFa doesn't "use namespaces"
12:13
<Philip`>
getElementsByTagName+lookupNamespaceURI is "not standard DOM methods"?
12:15
jgraham
is bemused
12:15
<hsivonen>
Philip`: context?
12:16
hsivonen
sees it now
12:17
<hsivonen>
Philip`: fwiw, the way lookupNamespaceURI is defined makes it theoretically brittle for RDFa-in-XML
12:17
<hsivonen>
Philip`: in the case of synthetic DOMs
12:18
<Lachy>
hsivonen, yeah, I also realised it wouldn't catch many cases where authors do things wrongly anyway
12:18
<hsivonen>
Philip`: lookupNamespaceURI looks at frozen element node namespace/prefix bindings in addition to looking at namespace declarations
12:18
<Philip`>
(The getNamespaces method seems to be non-standard, but you don't need that to do lookups)
12:18
<hsivonen>
Philip`: W3C specs say the darnest things
12:19
<Lachy>
In what way does RDFa not use namespaces?
12:20
<Lachy>
what's the point of them using xmlns and prefixes if they don't?
12:21
<Philip`>
I guess the point is to be compatible with XML Namespaces syntax (and nothing else matters)
12:21
<hsivonen>
Lachy: http://krijnhoetmer.nl/irc-logs/whatwg/20090904#l-414
12:21
<jgraham>
Lachy: Mark Birbeck's position appears to be that they use namespace syntax but must not be processed in a namespace-aware way
12:22
<Dashiva>
So why use the namespace syntax then?
12:22
<Lachy>
hmm, ok. Is that because they do that weird thing where they just prefix the namespace URI to the localname?
12:22
<Philip`>
jgraham: I don't remember anyone suggesting "must not", just "does not have to (but can if you've got a suitable API; it doesn't matter as long as you get the same result from the same syntax)"
12:23
<jgraham>
Philip`: "The original objection was that different processing is required for
12:23
<jgraham>
different DOMs, and I think we've shown that's not the case; all that
12:23
<jgraham>
is required is to iterate through the list of atttributes, and pull
12:23
<jgraham>
out those that begin "xmlns:"."
12:23
<jgraham>
As far as I can tell the only way to make sense of that would be to forbid using namespace aware processors
12:24
<Philip`>
jgraham: (Some) namespace aware processors let you iterate through the list of attributes and pull out xmlns:*
12:24
<hsivonen>
Lachy: the RDFa TF is programming with namespace-unaware tools (Mark Birbeck: IE, Shane: the perl lib, Ivan Herman: Python minidom)
12:24
<Philip`>
e.g. web browser DOMs do
12:24
<Philip`>
jgraham: so they still satisfy that requirement
12:25
<hsivonen>
Lachy: whereas people here are programming with libraries whose authors have drank the Infoset kool aid
12:25
<hsivonen>
Lachy: hence, diffent world views
12:25
<hsivonen>
*different
12:25
<zcorpan_>
we should see the light and stop caring about namespaces and namespace-aware APIs
12:26
<Dashiva>
I just don't get it. If they aren't using namespaces, why use the one set of attributes that cause trouble?
12:28
<jgraham>
Dashiva: Presumably because it scores brownie points with TAG
12:28
<Philip`>
Dashiva: Because it doesn't cause trouble
12:28
<Philip`>
(when you use namespace-unaware APIs)
12:29
<Dashiva>
jgraham: Nice order of constituencies :)
12:29
<Philip`>
and because it can reuse the normal xmlns processing model, rather than having to define its own that's almost identical but slightly different
12:29
<jgraham>
Philip`: Neither does a mechanism that doesn't cause trouble when you _do_ use namespace aware APIs
12:29
<Philip`>
(and thus avoid reinventing wheels)
12:30
<Dashiva>
processing model?
12:30
<Dashiva>
You mean doing a string split on :?
12:31
jgraham
isn't sure what Philip` means by "xmlns processing model" but Mark has explicitly stated that authors are expected to keep track of prefixes at the application level, independent of the ns machanisms
12:32
<Philip`>
jgraham: I just mean the definition of what namespace bindings are in scope at a particular element, which is (I'd hope) defined in the Namespaces spec and so RDFa just references that instead of redefining it (even if practical implementations will reimplement it)
12:36
<jgraham>
Philip`: It seems highly misleading to imply that processing happens per xmlns when in fact you are actually required to ignore xmlns and do your own prefix binding processing
12:36
<Philip`>
jgraham: You're not required to ignore it, you could do it with lookupNamespaceURI if you wanted
12:37
<Philip`>
and the result is equivalent to the normal proper xmlns processing
12:38
<Philip`>
(though it's sufficient (but not necessary) to just have an API that exposes a list of namespace-unaware attributes per element)
12:39
<hsivonen>
Dashiva: it's politically expedient in the W3C to use namespace-looking syntax. It gets you an applause even if you break the XML architecture all over.
12:39
<hsivonen>
Dashiva: remember the aria:foo debate
12:40
<Dashiva>
Will we ever learn?
12:40
Philip`
goes away
12:45
<Lachy>
This appears to be the first suggestion for renaming <sidebar> to <aside> http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2005-March/003355.html
12:45
<Lachy>
I can't see much discussion about the name itself, but the next e-mail I found that mentions it simply refers to it as though it's already in the spec
12:53
<hsivonen>
FWIW, I thought <aside> was for callouts foremost
13:00
<Lachy>
hsivonen, it's primary usecase seems to have shifted. <sidebar> was introduced to address the common case of <div id="sidebar">, then adjusted to also address callouts as being tangentially related content
13:00
<Lachy>
and then renamed to <aside> based on that
13:07
<jgraham>
FWIW <header> <footer> <content> and <sidebar> make sense as "structural" elements to me. <aside> or <callout> seems more like <section>
13:10
<jgraham>
More specifically, I would expect <sidebar><nav></nav></sidebar> to increment the heading depth by 1 whereas I would expect <aside><nav> to increment the heading depth by 2
13:12
<Lachy>
if we make <aside> inappropriate for page sidebars, then I would change it so that it's no longer a sectioning element
13:13
<Lachy>
and I would expect <sidebar> to be a sectioning element.
