- From: Jim Jewett <jimjjewett@gmail.com>
- Date: Wed, 8 Jul 2009 14:01:09 -0400
- To: John Foliot <jfoliot@stanford.edu>
- Cc: Janina Sajka <janina@rednote.net>, Henri Sivonen <hsivonen@iki.fi>, joshue.oconnor@cfit.ie, Laura Carlson <laura.lee.carlson@gmail.com>, Sam Ruby <rubys@intertwingly.net>, Michael Cooper <cooper@w3.org>, "W3C WAI Protocols & Formats" <w3c-wai-pf@w3.org>, Gez Lemon <gez.lemon@gmail.com>, wai-liaison@w3.org, www-archive <www-archive@w3.org>, HTML WG Public List <public-html@w3.org>
On Wed, Jul 8, 2009 at 12:52 PM, John Foliot<jfoliot@stanford.edu> wrote: > Jim Jewett wrote: >>... spend their time writing better alt text, and >> better captions, and better table header cells. > None of those others provide what summary provides. True, but they are also useful, and some to the same population -- and they aren't always provided, because of time. Would 7 extra alt tags be worth more than a mediocre table summary? > If summary is to be obsoleted, then deliver on the > specific need that it was trying to address, Agreed -- which is why I keep asking what a good summary *should* look like, so that I can try to address the need. So let me give some examples, and ask what would be acceptable. Starting with the same sample table at http://juicystudio.com/wcag/tables/complexdatatable.html, would any of the following be acceptable summaries? I ask because I think the browser could generate these automatically, so if one of these is acceptable, it could serve as the replacement for @summary. "Displays Child Investment, Type, Status, Allocation, TCO, ROI, NPV, Property, and Running Cost" "Displays Child Investment, Type, Status, Allocation, TCO, ROI, NPV, Property, and Running Cost (broken down by 12/12/2005, 12/19/2005, and 12/26/2005)" "7 columns with merged rows, a property column, and running cost for each of 12/12/2005, 12/19/2005, and 12/26/2005" "Displays merged rows of Child Investment, Type, Status, Allocation, TCO, ROI, and NPV. Each row is split by Property to show the Running Cost for each of 12/12/2005, 12/19/2005, and 12/26/2005." I will freely grant that it should be possible to do even better by hand. But I believe the browser could generate any of these four by hand -- so if any of these are "good enough", then such an algorithm should be specified -- and the question is whether to use it in all cases, or only when there is no (supposedly) handbuilt summary. (The algorithm isn't trivial, so it isn't worth specifying in detail unless it will be helpful.) -jJ
Received on Wednesday, 8 July 2009 18:02:14 UTC