02:03
<TabAtkins>
Hixie: In "Selecting names when defining vocabularies"'s list of properties that appear in the example at the bottom, you say "This example has one item with two types...". I only see the one type, though.
02:04
<Hixie>
fixed thanks
02:13
<TabAtkins>
What *possible* times would an *element* be needed for customization where a class wouldn't cut it?
02:13
<TabAtkins>
That... that just doesn't make sense. <div class="foo"></div> is functionally identical to <foo></foo> for purposes of extending the language. The only difference is that the first one works right now and is accepted by everyone.
02:15
<Hixie>
chapter 3 of http://www.rfc-editor.org/rfc/rfc5434.txt seems relevant
02:35
<othermaciej>
TabAtkins: I looked at some SVG files where Inkscape adds elements and not just attributes for its design-time metadata
02:35
<othermaciej>
TabAtkins: it does seem to use elements for some things
02:36
<othermaciej>
TabAtkins: it seems to be in cases where there are logical "objects" with a bunch of key-value pairs, where in principle you may have more than one such object per document or per affected parent element
02:36
<othermaciej>
TabAtkins: so you can't just add attributes to the affected element
02:36
<othermaciej>
TabAtkins: or at least, that would be awkward
02:36
<othermaciej>
you have to use a semantically null, non-rendering element
02:36
<othermaciej>
<div> is not great because, even though it has no semantics, it renders by default
02:37
<othermaciej>
<meta> might work, though only for things that have document-wide scope
02:41
TabAtkins
is sorry, was off making dinner.
02:42
<TabAtkins>
That makes sense, but we have tools for that now.
02:42
<TabAtkins>
Microdata is the obvious candidate.
02:43
<TabAtkins>
Plus the aforementioned just-use-a-comment suggestion.
02:43
<TabAtkins>
Hell, embed xml in your comment or whatever. If you have the source, that's more than adequate.
02:43
<TabAtkins>
And it doesn't mess with what anyone else sees, as intended.
02:46
<TabAtkins>
Though I see how comments would fail the "must roundtrip through editting tools" requirement.
02:51
<othermaciej>
yes, I think comments fail due to round-tripping
02:51
<othermaciej>
microdata could work in place of data-* attributes, but I think it has the same issue as RDFa of introducing overconstraining model structure
02:52
<othermaciej>
microdata doesn't by itself provide a solution to the kinds of use cases that would call for an element with null semantics and no rendering
02:53
<othermaciej>
(and which can be placed anywhere)
02:54
<othermaciej>
like if your editor wants to represent multiple "viewports" (where you might have different size/shape windows zooming into different parts of the image), there is likely no existing element in the document to associate each one with
02:54
<othermaciej>
and it's hard to express as a single set of key-value pairs on the root element
02:55
<othermaciej>
I suppose you could do it with sufficient extra levels of indirection, it would just be messy
02:57
<othermaciej>
OK, microdata can do it, using <meta itemprop itemscope>
02:59
<othermaciej>
but requiring a separate element for each property would be a significant burden, I think
03:00
<othermaciej>
it would be nice it you could add properties directly to the itemscope element with attributes of some kind, maybe something analogous to data-* attributes
03:10
<Hixie>
can't inkscape just use xml as its internal format, like, well, like it does already?
03:13
<othermaciej>
I think Inkscape is being used as a stand-in for HTML editing tools that may similarly want to record design-time information
03:14
<othermaciej>
(including hypothetical future SVG-in-HTML tools)
03:14
<othermaciej>
at least that's how I am thinking about it
03:16
<othermaciej>
(but it provides more of a concrete example than a purely hypothetical example)
03:25
<TabAtkins>
It seems like we're talking about tools here, so I'm not sure why lots of <meta>s is so bad. It would be inconvenient when hand-authoring, but that's not the use-case we're talking about.
03:26
<TabAtkins>
It's a bit weighty, but not horribly so.
03:27
<TabAtkins>
And with a lot of properties that are grouped, you can wrap them in a <div itemscope itemtype="foo" hidden></div> to save on the per-item weight.
03:32
<Hixie>
othermaciej: can't they use xhtml, then? or does the on-the-wire format have to have this stuff? it seems bad for bandwidth to be encouraging people to export this proprietary stuff onto their site
05:29
<boblet>
On a weblog homepage showing excerpts of recent articles (wrapped in <article>), these shouldn’t be inside a parent <article>, should they (<section> instead, right?)
07:33
<annevk42>
Hixie, wouldn't itemScope etc. be more consistent?
07:51
<MikeSmith>
Hixie: I thought you had mentioned dropping the predefined vocabularies
07:52
<annevk42>
are they not dropped?
07:54
annevk42
doesn't see them
07:54
<MikeSmith>
diff seems to just show that they moved
07:57
<annevk42>
I think it shows they are no longer in HTML5 but still in the source doc
07:58
<annevk42>
itemtype doesn't even allow predefined vocabs
07:59
<MikeSmith>
annevk42: yeah, OK, I see now
08:00
<othermaciej>
well, there's one issue we can resolve
08:01
<annevk42>
I guess the idea is to define org.w3.vcard or some such
17:08
<Dashiva>
http://www.jgc.org/blog/2009/09/javascript-must-die.html
17:13
<miketaylr>
deprecate javascript, that's cute
17:17
<gsnedders>
I think that'll have about as much effect as deprecating presentational markup has had.
17:36
<othermaciej>
there's a lot of sloppy thinking and/or sloppy communication in those slides
17:50
<Dashiva>
Oh, poor RDFa. It only supports RDF.