13:15
<jgraham>
Lachy: That is quite the opposite of what I would expect. Typically a sidebar doesn't have an overall heading but might contain several sections with headings e.g. <nav><h1>Archives <section><h1>Tags and so on
13:15
<Lachy>
so, e.g. on my site, I would do <sidebar><h1>Other Resources</h1> <section><h1>Friends</h1>...</section> <section><h1>HTML5 Stuff</h1> ...</section>
13:15
<jgraham>
Whereas a pullout box can often have a heading
13:16
<jgraham>
Lachy: Why not just use <nav> if it is purely links?
13:16
<Lachy>
because it's a sidebar. On my blog (when it was using the same template) it also contained the search form
13:18
<Lachy>
(maybe the individual <section> elements in what I wrote above could use <nav> though)
13:19
<Lachy>
if we fixed the sectioning algorithm as you previously suggested, so that sections without a heading don't affect the heading level, then it wouldn't be a problem.
13:19
<jgraham>
Lachy: True
13:22
<karlushi>
I have a tendency to classify elements like this: structure (p, ul, …) / semantic (q, cite, title, …) / functional (form, div, span, …)
13:23
<hsivonen>
karlushi: interesting categorization of div and span :-)
13:24
<karlushi>
hsivonen: yes, my reasoning is that their only purpose is that you need to attached something else to it which gives their intended meaning.
13:26
<karlushi>
The discussion around elements such as footer, header, etc. "bothers" me a little bit, because they seem awkward when things like <ads> or <logo> are not taken into account. But I'm not sure how to rationalize why it bothers me.
13:26
<erlehmann>
karlushi, is a list not semantic ? maybe you mean text-level-semantics ?
13:26
<erlehmann>
<ads> ahaha oh hwow
13:27
<karlushi>
erlehmann, a list can be of many types :) from shoping list to do list
13:27
<karlushi>
s/to do/to todo/
13:27
<erlehmann>
i can see it coming "implementors MUST provide a config option to disable rendering of <ad> elements"
13:27
<karlushi>
erlehmann, ;)
13:28
<erlehmann>
karlushi, lists are inherently semantic. that is why we have ordered and unordered and definition lists, to cater to that.
13:29
<karlushi>
that would be a nice and interesting side effects on the whole system. A bit like in magazine in France where publishers have to flag ads which are written like articles :)
13:30
<erlehmann>
karlushi, you always have to take the real-world into account. while things such as @autoplay can be there just to disable them, no one would use <ad> elemetns.
13:30
<karlushi>
erlehmann, depends on the definition of semantics. :) let's change the terms: Structure (list, paragraph, headers, sections, etc.) and Meaning (poem, dialogs, quotes, etc.)
13:31
<karlushi>
real world is not a good argument as everything is always real. The rest is cultural, economic context. :)
13:32
<erlehmann>
karlushi, i dialog is a special type of ordered list.
13:33
<karlushi>
erlehmann: yes. ordered list: a text *structure* which can have the *meaning* of a dialog
13:35
<erlehmann>
karlushi, i dont get the finer points of your statements. speak easy
13:43
<Lachy>
Interestingly, the first suggestion for a <content> element I've found came from Hixie in 2004 http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2004-August/002114.html
13:44
<hsivonen>
grr. that page triggers an elusive Necko bug
13:45
<Lachy>
Necko bug?
13:45
<Philip`>
hsivonen: "defer to the HTML5 rules for determining the language of a node" - http://html5.digitalbazaar.com/specs/rdfa.html#specifying-the-language-for-a-literal sounds like it's trying to defer to the HTML5 definition, so I don't think that's an especially controversial idea (though the details might be fuzzy in that spec)
13:45
<hsivonen>
Philip`: ah cool. Manu fixed the draft.
13:46
<Lachy>
hsivonen, what's the Necko bug? That page works fine for me in Minefield
13:46
<hsivonen>
Lachy: sometimes when reading WHATWG archives, Necko claims that the server sent bogus gzip data
13:47
<Philip`>
hsivonen: Hmm, I thought it always said that
13:47
<Lachy>
ah, like this "Content Encoding Error" I get in Minefield on this page? http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2005-April/003377.html)
13:47
<hsivonen>
Lachy: yes
13:47
<Philip`>
(i.e. since a few days ago when it was published)
13:47
<Lachy>
hsivonen, is there a bug filed about that in bugzilla?
13:48
<hsivonen>
Lachy: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2004-August/002114.html
13:48
<Lachy>
I think I've manged to track down sufficient information about the history of <sidebar> and <aside> now. Mail sent.
13:49
<Lachy>
hsivonen, yeah, that's the link I posted above. What about it?
13:50
<hsivonen>
Lachy: oops. sorry. clipboard problem.
13:50
<hsivonen>
https://bugzilla.mozilla.org/show_bug.cgi?id=501953
14:10
<jgraham>
I guess <figure>-as-<aside> might even make more sense than the current model
14:10
<jgraham>
Although I am increasingly unhappy with the whole sectioning root thing
14:11
<beowulf>
what bruce lawson describes as an aside seems to me tobe a section with a singular style in his design
14:14
<jgraham>
I think <figure>-as-<aside> would workj if you could do <figure><legend><h1>Box A</h1> and have it appear in the outline of the main document, but the complexity of that just made my head explode
14:14
<Lachy>
figure-as-aside doesn't make that much sense to me. I perceive <figure> for the kind of things that would typically be labelled "Figure #. Caption..." in a book, like images, tables, code fragments, etc.
14:14
<Philip`>
Can we just get rid of all these structural elements?
14:14
<Lachy>
no
14:14
<Philip`>
Hmph
14:15
<jgraham>
Lachy: Agreed that the name doesn't really work
14:15
<jgraham>
But the spec-wide definition does
14:15
<jgraham>
*wise
14:16
<Lachy>
I'd rather we keep <figure> as is, make <aside> for pullouts only, rather than sidebars, and introduce <sidebar> to replace the common cases of <div id="sidebar">, <div class="right">, etc.
