- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Wed, 3 Aug 2011 07:45:18 -0700
- To: Cameron Heavon-Jones <cmhjones@gmail.com>
- Cc: Manu Sporny <msporny@digitalbazaar.com>, W3C TAG <www-tag@w3.org>, "HTML WG (public-html@w3.org)" <public-html@w3.org>
On Wed, Aug 3, 2011 at 6:59 AM, Cameron Heavon-Jones <cmhjones@gmail.com> wrote: > On 01/08/2011, at 6:21 PM, Tab Atkins Jr. wrote: >> On Fri, Jul 29, 2011 at 7:58 AM, Cameron Heavon-Jones >> <cmhjones@gmail.com> wrote: >>> The inclusion of data-(*) attributes is possibly seen as one of the simplest and most liberating additions to the HTML language within this version's iteration. However their omission from greater semantic scope would seem to either render these attributes as opaque for semantic derivation or provide an input to be considered for capturing ontological information in the disparate environment of the web. >> >> You're correct; data-* attributes are required to be opaque for >> semantic derivation. They are meant only for page-local use, and must >> not be used to embed information for third parties to read. > > Yes, i'm aware of their currently intended scope but my open question was whether this is a prudent design decision? Yes, because they were not designed for, and are not appropriate for, marking up data for third-party machine-reading. The use-cases that were consulted in the creation of the data-* attributes have nothing to do with embedding third-party machine-readable data, and if you tried to evaluate the data-* attributes using the use-cases that were consulted in the creation of Microdata, you'll find that the data-* attributes are a fairly poor solution to them. > Should we be recommending that authors use a markup solution which is opaque? > > Is there such a thing as opaque data? Yes, there are multiple sources of opaque author-generated data, such as the 'id' and 'class' attributes. The data-* attributes were designed solely to make it easier to embed key/value pairs of data, which is somewhat difficult to do using the other attributes which authors control. > What the data-* attributes do bring to HTML over-and-above other markup solutions is their immediate targeting and application of CSS styles and rules, something which is completely missed from any discussion on data, its identity and computational comprehension. You can, if you wish, target CSS using the item* attributes or the RDFa attributes as well. data-* has no particular advantage here. All of them are inferior to just using 'class' to target CSS. ~TJ
Received on Wednesday, 3 August 2011 15:30:12 UTC