WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Table SUMMARY Tag

for

From: Jukka K. Korpela
Date: Mar 3, 2011 1:45PM

Joshue O Connor wrote:

> On 02/03/2011 18:09, Jukka K. Korpela wrote:
>> On the other hand, the HTML5 drafts designate the SUMMARY attribute
>> as obsolete and declare that authors must not use it. Whatever we
>> think of HTML5, it seems to be the way the world is going. I guess
>> the reason for making SUMMARY as obsolete is twofold: first, it has
>> fairly limited support and usage; second, it is better replaced by
>> explanations that every user can make use of.
>
> FWIW For the first point, @Summary is very well supported in most
> browsers, it is recognised by most screen reading AT and announced as
> soon as the table is given focus.

I'm not sure whether there is reliable information about support to
different features across AT software and whether we can estimate the impact
(do we know the usage share, worldwide?). But no doubt the summary attribute
is supported to a significant degree - in AT. In normal browsing, the user
does not even notice its existence, and the great majority of users are not
using any AT. This is what I meant by "fairly limited support". And what
about the usage? I don't think the summary attribute is used on more than
perhaps one page out of a thousand.

I must correct a previous statement of mine about the status of the
attribute in HTML5 drafts. The HTML5 terminology is complex and confusing,
and the drafts use different expressions of dislike. The summary attribute
is not forbidden. It is in a small set of "Obsolete but conforming
features", described at
http://www.whatwg.org/specs/web-apps/current-work/multipage/obsolete.html#obsolete-but-conforming-features
So in everyday terms, the drafts don't say that authors must not use the
summary attribute, just that they should not. On the other hand, they do not
require or even recommend that browsers communicate the attribute value to
users; they use the mild word "may": "If a table element has a summary
attribute, the user agent may report the contents of that attribute to the
user."

> For the second, while on paper this looks like a good idea, removing
> @summary from HTML 5 because it couldn't be used by _everybody_
> became a Shibboleth that lost all meaning. @summary content could be
> exposed to sighted users if browsers were wired to do so.

Authors could even now do such things if they wanted to. With a suitable
piece of Javascript code, the summary attribute value could be disclosed on
request. The main problem would be, I guess, the design of the user
interface. How would you convey the idea "there is a description of the
structure or the purpose of this table available"? Someone (maybe me) might
say "Oh, you could just use those words, in a statetement before the table,
with the word "description" turned to a pseudo-link (Javascript-powered).
And this might work up to a point. But many people who would need the
description would not notice it. People don't read web pages that much. If a
table looks messy, they go elsewhere or just get confused.

Anyway, my somewhat naive idea resembles what the HTML5 draft seems to be
saying, more or less. And you can achieve without using a summary attribute.
Just write the explicit text, with a (pseudo) link, perhaps putting some
easy-to-notice icon, symbolizing "help" or "info", adjacent to it. Screen
reader users won't need the icon, so it could have alt="". Screen readers
users would hear, before the table, something like "There is a (link)
description of the structure or the purpose of the following table
available". Simple, isn't it?

> While
> philosophies like Universal Design are great, I would rather see
> elements and attributes that support some user groups very well, than
> elements and attributes that serve all users poorly.

Well I think it partly depends on how poor "poorly" is. But most
importantly, elements and attributes don't really do anything; people do
things with them. And authors - even accessibility-aware authors - don't
seem to have become excited about the summary attribute.

Therefore I think it is more realistic to try to make authors do some effort
in explaining tables, in prose before the table or in its caption or maybe
sometimes elsewhere. They need to learn some new thinking and design
principles, and to apply them, but meaningful use of the summary attribute
would require that, too.

--
Yucca, http://www.cs.tut.fi/~jkorpela/