14:16
<jgraham>
Lachy: I think that is my prefered solution too. But I could live with dual purposing figure
14:16
<jgraham>
(Note the definition is "The figure element represents some flow content, optionally with a caption, that is self-contained and is typically referenced as a single unit from the main flow of the document.")
14:17
<Lachy>
jgraham, if we dual-purpose <figure>, would you introduce <sidebar> or keep <aside> for that?
14:17
<lazymanc>
I'd prefer a defined list of roles and just using <div role="sidebar"> etc
14:17
<jgraham>
<sidebar>
14:17
<Lachy>
agreed.
14:18
<Lachy>
lazymanc, no, that makes styling more difficult for authors since it requires the less commonly used attribute selectors that not all authors are familiar with, and which IE still doesn't support.
14:19
<Lachy>
oh, actually, they were CSS2, so it might
14:20
Philip`
forgot to mention xmlns:xml and xmlns:xmlns in his latest email
14:20
<Lachy>
yep, IE8 does support them.
14:20
<lazymanc>
IE doesn't support <sidebar> <header> <footer> etc though
14:20
<Lachy>
there's a simple workaround to make IE parse those elements reasoanbly, and styling them isn't much of a problem
14:20
<Philip`>
Are there any more cases that are well-formedness errors in XML/Namespaces, where a non-XML processor would need special rules?
14:21
<lazymanc>
if you're writing a spec for the future obviously you've got to take into account backwards compat, but you're going to introduce stuff that's not supported by the vendors yet
14:22
<lazymanc>
I'd like to see textarea and select as input types
14:22
<lazymanc>
for consistency and ease of styling
14:22
<lazymanc>
obviously keeping the old tags for backwards compat
14:23
<Lachy>
lazymanc, not going to happen, and it's not clear how using <input> would make styling any easier for them
14:25
<lazymanc>
Lachy: well say you want to set a default font style / colour for all your inputs, input { // foo: bar; } is quicker and easier to remember than input, textarea, select { foo: bar; }
14:26
<Philip`>
How would parsing work?
14:26
<AryehGregor>
Same as it works now . . .
14:26
<Philip`>
<input type=text value=foo> vs <input type=textarea>foo</input> ?
14:26
<Philip`>
vs <input type=select><option>...</input> ?
14:26
<AryehGregor>
The philosophy in HTML 5 has generally been to not introduce new elements solely for greater future consistency.
14:27
<AryehGregor>
Hmm, that would make <input> a non-void element, you're right.
14:27
<lazymanc>
what's wrong with <input type="textarea" value="some content" />
14:27
<Lachy>
Philip`, it wouldn't work for type=select, but presumably type=textara could just be limited to the value attribute, which isn't such a good idea anyway
14:27
<Philip`>
lazymanc: That's hard to use when you've got multiple lines of content
14:28
<AryehGregor>
lazymanc, it's identical to <textarea>. HTML 5 doesn't usually introduce new features that are identical to existing features for the sake of consistency.
14:28
<AryehGregor>
Arguably that would be a good move.
14:28
<Philip`>
since you'd have to encode newlines as &#9; or something I guess
14:28
<AryehGregor>
Hmm, that's a good point.
14:28
<AryehGregor>
&#a;. :)
14:28
<AryehGregor>
In this case there's actually a good reason to not have it be <input>.
14:28
<lazymanc>
AryehGregor: I would argue that it should, we should fix the inconsistencies while we have the chance to benefit future content authors
14:28
<Philip`>
AryehGregor: &#xa; :-P
14:28
<lazymanc>
at least while it doesn't cause problems with BC
14:29
<AryehGregor>
Er, right, or &#10;.
14:29
Philip`
knows CRLF=0D0A, so he's not sure how he arrived at &#9;
14:29
<Lachy>
lazymanc, we don't really have a chance to fix this now, and your proposed "fix" has already been demonstrated to be flawed.
14:29
<AryehGregor>
lazymanc, we could reasonably do something like require UAs to support the feature, but make it invalid markup. And then later, when all UAs support the feature, make the old feature non-conforming (but keep it specced).
14:30
<AryehGregor>
But it seems the important people don't want to do that.
14:31
<Philip`>
lazymanc: We can't remove support for old syntax, and can't remove old content that uses it (and tutorials and editors etc), so adding new 'consistent' syntax for the same features will just result in the language being twice as large and even more inconsistent
14:31
<lazymanc>
Lachy: i'm not saying that's the way it has to be done, i'm sure there are alternatives
14:31
<AryehGregor>
E.g., require UAs to support <style src="foo"></style>, but make it non-conforming. Then 15 years from now when all UAs anyone uses support that, make it conforming and make <link rel=stylesheet> nonconforming.
14:31
<AryehGregor>
Philip`, not in the long term, though. In the long term you could eventually make the old way of doing it nonconforming.
14:31
<AryehGregor>
Then the conforming part of the language would be just as small and more consistent.
14:31
<lazymanc>
I'm not suggesting removing <textarea> and <select>, just to have the option of <input type="textarea" and then maybe depreciating <textarea> at some point down the line
14:32
<Lachy>
lazymanc, you wanted a solution using the <input> element, and there isn't one. The only other alternative is to use another element, which we already do. So problem solved.
14:32
<Philip`>
AryehGregor: Then the conforming part of the language wouldn't correspond to existing content, or to existing tutorials etc, so it'd become even more confusing
14:32
<AryehGregor>
lazymanc, value="" is annoying for multiline input.
14:32
<AryehGregor>
Philip`, that's already the case.
14:32
<AryehGregor>
And will be whenever old content is made non-conforming.
14:32
<AryehGregor>
Hmm, well.
14:32
<lazymanc>
AryehGregor: agreed, is there a reason the input element cannot be optionally void?
14:32
<AryehGregor>
I guess HTML 5 actually doesn't tend to remove stupid things people were formerly advised to do.
14:32
<AryehGregor>
Like xmlns and /> and whatever.
