W3C home > Mailing lists > Public > public-html@w3.org > January 2010

Re: Decentralised extensibility idea (ISSUE-41)

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Sun, 17 Jan 2010 09:50:04 -0800
Message-ID: <dd0fbad1001170950y7ae5c16cga1c62f3a54ca4d67@mail.gmail.com>
To: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Cc: Julian Reschke <julian.reschke@gmx.de>, Toby Inkster <tai@g5n.co.uk>, Philip Jägenstedt <philipj@opera.com>, "public-html@w3.org" <public-html@w3.org>
On Sat, Jan 16, 2010 at 8:27 AM, Leif Halvard Silli
<xn--mlform-iua@xn--mlform-iua.no> wrote:
> If @profile is of limited use, now, then Tobys DE proposal increases
> its precision w.r.t identifying where exactly the profile which the
> @profile points to is applied, by the fact that
>
>  * @profile can be placed on any element
>
> The precision is increased even more by because HTML5 allows @class on
> all elements.

Not really.  At best you reduce the area of ambiguity.  Without
vocabulary knowledge, though, you still can't extract information from
anything in that subtree.

As well, if you limit the profile to a subtree, that limits your
ability to spread information around the page, for styling or
organizational purposes.  This was a use-case identified during the
creation of Microdata.  Microdata suffers similarly, and solves the
problem with @itemref.  RDFa avoids the problem entirely by going for
a more abstract mapping that doesn't hook into the page structure
quite as strictly.  This proposal doesn't have anything to address
that (yet).

> A best practise which included the use of prefixes in the class names
> and the data-* attributes, could increase the precision probably close
> to the level of a namespaced solution.

Closer to, and yet still too ambiguous to be extracted by generic
tools.  It is *required* that toolmakers recognize your specific
vocabulary for it to be usable.  (Either that, or the profile url has
to be dereferenceable to a usable vocabulary description, which, as I
said before, brings up all the problems that DOCTYPE urls have already
suffered from.)

> I am puzzled by the remark from Tab. Profiles only makes use of
> existing features within HTML. Thus it is the most generic solution of
> all D.E. solutions.

This does not seem to follow - I don't see how genericness comes from
using only existing features.

> It also uses the same extension method that HTML5
> iself uses: @profile is only a way for pointing to an applicable
> specification without having to wade through the wg desert.

Indeed, and the Microformats community has made it rather clear that
HTML's legacy extension mechanisms aren't very capable.  You can twist
them into working, but you need vocabulary-specific knowledge to use
them, and there's still some common cases that are fairly difficult to
get to work correctly.  (See the extended problems with embedding
datetimes in Microformat vocabs - it's finally become a solved problem
using HTML4, and HTML5 introduces an entire element pretty much for
the sole purpose of solving the problem in an easier way.)

~TJ
Received on Sunday, 17 January 2010 17:50:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:57 GMT