- From: Charles McCathieNevile <charles@sidar.org>
- Date: Tue, 22 Apr 2003 15:54:05 +0200
- To: Isofarro <w3evangelism@faqportal.uklinux.net>
- CC: Josh Hughes <josh@deaghean.com>, w3c-wai-ig@w3.org
The HTML specification defines an algorithm for determining table headings: http://www.w3.org/TR/html401/struct/tables.html#h-11.4.3 chaals Isofarro wrote: >>I'm developing a data table interpreter. I almost done with it, but I'm >>a little stumped on what to do about table headers that haven't been >>given a scope or id/header association. >> >> > >Possibly the next step (after col, colgroup, thead and tfoot) is to look at >elements around it. I'd hazard a guess that if there's table data in the >same row, either immediately before or immediately after the th, then the >scope of the header could be row. If there's table data in the corresponding >cell on the previous or next row, then the scope may be col. > >There are probably quite a few exceptions to these pointers. > > > >>My first inclination was to have scope="col" be the default behavior, >>but I've run into instances where scope="row" is assumed as well. Is >>the user agent expected to guess in these situations? >> >> > >Seems that way :( > > > >>This is going to be part of a validator, so I'm not interested in error- >>handling; I want to know what is expected from a standards perspective. >> >> > >I don't believe there's anything further defined after scope header and id >attributes. Although there's still the colgroup / col elements that could >allow grouping. Correspondingly, the thead and tbody give inklings to col >type scopes. > > >Mike. > > > >
Received on Tuesday, 22 April 2003 09:54:58 UTC