14:33
<AryehGregor>
So scratch what I said.
14:33
<Philip`>
AryehGregor: Sure, but it's something that should be minimised, by avoiding making things non-conforming just for the sake of consistency
14:33
<AryehGregor>
lazymanc, no element in the HTML syntax is optionally void.
14:33
<AryehGregor>
Philip`, I see your point. It's pretty annoying, though.
14:33
<Philip`>
AryehGregor: HTML is pretty annoying :-)
14:33
<Lachy>
lazymanc, yes, because it would make parsing dependent on the value of an attribute, which means feeding back information from the tree construction to the tokenisation stage
14:33
<AryehGregor>
Just because we acknowledge that pragmatically we have to support old content, that doesn't mean we have to *totally* give up on making the language nicer in the long term.
14:34
<AryehGregor>
Lachy, c'mon, that's not half as horrible as a lot of things HTML parsers have to do anyway. :)
14:34
<Lachy>
actually, no, that's wrong
14:34
<Philip`>
I think the usual response to people wanting to make the language cleaner was "you should join the XHTML2 WG", but that response is not going to work any more :-/
14:34
<AryehGregor>
But it would *decrease* consistency to have exactly one optionally void element.
14:34
<Philip`>
Parsing is already dependent on the value of an attribute
14:34
<AryehGregor>
Philip`, sure it will, it will be even more cutting!
14:34
<Philip`>
because of <input type=hidden> in foster-parented table content
14:35
<Lachy>
Philip`, in this case, I don't think lazymanc would like that anyway, since XHTML2 uses XForms, and they make the problem worse by replacing <input> with new elements for all new controls
14:36
<Lachy>
AryehGregor, the argument that HTML5 parsers already have to do a lot of horrible things is not a reason to make it do more horrible things
14:36
<AryehGregor>
Sure.
14:42
<lazymanc>
ok so I agree that there are problems with the suggested fixes, it's just annoying that with new input types being added, these two existing ones can't be bent into shape
14:46
<Philip`>
The language is like a paperclip
14:46
<Philip`>
It's bent, but if you try to bend it more to straighten it out, it'll end up with an infuriating little wobble in the middle instead of actually being straight, and if you try enough times it'll snap in half
14:46
Philip`
isn't sure this is the best analogy :-(
14:50
<lazymanc>
Philip`: it's pretty good
14:51
<lazymanc>
with regards to BC, will the new elements be ignored by older browsers without support?
14:51
<Lachy>
lazymanc, what are you using "BC" to stand for?
14:52
<lazymanc>
or will using html5 elements actually break the rendering? Lachy: backwards compatibility
14:52
<Lachy>
ah
14:53
<Lachy>
lazymanc, it depends what you mean by "break the rendering". If you're not careful, the elements won't have the correct default style. e.g. you need to specify display: block for the new sectioning elements
14:54
<Lachy>
and you also need to make sure you explicitly close <p> elements with </p> if it precedes one of the new sectioning elements like <section>
14:54
<Lachy>
but their presence in the DOM won't inherently break things
14:54
<lazymanc>
Lachy: ok thanks
15:01
<hsivonen>
Hixie: bz says the server for the WHATWG archives is buggy: https://bugzilla.mozilla.org/show_bug.cgi?id=501953#c10
15:08
<Lachy>
hsivonen, should the second response from the server also be gzipped?
15:10
<hsivonen>
Lachy: I expect so
16:38
<Philip`>
annevk3: http://philip.html5.org/data/charsets-2.html exists though I have no idea how correct it is
16:39
<Philip`>
(Particularly the error percentages seem totally bogus)
16:40
<Philip`>
(and also there's no meta-sniffer data, because it took ages and the machine on which I ran the SSH session to the machine running the analysis died so it stopped halfway through)
20:11
<Hixie>
hah, now zeldman is claiming credit for the <footer> change
20:11
<Hixie>
even though it was made in response to unrelated feedback
20:14
<miketaylr>
Hixie: i saw that too
20:20
<Lachy>
hsivonen, yt?
20:21
<Lachy>
hsivonen, I found a bug in your HTML5 parser in Gecko relating to <textarea> followed by LFs
20:21
<Lachy>
I will file a bug shortly
21:33
<Lachy>
Hixie, was <footer> changed based on e-mails that were sent months ago?
21:34
<Hixie>
iirc it was based on an e-mail that was a reply to maciej's, influenced by the feedback from a few months ago, yes
21:40
<Lachy>
wow, I don't understand JF's latest complaint about IRC discussions
21:41
<Lachy>
the discussion he linked to from earlier today clearly didn't reach any conclusions or cause any relevant spec changes
21:47
<othermaciej>
Lachy: where is his complaint to be found?
21:48
<Lachy>
http://lists.w3.org/Archives/Public/public-html/2009Sep/0267.html
21:57
<Dashiva>
Lachy: Not to mention the IRC logs are posted publicly
22:00
<Lachy>
Dashiva, I think he's made the assumption that because there wasn't much discussion about the name change from sidebar to aside, that it must have occured on IRC
22:00
<Hixie>
wait, isn't JF one of the members of the html4all self-styled cabal?
22:00
<Lachy>
yes
22:00
<Hixie>
and he's complaining about us talking in public on irc?
22:00
<Lachy>
yes
22:00
<Hixie>
i'm gonna need a new irony meter
22:01
<Hixie>
my current one is just a consumer model
22:01
<Hixie>
i need the advanced "prosumer" version
22:01
<Hixie>
with the extended gague
22:01
<Hixie>
god i hate that word
22:01
<Hixie>
Cannot Spell Gague
22:01
<Hixie>
GAUGE
22:01
<Hixie>
GAH
22:02
<Lachy>
they're aware of the irony of the situation. I think they justify the secrecy so that they can discuss issues without worrying about anything they write being ridiculed in any way
22:02
<Hixie>
well it's a good thing we don't mind being ridiculed :-)
22:02
<Lachy>
I think that's why they stopped using their public html4all list and migrated to the private list
22:03
<Dashiva>
Let's make it <rouge> instead
22:03
<TabAtkins>
It sounds trite, but if you're afraid of being ridiculed, you're probably being ridiculous.
