Re: How did the summary attribute become part of HTML 4.0?

Steven Faulkner On 09-08-07 12.42:

> hi henri,


>> Regardless of the reason why there's no record to be found, it would be

>> good if someone who was there could recount what lead to the addition of
>> the summary attribute to HTML 4.0.


> agreed, but I do not consider that the title attribute would ahve been
> then or is now an appropriate container for summary  information of this
> kind as i consider it would be annoying to many people if the summary
> information was displayed whenever a user moved their mouse over the table

HTML 4.01's Appendix A, section "A.3 Changes  between HTML 3.2 and 
HTML 4.0 (18 December 1997)") [1] links @summary to the issue of 
"long descriptions":

    "Authors may provide long descriptions of tables
     (see the summary attribute), images and frames
     (see the longdesc attribute)."

In the December 1997 version (I looked it up), the above quote 
does not contain the phrase "(see the summary attribute)". 
Therefore HTML 4.01 points out, that... [2]

    "The longdesc attribute was said to be specified
     for tables. It is not. Instead, the summary
     attribute allows authors to give longer
     descriptions of tables."

Does this hint that <table longdesc="URI"> was considered as an 
option? Or, opposite, what about @summary for images - instead of 

Answer: Probably table@summary was thought to have a much more 
limited role than longdesc pages could have, namely, the rather 
brief task of describing the table structure. (A longdesc image 
description could itself contain a table ...)

The choice of @summary instead of @longdesc also points out that 
the summary is intended to be read quickly - just before reading 
the table. Whereas image descriptions are an extra option - when 
@alt isn't enough. (Of course, @summary also becomes an option - 
or something extra - if the UA doesn't support it ...)

Ferg, btw, when he describes how to use @headers in HTML 4 tables, 
points out that the HTML 4 table design was affected by the need 
for "incremental rendering", as tables could be slow to download 
and render on slow modem/computers[3]. Thus one should think that 
a short @summary text would also save bytes (compared with an 
side-by-side descriptive paragraph). Another variant of the fact 
that "authors might have reasons to not insert a visible 
description" ... (Placing the long image description in a separate 
@longdesc document also helped save bytes - by the way...)

So, might it be that table@summary was meant to be presented to AT 
users /before/ the entire table had been rendered?  If so, then 
this might also explain why some @summary texts perhaps are more 
wordy than we today consider kosher: In such a scenario it would 
perhaps not matter if the summary repeated bits of what the user 
later would read inside a caption. (For instance, perhaps the user 
would use the @summary info to simply skip reading/loading the table?)

Of course - just speculations.

As for truncated @title tooltips: it seems natural that they 
*considered* using @title before going for @summary. Even today, 
@title and @alt are often discussed in context ...

But it has been pointed out to me that only expert AT users access 
the @title information. And @title is also, in general, too 
semantically unspecific. With some exceptions, such as <abbr>, 
that prove that it sometimes is /possible/ to define specific 
@title semantics. (In my view, it ought to have been possible for 
caption@title - though the right moment for that idea probably was 
10-12 years ago.)

Conclusion: 1) We need a time machine. 2) @summary doesn't have 
the same role to play as it potentially had.

leif halvard silli

Received on Saturday, 8 August 2009 00:17:43 UTC