W3C home > Mailing lists > Public > www-html@w3.org > May 2007

Re: "Pave The Cowpaths" Design Principle

From: Lachlan Hunt <lachlan.hunt@lachy.id.au>
Date: Sun, 13 May 2007 01:24:05 +1000
Message-ID: <4645DC15.7050206@lachy.id.au>
To: "Patrick H. Lauke" <redux@splintered.co.uk>
CC: www-html@w3.org

Patrick H. Lauke wrote:
> Henri Sivonen wrote:
>> For example, it is more reasonable to require AT to walk the table 
>> model up and to the left to find the header cell for a given data cell 
>> by *default* than to expect authors to use id/headers in the simple 
>> cases.
> Yes, but the two aren't mutually exclusive:

I don't think Henri was saying the approaches were mutually exclusive.

> *if* there are no id/headers, or explicit scope attributes set to 
> TH/TD, do the heuristics (as an error recovery or repair mechanism, 
> if you will).

I don't think omitting headers attributes should be considered an error. 
  There should be a well defined algorithm that covers the vast majority 
of cases reliably without explicit association.

> However, if id/headers etc are detected, do NOT attempt to override them 
> via heuristics, as this is what some AT currently does and it completely 
> undermines any effort by conscientious authors.

Assuming the headers attributes refer to ids of cells within the table, 
that's fine.  But the algorithm should handle references to bogus ids or 
ids that aren't cells within the same table.

> And worth noting that heuristics can only deal with simple tables...as soon 
> as a table is complex, heuristics break. And there's often no easy way to 
> tell simple from complex.

It would be really helpful if some real world example tables, which 
either already use, or would benefit from the use of, headers 
attributes, could be presented as evidence to support your case.  We can 
then evaluate whether or not headers attributes really are needed, or if 
those cases could be handled with a better algorithm.

I have one example that I built early last year for a project I was 
working on.  Although I don't believe the site ever actually launched (I 
think they went out of business or something), I've published an example 
of one of the tables.


Lachlan Hunt
Received on Saturday, 12 May 2007 15:24:18 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 30 April 2020 16:21:03 UTC