22:03
TabAtkins
has bad memories of all the rouges sneak-attacking people.
22:04
<TabAtkins>
It just embarassing when you get knifed by a *color*.
22:04
<Dashiva>
The upside is that nobody will get it wrong
22:05
<TabAtkins>
What, 'cause they'll all assume it was meant to be <rogue>, then mistype it as <rouge> anyway?
22:05
<Hixie>
hah
22:05
<Dashiva>
TabAtkins: That's half the users. The other half just know how to spell.
22:06
<TabAtkins>
Ah, clever. Get it from both sides.
22:06
<Hixie>
I'm writing an e-mail about <meter>
22:06
<Dashiva>
It's too bad the name has nothing to do with gauges
22:06
<Hixie>
and i've mistyped the word "gauge" about 50 times so far
22:07
<Lachy>
Hixie, Irony Level: <meter class="irony" min=0 max=1000 low=0 high=100 optimum=50>999%</meter>
22:07
<Hixie>
it's only in the e-mail about 5
22:07
<Dashiva>
I don't get why anyone would typo gauge, though
22:07
<Dashiva>
Both pronounciations make it clear there's no u after the g
22:07
<TabAtkins>
I try to say guage all the time. Hixie keeps trying to type gague.
22:07
<Lachy>
Dashiva, how can gauge be pronounced in two ways?
22:07
<TabAtkins>
gahj, or gaej
22:08
<Hixie>
Dashiva: language="" was one of the most-typoed attribute names i found when doing the survey a few years ago
22:08
<TabAtkins>
langugae?
22:08
<Dashiva>
Hixie: No obvious failure
22:08
jgraham
guesses it can be pronounced in many more ways
22:08
<Dashiva>
jgraham: I had "the right way" and "the wrong way", are there more?
22:08
<Lachy>
if by gaej, you mean that it rhymes with "rage", then that's the correct pronunciation
22:08
<Hixie>
Dashiva: sure, but i'd rather avoid failures at all, even during debugging
22:09
<Dashiva>
Hixie: Better a typo failure than a conceptual failure, wouldn't you say?
22:09
<TabAtkins>
Lachy: Yeah, but I hear it rhyming with lodge sometimes too.
22:09
<Lachy>
wtf?
22:09
<Dashiva>
Isn't that gouge, TabAtkins?
22:09
<TabAtkins>
Because "au" sounding like "ae" doesn't make any sense.
22:09
<Hixie>
Dashiva: the conceptual failure would also be caught at debugging, no?
22:09
<Lachy>
what's the native language of people who pronounce it like that? It can't be english
22:09
<Philip`>
http://philip.html5.org/data/attr-count-total.txt - "96 langauge"
22:09
<TabAtkins>
Definitely English. ^_^
22:09
<Philip`>
Doesn't seem very common there
22:10
<Hixie>
Philip`: it's a long tail thing
22:10
<Hixie>
Philip`: lots and lots and lots of misspellings of various kinds
22:10
<TabAtkins>
"au" is *supposed* to sound like "ah" or "ow".
22:10
<Hixie>
Philip`: all individually rare
22:10
<Dashiva>
Hixie: Only incorrect inclusion, not incorrect exclusion
22:10
<Dashiva>
If people skip using the element because they think it doesn't apply, it won't show up
22:10
<Lachy>
we can't worry about mispellings too much for elements that have a noticable effect if you get it wrong.
22:11
<Hixie>
languaje="" was one of the most common attributes on <script>, e.g.
22:11
<Dashiva>
And besides, <metre> will be there, I guarantee it
22:11
<TabAtkins>
Dashiva: No, 'gouge' sounds different entirely.
22:11
<AryehGregor>
<Lachy> Hixie, Irony Level: <meter class="irony" min=0 max=1000 low=0 high=100 optimum=50>999%</meter> <-- That would work out to a value of 9.99 out of 1000, wouldn't it?
22:11
<TabAtkins>
...languaje? Seriously?
22:11
<Hixie>
Dashiva: one possible typo is better than however many <gague> has :-)
22:11
<Hixie>
<gauge>!
22:11
<Hixie>
gah!
22:12
<Hixie>
Philip`'s data also has 15 langage
22:12
<Lachy>
AryehGregor, I probably needed to include a value="999" attribute
22:12
<Hixie>
13 languague
22:12
<Hixie>
10 lanugage
22:12
<Dashiva>
Hixie: A typo is a typo, but a confusing name is forever
22:12
<AryehGregor>
Lachy, or do min=0 max=10 high=1 optimum=0.5
22:12
<annevk3>
Hixie, there's no such thing as an IRI scheme
22:12
<Hixie>
Dashiva: people learn concepts, they don't get over hard-to-type names
22:13
<TabAtkins>
Dashiva: I totally don't think <meter> is confusing.
22:13
<annevk3>
Hixie, IRI is a syntax, URI is the thing
22:13
<Hixie>
annevk3: so julian told me
22:13
<Hixie>
annevk3: but he wants me to give the URI syntax
22:13
<annevk3>
Hixie, and he's right
22:13
<AryehGregor>
People spell "gauge" as "gague" because it really sounds like it should be spelled "gage", so "gag" comes out and then the "u" finds its way in later. Or something.
22:13
<Hixie>
annevk3: which seems to be incompatible with the idea of IRI having a syntax
22:13
<annevk3>
Hixie, no
22:13
<Hixie>
annevk3: so where do i say that if you use ws: IRIs, you can't give a fragment identifier?
22:13
<annevk3>
Hixie, somewhat wrong, but close enough: IRI is like HTML and URI is like the DOM
22:14
<AryehGregor>
It's definitely pronounced to rhyme with "rage", not "lodge". The latter is just wrong.
22:14
<Hixie>
annevk3: no, both URI and IRI define syntax
22:14
<Dashiva>
What about the people who pronounce it gawdje?
22:14
<Hixie>
annevk3: so if i'm to define the URI syntax subset for ws:, where do I define the IRI syntax subset?
