- From: Tantek Çelik <tantek@cs.stanford.edu>
- Date: Mon, 05 Apr 2004 12:03:49 -0700
- To: Ian Hickson <ian@hixie.ch>
- Cc: <Matthew.van.Eerde@hbinc.com>, <www-style@w3.org>
On 4/5/04 11:18 AM, "Ian Hickson" <ian@hixie.ch> wrote: > On Mon, 5 Apr 2004, Tantek [ISO-8859-1] ?lik wrote: >> >> I think there can be many examples where the table-ness of elements (i.e. >> their table semantics) MUST be determined by the language (e.g. say you were >> marking up a sparse matrix -- the mathematical kind), and thus it would be >> possible to create pseudo-classes that selected the semantic table-ness of >> those elements rather than their presentational table-ness. > > Yeah, that might make sense. It would mean that :column(3) would select > cells in the third column in the markup, as opposed to in the rendering, > but I expect that's probably ok. > > It would mean that not-over-the-wire scenarios where UAs do not have > knowledge of the language being used are disadvantaged, but that's not a > big problem. (Over-the-wire scenarios don't have this problem since you > should never send unknown languages over the wire.) It would be no different than the various UI related selectors which require the UA to have knowledge of the form-ness of elements defined in the language being used, and certainly we've already agreed that this is not a problem (e.g. Selectors CR, CSS3 UI (CR soon)). A good place to start (to define table related pseudo-classes) may be to understand how matrices are marked up in MathML, and how the table-ness defined there in syncs up with the semantic tables in XHTML. Pseudo-classes could certainly be defined based on a model that would handle those two semantic languages to begin with. Tantek
Received on Monday, 5 April 2004 15:03:52 UTC