Re: Microdata

Hi Chaals,

I think there's a tangle of issues here. I'd like to disentangle some 
pieces of information before diving in.

First: the current HTML 5.0 CR does *not* include Microdata under any 
form. That may or may not have been a good decision. At any rate, I 
think the ship on changing that sailed some time in last December.

Second: the proposal to incorporate MD into HTML 5.1 was only for the 
"integration points". That is to say, at this point in time, if you look 
at the 5.1 draft it has things like "when using Microdata attribute foo, 
then this element can also do such and such" but it does not include the 
bulk of MD.

Third, as Jirka points out, there's the syntax and then there's the API. 
The API is meant for browsers. If we are certain that we're not going to 
get the required support to ship that, we shouldn't do it just for the 
fun of it.

On 20/05/2013 11:13 , Charles McCathie Nevile wrote:
> Microdata is included in the HTML spec, and RDFa Lite (which does the
> same thing) isn't.

Not really, see above.

> There is a proposal to remove it from HTML5 and delay it to HTML 5.1.
> This seems a very bad idea, and we object.

That's not really the proposal. Currently the integration points are in 
the 5.1 draft, the bulk of MD itself is only in its own document (which 
isn't at CR yet).

 From this point on, there are several things that we can possibly do.

A) Push the current separate MD draft to CR as is. That's a bad idea 
because we know that it won't come out of it unless some miracle makes 
the APIs implemented in enough browsers.

B) Fold the whole thing (not just the integration points) into HTML 5.1 
and figure out how the market stabilises between now and 2016. I don't 
think that hurts anyone; it does make the editors' job easier. It is, 
however, a bit of a cop out.

C) Ship a separate MD specification (I would assume as a regular WD, at 
most as an LCWD) with the API pulled out (as well as anything 
browser-related) and only the syntax parts included. This has two 
sub-options:
   C.1) Keep the integration points in HTML 5.1.
   C.2) Pull the integration points from HTML 5.1, and make it so that 
they get included into the generated MD draft.

I suspect that the consensus position, and the position that can 
actually lead to a useful specification that corresponds to real-world 
implementations, is C.

If we go with C, I would propose that we go with C.1 for the time being 
as that's what requires the least editorial work but that we move to C.2 
if either the new MD turns out to require being on a substantially 
different timeline from HTML 5.1 (which is likely, but can be dealt with 
when we get there) or if MD peters out (at which point we won't want to 
mention it at all).

Thoughts?

> Microdata is the format most heavily promoted by schema.org, and has
> wide implementation.

I was under the impression that this was no longer the case, or at least 
shifting. I've copied Danbri (Hi Dan!) on this so that he may weigh in 
with his usual wisdom.

> We are sympathetic to the argument that taking microdata out of HTML5
> improves modularity and is therefore good, and to the argument that
> removing microdata puts it on a more logical level footing with RDFa
> Lite in public perception, reducing the risk of suggesting one is better
> than the other for use with HTML, instead of leaving it to the market to
> determine. Although these are fundamentally political, rather than pure
> technical arguments, they are not incorrect.

They also have the appeal that Microdata is *already* outside of HTML :)

> Unfortunately we are currently unable to provide significant editing
> resources of the calibre and experience with the HTML specification that
> is already available to the HTML Working group. If those editors are
> really unable to extract the spec and progress it to Recommendation, we
> believe that the next-best option is to keep it in HTML5.

Just to be clear: the core of the spec is extracted. In HTML 5.0, even 
the integration points (which are basically modifiers on HTML that are 
sprinkled throughout the spec) are extracted (though not very well — 
they don't appear in any document at all, which is a problem that would 
need a solution). In HTML 5.1 the integration points are included, but 
the core of the specification itself is still extracted.

It's the editors' job to shepherd this spec to Rec if that's what the 
group wants. I have to note however that there is no point in us 
bringing anything along the process that's just going to die there. So 
if the consensus is to keep MD alive, at least for a while, then at the 
very least it needs to have its browser parts culled and it needs to 
return to an earlier stage in the Process for further review.

-- 
Robin Berjon - http://berjon.com/ - @robinberjon

Received on Tuesday, 21 May 2013 14:54:02 UTC