22:14
<Dashiva>
That's not too far from lodge
22:14
<AryehGregor>
"XauXe" is pretty rare in English. I can't immediately think of another example.
22:14
<TabAtkins>
AryehGregor: "wrong" doesn't mean much. ^_^
22:14
<AryehGregor>
Dashiva, definitely wrong.
22:14
<AryehGregor>
TabAtkins, "nonstandard" if you like.
22:14
<Dashiva>
"wrong"
22:15
<annevk3>
Hixie, all URIs are IRIs so you do not need to do that
22:15
<AryehGregor>
Disagrees with every dictionary I've checked.
22:15
<Hixie>
(i used to say gawje. now i say gayje.)
22:15
<TabAtkins>
Dashiva: I think we're hitting minimal-pair errors. In my accent the 'gawdje' and 'lodge' have the same vowel.
22:15
<AryehGregor>
Mine too, I think.
22:15
<Dashiva>
Well, most people don't check the dictionary when they see a new word
22:15
<Hixie>
annevk3: what?
22:15
<Dashiva>
They guess and go on until corrected
22:15
<annevk3>
Hixie, I guess you're right about my analogy being flawed, can't really come up with something better
22:15
<AryehGregor>
No, but the people who are confident in how to pronounce "gauge" will think "ha, this person is an ignoramus who doesn't know how to pronounce 'gauge'" if they hear you say "godge".
22:15
<Dashiva>
Now, assume you're in a place where nobody else knows how to pronounce it either
22:16
<Dashiva>
It proliferates fast
22:16
<AryehGregor>
So practically speaking, it's useful to stick to the standard educated pronunciation.
22:16
<annevk3>
Hixie, the parse URL already takes care of IRI -> URI
22:16
<Hixie>
annevk3: i want ws://example.com/©/ to be a valid ws: IRI
22:16
<Hixie>
annevk3: and i want ws://example.com/©/#test not to be
22:16
<TabAtkins>
AryehGregor: No? Generally, both sides will think the other one is wrong. It's only if one of them consults a dictionary that it matters. And I'll bet that the average English-speaker has *never* consulted a dictionary for the pronunciation of 'gauge'.
22:16
<Hixie>
annevk: if i say the syntax of ws: URIs is "hier-part query?", then how is the former valid?
22:16
<Dashiva>
AryehGregor: Besides, anything you say is moot until you guys decide on one pronounciation for tomato
22:17
<annevk3>
Hixie, after conversion it's valid
22:17
<AryehGregor>
TabAtkins, the nonstandard side will usually either a) not be as sure, or b) accept the other pronunciation as legitimate. The standard pronunciation is always at least common, even when it's not ubiquitous.
22:17
<Hixie>
annevk3: i want it to be valid before conversion.
22:17
AryehGregor
$ grep 'au..\?e$' /usr/share/dict/american-english
22:18
<Hixie>
annevk3: and i want the frag-id case to be invalid before conversion
22:18
<Hixie>
annevk3: or are you saying that IRIs can't be syntax-checked without converting them to UR
22:18
<Hixie>
Is?
22:18
<AryehGregor>
"applause", "because", "cause", "clause", "gauze", "mauve", "pause", "sauce". I think "gauge" is sui generis here, yep.
22:18
<annevk3>
Hixie, I think so, but now I'm no longer sure whether I'm right :/
22:18
<TabAtkins>
AryehGregor, you have a bad conception of how people learn language. They stick with what they heard first, only changing with difficulty. I can assure that most of the people rhyming with "rage" did *not* develop their pronunciation by consulating a dictionary.
22:19
<Hixie>
i hate the uri/iri BS
22:19
<Hixie>
the whole space is a big giant mess
22:19
<annevk3>
Hixie, no disagreement there
22:19
<AryehGregor>
TabAtkins, no, they developed it by hearing other people speak. And a large percentage of people who use words like "gauge" in ordinary conversation will be using the standard pronunciation.
22:19
<Hixie>
they should scrap all those specs and do a proper job with built-in i18n, no ASCII nonsense, and defining error handling, etc.
22:19
<Dashiva>
Errors don't exist on the internet, silly Hixie
22:20
<AryehGregor>
Hixie, are there any application-level Internet standards other than HTML 5 that you think are done right?
22:20
<Hixie>
CSS does a reasonable job
22:20
<TabAtkins>
AryehGregor, large percentage != overwhelming percentage. The fact that I still hear it rhyming with 'lodge' plenty attests to that.
22:20
<Hixie>
Unicode is pretty good
22:20
<othermaciej>
according to timbl, even the original term "URL" came from the IETF's desire to rename his original term (UDI)
22:20
<Hixie>
XForms is pretty good
22:20
<annevk3>
of course everything around Unicode is pretty bad
22:20
<Dashiva>
I found out the other day that CSS doesn't place any restrictions on the length of a real number
22:21
<annevk3>
Dashiva, that's a feature
22:21
<Hixie>
SVG Tiny 1.2 is pretty good
22:21
<Hixie>
(XForms and SVG have their own issues, but we're just talking spec-wise here)
22:21
<Dashiva>
annevk3: A feature that leads to getting three different results from four browsers
22:21
<AryehGregor>
TabAtkins, thus everyone who's sure about pronunciation will accept "gage" as a valid pronunciation. Those who have checked in a dictionary, or heard it only from more reliably standard speakers, will not accept "godge" as a valid pronunciation. Thus you may as well use the standard one.
22:21
<AryehGregor>
FWIW, I've never heard anyone pronounce "gauge" as anything but "gage".
22:22
<Hixie>
(e.g. XForms failed hard at being relevant for the Web, and SVG is an overly-complex language and 1.2 introduces too many new features)
22:22
<Dashiva>
I've heard godge many times
22:22
<Hixie>
(but they are both written reasonably well, compared to the rest of the space)
22:22
<AryehGregor>
Anyway, it does matter what people think of your pronunciation if you're being interviewed for a job, etc. Not a completely trivial thing, in life.
22:22
<Dashiva>
If you say gauge during an interview, they might just think you're pretentious too
22:23
<AryehGregor>
In some jobs, that's a plus!
22:23
<TabAtkins>
AryehGregor: Then you're clearly in an abnormal area of the english-speaking world. ^_^ Again: dictionaries are *not* consulted in general. There is a sense of learning from people you perceive to be more 'correct', but that's half sawdust and lies anyway.
22:23
<AryehGregor>
(says the Ph.D. student)
22:23
<Hixie>
XHR is another good spec
22:23
<Dashiva>
Busted
22:23
<Hixie>
CORS, too
22:23
<AryehGregor>
TabAtkins, what's perceived to be "correct" is totally arbitrary, for sure. But it's worthwhile to go with it anyway, all things being equal. (Which sometimes they're not.)
22:24
<annevk3>
Dashiva, that's nothing, most features have six different results in four browsers
22:24
<annevk3>
cheers Hixie
22:24
<annevk3>
I kind of like HTML5 too
22:24
<Dashiva>
AryehGregor: You still haven't addressed the cases where there are multiple "correct" alternatives
22:24
<annevk3>
:p
22:24
<Hixie>
annevk3: :-P
22:24
<Lachy>
AryehGregor, the core of JF's argument was that decisions were being made on IRC. You should have defeated his argument by pointing out that no decisions were in fact made in the IRC disucssion he complained about.
22:24
<AryehGregor>
Dashiva, then use whichever you feel like. Why is that a problem?
22:25
<Dashiva>
Because some people will contest the correctness of some of them
22:25
<Hixie>
i think a big part of hte problem i that a lot of people writing specs started when the internet was basically a small community project, mostly military and academia
22:25
<Lachy>
(I almost responded, but decided not to as it wouldn't have been too constructive)
22:25
<TabAtkins>
AryehGregor: Not if you think it's wrong, which 'gauge'='rage' clearly is. ^_^ Your search through the dictionary for other uses of that vowel pair showed that it's a crazy outlier, and should be corrected in common use to increase consistency.
22:25
<Hixie>
or they learnt abotu specs from people who were already in that mindset
22:25
<AryehGregor>
Lachy, well, decisions like "that's unlikely to be added to HTML 5" occur here pretty regularly, which effectively halt changes that might otherwise have been pursued.
22:25
<AryehGregor>
I don't think it's a problem even if some decisions are made here.
22:26
<AryehGregor>
Dashiva, so life isn't perfect. Pick the least controversial or give up and don't care.
22:26
<Lachy>
which decisions like that are you referring to?
22:26
<AryehGregor>
Lachy, like that guy who came in here yesterday asking that <textarea> be rolled into <input>.
22:27
<AryehGregor>
We basically shot him down here. That's a decision of sorts, although it resulted in no change to the spec.
22:27
<Lachy>
that wasn't just decided yesterday. It was just explained to him why it wouldn't work
22:27
<AryehGregor>
TabAtkins, "correct" pronunciation signals your level of education and socioeconomic status, basically. It shows that you're the kind of person who actually does things like look stuff up in dictionaries if you're unsure. Take it or leave it.
22:28
<AryehGregor>
Lachy, sure. But it stopped the proposal from going forward. If you accept the premise that IRC is non-public, then it would have been more transparent if the reasons why his proposal wouldn't work were posted publicly on a mailing list.
22:29
<Lachy>
I don't accept the premise that IRC is non-public
22:29
<AryehGregor>
Nor do I, and that's what my post said.
22:29
<TabAtkins>
Indeed, but you're talking to a Texan who was raised with a healthy does of 'ghetto' pronunciation as well. My casual speech is already halfway-retarded to the sort of person who consults the dictionary for pronunciation.
22:29
<Lachy>
he didn't raise his issue on the mailing list, why should we bother everyone else with a response to an issue that wasn't even raised there and was easily explained to the guy that raised it where he raised it
22:29
<AryehGregor>
I think some decisions *are* made here. Not as important as the ones made on the mailing list, but some. Which is why I didn't make the point as you suggested.
22:30
<AryehGregor>
TabAtkins, while you're talking to a New York Jew who's a Ph.D. student, who was brought up in a house where dictionaries/encyclopedias/atlases/globes/etc. were and are routinely consulted during family meals to resolve arguments. Different perspectives, I guess. :)
22:30
<Lachy>
AryehGregor, the specific discussion he linked to, relating to the section/article/content element discussion, absolutely made no decisions
22:31
<AryehGregor>
Lachy, but others did, IMO. Or if not decisions, then at least things that should be public if we want a transparent spec development process. But they are public, since they're on IRC, so that's okay.
22:31
<Lachy>
anyway, doesn't matter. Best to just ignore the complaint anyway
22:34
virtuelv
responds to the fakepath thread
22:42
<cardona507>
the spec says "MOST elements that are categorized as phrasing content can only contain elements that are themselves categorized as phrasing content, not any flow content." - What phrasing content elements can contain flow content?
22:44
<TabAtkins>
Is <a> a phrasing content element?
22:44
<TabAtkins>
If so, that one.
23:29
<annevk3>
Hixie, I just read some more of that thread and if Addison Phillips is right, and that seems more likely than me being right, I was wrong above about there not being such a thing as an IRI scheme
23:31
<Lachy>
TabAtkins, I'm really struggling to understand your arguments. They just don't seem rational and it's not at all clear what you're arguing for
23:32
<Lachy>
(re the header/footer/sidebar discussion)
23:32
<TabAtkins>
Lachy: one sec, in a call at work
23:32
<Hixie>
annevk3: well fwiw, i'm struggling to understand any of that thread
23:33
<Hixie>
annevk3: i just wish people (especially julian) would stop whining about what the rules are and just say wtf they would like the spec to say to satisfy them
23:33
<annevk3>
Hixie, yeah, he's not always very constructive :/
23:36
<TabAtkins>
Lachy: My point is that trying to draw semantic distinctions between header/footer/aside (used as website structure) doesn't do any good. If you can put some item in one of those locations, it's almost guaranteed to be attested in the other two across the web as well.
23:37
<TabAtkins>
So, semantically, header/footer/aside as *website structure* are very nearly the same (header being a slight exception, as it nearly always contains the site header).
23:37
<TabAtkins>
header/footer/aside used in an article carry very real and distinctly different meanings, of course. These are currently expressed in the spec.
23:37
<Lachy>
and as I tried to point out on the list, you're placing too much significance on the content alone, and ignoring the structural and contextual meanings conveyed by each
23:38
<TabAtkins>
What's the structural and contextual meaning of a blogroll being placed in the header as opposed to the footer or aside?
23:38
<Lachy>
when is a blog roll ever placed in a header?
23:39
<Lachy>
it's not introductory content, it doesn't belong there
23:39
<TabAtkins>
It's often not, because of (purely visual) size considerations. If you would like, just consider it in a sidebar or footer. What's th difference?
23:39
<Lachy>
it's not just visual consideration. It's structural and semantic
23:39
<TabAtkins>
Okay, look at, frex, http://www.igofigure.com.
23:39
<TabAtkins>
The heading area contains 6 distinct things.
23:40
<TabAtkins>
One of those is definitely appropriate for <header> (the site heading). One is possibly best in header (the main nav). The other four could be put anywhere equally well.
23:40
<TabAtkins>
Those four are placed in the heading area for purely visual reasons.
23:41
<cardona507>
is this worded correctly? "The value must be a short free-form string that giving the name of the Web application that the page represents. "
23:42
<TabAtkins>
cardona507: remove the "that" from "string that giving".
23:42
<cardona507>
its from the spec
23:42
<TabAtkins>
Then it's a spec bug. ^_^
23:43
<Hixie>
file it using the bug filing widget please :-)
23:44
<cardona507>
its also in the W3C spec
23:45
<Hixie>
it's the same document
23:46
<cardona507>
for some reason I felt like the whatwg was "fresher" :)
23:46
<Hixie>
it is :-)
23:46
<Hixie>
i edit the whatwg one live
23:46
<Hixie>
then when i'm done with an editor, i run a script that commits it to svn and uploads it to the w3c with the w3c header on it
23:47
<Hixie>
er, done with an edit, rather
23:47
<Lachy>
TabAtkins, I'm not disagreeing that their is partial overlap in the kinds of content that could potentially be included in each. But by placing, e.g., the language/country selector in the header, the designers intended for it to be part of the introductory content so it is among the first things encountered in the site.
23:47
<Hixie>
(and then every few months the w3c make a snapshot of the document to copy to the /tr/ page)
23:47
<TabAtkins>
I'm the designer, btw. ^_^
23:47
<Lachy>
While it could technically be placed in the sidebar or footer, placing it structurally near the beginning has a clear purpose
23:47
<Hixie>
(though i can't for the life of me work out why!)
23:48
<TabAtkins>
And it's placed in the heading area because the sidebar is already full enough, and putting more stuff in it would push down things that I want to be visible on pageload.
23:49
<TabAtkins>
There was no intention that it was introductory content. That blue bar up top is treated, for all intents and purposes, as an <aside> to me.
23:50
<TabAtkins>
And, given appropriate support from CSS, that totally *would* be an <aside>, placed somewhere else in the document.
23:50
<Lachy>
TabAtkins, why do you think it's most common for sites that include country/language selectors near the top of the page, rather than sticking them in sidebars or footers?
23:50
<TabAtkins>
Because you want to ensure it's seen on pageload, without taking up valuable sidebar inches.
23:50
<Lachy>
clearly, there has to be more than a purely asthetics reason for it
23:50
<TabAtkins>
It's a practical concern.
23:51
<cardona507>
hixie - is the bug filing widget the little arrow at the bottom right of the screen?
23:51
<cardona507>
on the spec that is
23:51
<Lachy>
right, a practical concern greatly affected by the chosen structure
23:51
<Hixie>
cardona507: yes
23:52
<TabAtkins>
It is, however, entirely tangential to the page itself, and certainly isn't introductory content.
23:52
<cardona507>
limited characters in the bug report widget?
23:52
<TabAtkins>
Lachy: It's only affected by the chosen structure because page layout is so impoverished on the web.
23:53
<Hixie>
cardona507: i don't think so
23:54
<TabAtkins>
Hixie: iirc, there's a limit. I was lucky that my last bug was *exactly* long enough to fit. One or two characters more would have done it in.
23:55
<Hixie>
huh
23:55
Hixie
checks
23:55
<cardona507>
255 characters
23:55
<Lachy>
TabAtkins, personally, I probabaly wouldn't include that witin the header element itself anyway, but it would certainly near the top, either before or immeidately after the <header>
23:55
<Hixie>
huh, i wonder why i have a maxlength
23:56
<Hixie>
i've removed the maxlength
23:56
<Lachy>
I likely wouldn't include the <nav> within the <header>, but that doesn't matter so much
23:56
<TabAtkins>
Indeed, somewhere near the top. You want it exposed early on, even if it's not part of the page header. Hell, you may even want it as an <aside> appearing right in the beginning of the page stream.
23:56
<TabAtkins>
Lachy: Yeah, again, that's due to layout concerns.
23:57
<TabAtkins>
Given something appropriately powerful like Template Layout, the structure would be a goodly bit different.
23:57
TabAtkins
notes that that's the major reason he joined the CSSWG - so he can ensure that things like layout are driven properly.
23:58
<Lachy>
no, it's because I'd want it near the beginning, irrelevant of the design. It could just as easily be positioned at the bottom of the page with CSS
23:58
<TabAtkins>
Yo, Maciej.
23:58
<othermaciej>
hello!
23:59
<cardona507>
*bug filed
23:59
<TabAtkins>
Lachy: Ideally, I'd probably do something like <aside #lang-selector /><header /><nav #main-nav /><article /><lots of navs and asides>
23:59
<Lachy>
TabAtkins, for the same reason a lot of developers include the navigation structually after the main content, but style it to be at the top
23:59
<Lachy>
sure, that would be a reasonable approach